Articles de revues sur le sujet « Multi-Branch generative models »

The study of generative models is a promising branch of deep learning techniques, which has been successfully applied to different scenarios, such as Artificial Intelligence and the Internet of Things. While in most of the existing works, the generative models are realized as a centralized structure, raising the threats of security and privacy and the overburden of communication costs. Rare efforts have been committed to investigating distributed generative models, especially when the training data comes from multiple heterogeneous sources under realistic IoT settings. In this paper, to handle this challenging problem, we design a federated generative model framework that can learn a powerful generator for the hierarchical IoT systems. Particularly, our generative model framework can solve the problem of distributed data generation on multi-source heterogeneous data in two scenarios, i.e., feature related scenario and label related scenario. In addition, in our federated generative models, we develop a synchronous and an asynchronous updating methods to satisfy different application requirements. Extensive experiments on a simulated dataset and multiple real datasets are conducted to evaluate the data generation performance of our proposed generative models through comparison with the state-of-the-arts.

Achieving super-resolution with satellite images is a critical task for enhancing the utility of remote sensing data across various applications, including urban planning, disaster management, and environmental monitoring. Traditional interpolation methods often fail to recover fine details, while deep-learning-based approaches, including convolutional neural networks (CNNs) and generative adversarial networks (GANs), have significantly advanced super-resolution performance. Recent studies have explored large-scale models, such as Transformer-based architectures and diffusion models, demonstrating improved texture realism and generalization across diverse datasets. However, these methods frequently have high computational costs and require extensive datasets for training, making real-world deployment challenging. We propose the multi-branch generative prior integration network (MBGPIN) to address these limitations. This novel framework integrates multiscale feature extraction, hybrid attention mechanisms, and generative priors derived from pretrained VQGAN models. The dual-pathway architecture of the MBGPIN includes a feature extraction pathway for spatial features and a generative prior pathway for external guidance, dynamically fused using an adaptive generative prior fusion (AGPF) module. Extensive experiments on benchmark datasets such as UC Merced, NWPU-RESISC45, and RSSCN7 demonstrate that the MBGPIN achieves superior performance compared to state-of-the-art methods, including large-scale super-resolution models. The MBGPIN delivers a higher peak signal-to-noise ratio (PSNR) and higher structural similarity index measure (SSIM) scores while preserving high-frequency details and complex textures. The model also achieves significant computational efficiency, with reduced floating point operations (FLOPs) and faster inference times, making it scalable for real-world applications.

In the scenario of limited labeled remote-sensing datasets, the model’s performance is constrained by the insufficient availability of data. Generative model-based data augmentation has emerged as a promising solution to this limitation. While existing generative models perform well in natural scene domains (e.g., faces and street scenes), their performance in remote sensing is hindered by severe data imbalance and the semantic similarity among land-cover classes. To tackle these challenges, we propose the Multi-Class Guided GAN (MCGGAN), a novel network for generating remote-sensing images from semantic labels. Our model features a dual-branch architecture with a global generator that captures the overall image structure and a multi-class generator that improves the quality and differentiation of land-cover types. To integrate these generators, we design a shared-parameter encoder for consistent feature encoding across two branches, and a spatial decoder that synthesizes outputs from the class generators, preventing overlap and confusion. Additionally, we employ perceptual loss (LVGG) to assess perceptual similarity between generated and real images, and texture matching loss (LT) to capture fine texture details. To evaluate the quality of image generation, we tested multiple models on two custom datasets (one from Chongzhou, Sichuan Province, and another from Wuzhen, Zhejiang Province, China) and a public dataset LoveDA. The results show that MCGGAN achieves improvements of 52.86 in FID, 0.0821 in SSIM, and 0.0297 in LPIPS compared to the Pix2Pix baseline. We also conducted comparative experiments to assess the semantic segmentation accuracy of the U-Net before and after incorporating the generated images. The results show that data augmentation with the generated images leads to an improvement of 4.47% in FWIoU and 3.23% in OA across the Chongzhou and Wuzhen datasets. Experiments show that MCGGAN can be effectively used as a data augmentation approach to improve the performance of downstream remote-sensing image segmentation tasks.

Parametric modeling of transition tubes were implemented based on constant cross section area assumption along the main central routes on CATIA software. The key objective of modeling these similar structures is to provide more geometric configuration options and modifications of micro channels for multi phase flow systems. The modeling processes were parameterized and analyzed by CATIA “Generative Shape Design” module with the help of “Parameters” and “Relations” functions. The surface models are all designed in circular cross sections that are constrained in two ways: one is perpendicular to the main central routes of the tube for planar transitional junction, and another is, perpendicular to the sub-branch central routes for oblique transitional junction three dimensionally. Next work is emphasized on numerical simulation and experimental investigation with these geometric structures in a multi phase flow system.

With the flourishing development of generative models, image manipulation is receiving increasing attention. Rather than text modality, several elegant designs have delved into leveraging audio to manipulate images. However, existing methodologies mainly focus on image generation conditional on semantic alignment, ignoring the vivid affective information depicted in the audio. We propose an Emotion-aware StyleGAN Manipulator (EmoStyle), a framework where affective information from audio can be explicitly extracted and further utilized during image manipulation. Specifically, we first leverage the multi-modality model ImageBind for initial cross-modal retrieval between images and music, and select the music-related image for further manipulation. Simultaneously, by extracting sentiment polarity from the lyrics of the audio, we generate an emotionally rich auxiliary music branch to accentuate the affective information. We then leverage pre-trained encoders to encode audio and the audio-related image into the same embedding space. With the aligned embeddings, we manipulate the image via a direct latent optimization method. We conduct objective and subjective evaluations on the generated images, and our results show that our framework is capable of generating images with specified human emotions conveyed in the audio.

The belt conveyor is an essential piece of equipment in coal mining for coal transportation, and its stable operation is key to efficient production. Belt surface of the conveyor is vulnerable to foreign bodies which can be extremely destructive. In the past decades, much research and numerous approaches to inspect belt status have been proposed, and machine learning-based non-destructive testing (NDT) methods are becoming more and more popular. Deep learning (DL), as a branch of machine learning (ML), has been widely applied in data mining, natural language processing, pattern recognition, image processing, etc. Generative adversarial networks (GAN) are one of the deep learning methods based on generative models and have been proved to be of great potential. In this paper, a novel multi-classification conditional CycleGAN (MCC-CycleGAN) method is proposed to generate and discriminate surface images of damages of conveyor belt. A novel architecture of improved CycleGAN is designed to enhance the classification performance using a limited capacity images dataset. Experimental results show that the proposed deep learning network can generate realistic belt surface images with defects and efficiently classify different damaged images of the conveyor belt surface.

Deep learning, a pivotal branch of artificial intelligence, has increasingly influenced the financial domain with its advanced data processing capabilities. This paper introduces Factor-GAN, an innovative framework that utilizes Generative Adversarial Networks (GAN) technology for factor investing. Leveraging a comprehensive factor database comprising 70 firm characteristics, Factor-GAN integrates deep learning techniques with the multi-factor pricing model, thereby elevating the precision and stability of investment strategies. To explain the economic mechanisms underlying deep learning, we conduct a subsample analysis of the Chinese stock market. The findings reveal that the deep learning-based pricing model significantly enhances return prediction accuracy and factor investment performance in comparison to linear models. Particularly noteworthy is the superior performance of the long-short portfolio under Factor-GAN, demonstrating an annualized return of 23.52% with a Sharpe ratio of 1.29. During the transition from state-owned enterprises (SOEs) to non-SOEs, our study discerns shifts in factor importance, with liquidity and volatility gaining significance while fundamental indicators diminish. Additionally, A-share listed companies display a heightened emphasis on momentum and growth indicators relative to their dual-listed counterparts. This research holds profound implications for the expansion of explainable artificial intelligence research and the exploration of financial technology applications.

Abstract. The technology for 3D reconstruction of tree models based on point clouds has been extensively researched, necessitating effective datasets for the study of branch and leaf separation, skeleton point extraction, and tree parameter extraction methods. However, existing datasets for 3D tree models face several challenges, including insufficient data volume for deep learning network training, low accuracy of model ground truth impeding effective method precision evaluation, and a lack of dataset richness to satisfy the needs of multi-type method assessments. In response to these challenges, This paper introduces, for the first time, a fully automated method for generating structured three-dimensional synthetic tree models, and constructs a large-scale 3D synthetic tree dataset enriched with comprehensive structural information. This method facilitates automated computation across several processes, including the mass generation of simulated trees, separation of branches and leaves, noise generation, extraction of skeleton points, and volume calculation. To validate the usability of this dataset across various applications, this paper employs state-of-the-art (SoTA) algorithms to verify the accuracy of methods in 3D tree model reconstruction and carbon stock calculation, thereby thoroughly demonstrating the dataset’s effectiveness.

One of the promising directions of small-scale distributed power generation for Russia is the use of biomass. The present work is devoted to studies of an mini-CHP based on multi-stage biomass gasification. Mathematical models of elements and mini-CHP in general based on technological schemes were constructed. The mathematical models were constructed with the software developed at Melentiev Energy Systems Institute of Siberian Branch of the Russian Academy of Sciences. The calculations were made for two sizes of internal combustion engines. Thus, we obtained the values of flow rates, temperatures of heat carriers at various points of flow charts of the plants.

AbstractData-to-Text Generation (DTG) is a subfield of Natural Language Generation aiming at transcribing structured data in natural language descriptions. The field has been recently boosted by the use of neural-based generators which exhibit on one side great syntactic skills without the need of hand-crafted pipelines; on the other side, the quality of the generated text reflects the quality of the training data, which in realistic settings only offer imperfectly aligned structure-text pairs. Consequently, state-of-art neural models include misleading statements –usually called hallucinations—in their outputs. The control of this phenomenon is today a major challenge for DTG, and is the problem addressed in the paper. Previous work deal with this issue at the instance level: using an alignment score for each table-reference pair. In contrast, we propose a finer-grained approach, arguing that hallucinations should rather be treated at the word level. Specifically, we propose a Multi-Branch Decoder which is able to leverage word-level labels to learn the relevant parts of each training instance. These labels are obtained following a simple and efficient scoring procedure based on co-occurrence analysis and dependency parsing. Extensive evaluations, via automated metrics and human judgment on the standard WikiBio benchmark, show the accuracy of our alignment labels and the effectiveness of the proposed Multi-Branch Decoder. Our model is able to reduce and control hallucinations, while keeping fluency and coherence in generated texts. Further experiments on a degraded version of ToTTo show that our model could be successfully used on very noisy settings.

Botnets pose a significant challenge in network security by leveraging Domain Generation Algorithms (DGA) to evade traditional security measures. Extracting DGA domain samples is inherently complex, and the current DGA detection models often struggle to capture domain features effectively when facing limited training data. This limitation results in suboptimal detection performance and an imbalance between model accuracy and complexity. To address these challenges, this paper introduces a novel multi-scale feature fusion model that integrates the Transformer architecture with the Rapid Selective Kernel Network (R-SKNet). The proposed model employs the Transformer’s encoder to couple the single-domain character elements with the multiple types of relationships within the global domain block. This paper proposes integrating R-SKNet into DGA detection and developing an efficient channel attention (ECA) module. By enhancing the branch information guidance in the SKNet architecture, the approach achieves adaptive receptive field selection, multi-scale feature capture, and lightweight yet efficient multi-scale convolution. Moreover, the improved Feature Pyramid Network (FPN) architecture, termed EFAM, is utilized to adjust channel weights for outputs at different stages of the backbone network, leading to achieving multi-scale feature fusion. Experimental results demonstrate that, in tasks with limited training samples, the proposed method achieves lower computational complexity and higher detection accuracy compared to mainstream detection models.

As an important branch of the development of computer science, artificial intelligence not only belongs to the basic component of computer science and technology research, but also has been widely used in computer engineering. As the basic environment for the research of artificial intelligence system, it is the focus of Chinese researchers to deeply explore the intelligent model and describe the characteristics and future development direction of language. Therefore, after understanding the development trend of artificial intelligence and intelligent language, this paper mainly studies how to use modeling elements in UML to describe infrastructure, cooperation model, information model and behavior model in business models, and describes language and algorithm generation according to intelligent models oriented to multi-objective problems, providing an effective basis for artificial intelligence language research in the new era.

The remarkable development of deepfake models has facilitated the generation of fake content with various modalities, such as forged images, manipulated audio, and modified video with (or without) corresponding audio. However, many existing methods only analyze content with known and fixed modalities to identify deepfakes, which restricts their focus on intra-domain inconsistencies, and they fail to explore diverse modal and inter-domain hierarchical inconsistencies. In this work, we propose a novel unified neural network named MGDL-Net (Modal-Guided Domain Learning Network), which contains a spatial branch, a temporal branch, and a frequency branch. This diverse combination of branches endows our network with the ability to detect face-related input with flexible modalities and perceive both intra- and inter-domain inconsistencies, such as unimodal, bimodal, and trimodal modalities. To effectively and comprehensively capture the various inconsistencies, we propose implementing heterogeneous inconsistency learning (HIL) with a three-level joint extraction paradigm. In particular, HIL performs heterogeneous learning from spatial, temporal, and frequency perspectives to generate more generalized representations of forgery and eliminate the interference of static redundant information. Furthermore, a multi-modal deepfake dataset is also constructed. We have conducted extensive experiments, and our results have demonstrated that the proposed method can achieve an outstanding performance compared to that of numerous state-of-the-art methods, which implies that the cross-modal inconsistency learning we propose is beneficial for multi-modal face forgery detection.

The production of highly accurate land cover maps is one of the primary challenges in remote sensing, which depends on the spatial resolution of the input images. Sometimes, high-resolution imagery is not available or is too expensive to cover large areas or to perform multitemporal analysis. In this context, we propose a multi-task network to take advantage of the freely available Sentinel-2 imagery to produce a super-resolution image, with a scaling factor of 5, and the corresponding high-resolution land cover map. Our proposal, named SEG-ESRGAN, consists of two branches: the super-resolution branch, that produces Sentinel-2 multispectral images at 2 m resolution, and an encoder–decoder architecture for the semantic segmentation branch, that generates the enhanced land cover map. From the super-resolution branch, several skip connections are retrieved and concatenated with features from the different stages of the encoder part of the segmentation branch, promoting the flow of meaningful information to boost the accuracy in the segmentation task. Our model is trained with a multi-loss approach using a novel dataset to train and test the super-resolution stage, which is developed from Sentinel-2 and WorldView-2 image pairs. In addition, we generated a dataset with ground-truth labels for the segmentation task. To assess the super-resolution improvement, the PSNR, SSIM, ERGAS, and SAM metrics were considered, while to measure the classification performance, we used the IoU, confusion matrix and the F1-score. Experimental results demonstrate that the SEG-ESRGAN model outperforms different full segmentation and dual network models (U-Net, DeepLabV3+, HRNet and Dual_DeepLab), allowing the generation of high-resolution land cover maps in challenging scenarios using Sentinel-2 10 m bands.

The focus of this paper is on computer-aided process planning (CAPP) for parts manufacture in systems of definite processing capabilities, involving multi-axis machining centers. A methodical approach is developed to optimally solve for process planning problems, which consists in the identification of process alternatives and sequencing working steps. The approach involves the use of the branch and bound (B&B) concept from the field of artificial intelligence (AI). A conceptual scheme for generation of alternative process plans in the form of a network is developed, based on part design data modeling in terms of machining features. A relevant algorithm is proposed for creating such a network and searching for the optimal process plan solution from the viewpoint of its operational performance, under formulated process constraints. The use of the approach is studied numerically with regard to a real life case study and diverse machine tools with relevant tooling are considered. Generated process alternatives for complex machining with given systems are studied using models programmed in the Matlab environment.

Both the European and the Hungarian rural areas suffer multi dimensional problems. Beside infrastructural under development the most important difficulty is employment. Unemployment is significant in the rural areas, while other structural characteristics like education, profession, work circumstances and seasonality worsen this unfavourable situation. It can be stated that the challenge with the highest priority in rural and spatialdevelopment is to create jobs and to strengthen local employment. The authors examine the job generating possibilities of energetic biomass of agricultural origin in a structural point of view. The aim is to develop spatial biomass product line models that permanently support the raise of employment by utilizing the possibilities of the European Union support policy and the popularity of this branch.

Bed topography and roughness play important roles in numerous ice-sheet analyses. Although the coverage of ice-penetrating radar measurements has vastly increased over recent decades, significant data gaps remain in certain areas of subglacial topography and need interpolation. However, the bed topography generated by interpolation such as kriging and mass conservation is generally smooth at small scales, lacking topographic features important for sub-kilometer roughness. DeepBedMap, a deep learning method combined with multiple surface observation inputs, can generate high-resolution (250 m) bed topography with realistic bed roughness but produces some unrealistic artifacts and higher bed elevation values in certain regions, which could bias ice-sheet models. To address these issues, we present MB_DeepBedMap, a multi-branch deep learning method to generate more realistic bed topography. The model improves upon DeepBedMap by separating inputs into two groups using a multi-branch network structure according to their characteristics, rather than fusing all inputs at an early stage, to reduce artifacts in the generated topography caused by earlier fusion of inputs. A direct upsampling branch preserves large-scale subglacial landforms while generating high-resolution bed topography. We use MB_DeepBedMap to generate a high-resolution (250 m) bed elevation grid product of Antarctica, MB_DeepBedMap_DEM, which can be used in high-resolution ice-sheet modeling studies. Moreover, we test the performance of MB_DeepBedMap model in Thwaites Glacier, Gamburtsev Subglacial Mountains, and several other regions, by comparing the qualitative topographic features and quantitative errors of MB_DeepBedMap, BEDMAP2, BedMachine Antarctica, and DeepBedMap. The results show that MB_DeepBedMap can provide more realistic small-scale topographic features and roughness compared to BEDMAP2, BedMachine Antarctica, and DeepBedMap.

The dynamic characteristics of thermal energy play an important role of multi-scale coupling among heterogeneous energy sources in integrated energy systems (IES). In Part I, for the purpose of accurately describing the dynamic processes of thermal energy transmission, the theoretical approach and models were proposed and verified by numerical simulation. In this part, an innovative analytical method based on entropy flow was derived on the basis of theory developed in Part I, which can assess the quantity and quality of thermal transport. A comparative case study indicates that the change trend of entropy flow of each node is consistent with the change of available power, but independent of temperature. The node entropy flow is increased compared with the algebraic sum of branch entropy flow, which reflects the phenomenon of entropy generation in the mixing process; while the change of available power is just the opposite. This means the irreversible entropy generation at the node leads to loss of the available thermal power. Therefore, it is more accurate to describe the dynamic thermal transmission process on the scale of entropy. This proves the effectiveness of the models proposed in Part I as well as the methods in this part from the perspective of application.

Cropland segmentation of satellite images is an essential basis for crop area and yield estimation tasks in the remote sensing and computer vision interdisciplinary community. Instead of common pixel-level segmentation results with salt-and-pepper effects, a parcel-level output conforming to human recognition is required according to the clients' needs during the model deployment. However, leveraging CNN-based models requires fine-grained parcel-level labels, which is an unacceptable annotation burden. To cure these practical pain points, in this paper, we present PARCS, a holistic deployment-oriented AI system for PARcel-level Cropland Segmentation. By consolidating multi-disciplinary knowledge, PARCS has two algorithm branches. The first branch performs pixel-level crop segmentation by learning from limited labeled pixel samples with an active learning strategy to avoid parcel-level annotation costs. The second branch aims at generating the parcel regions without a learning procedure. The final parcel-level segmentation result is achieved by integrating the outputs of these two branches in tandem. The robust effectiveness of PARCS is demonstrated by its outstanding performance on public and in-house datasets (an overall accuracy of 85.3% and an mIoU of 61.7% on the public PASTIS dataset, and an mIoU of 65.16% on the in-house dataset). We also include subjective feedback from clients and discuss the lessons learned from deployment.

Abstract. The Met Office operational wave forecasting modelling system runs four times a day to provide global and regional forecasts up to 7 d ahead. The underpinning model uses a recent development branch of the third-generation spectral wave model WAVEWATCH III® (version 7.12) that includes several updates developed at the Met Office. These include the spherical multiple-cell (SMC) grid, a rotated pole grid formulation for mid-latitudes, enhancements to OASIS coupling and updates to the netCDF postprocessing. Here we document the technical details behind the system with a view to further developments. The operational system includes a global forecast deterministic model (GS512L4EUK) and two regional models nested one-way covering the Northwest (NW) European shelf and UK waters (AMM15SL2, where AMM is for Atlantic Margin model) in addition to an Atlantic wave ensemble (AS512L4EUK). GS512L4EUK and AS512L4EUK are based on a multi-resolution four-tier SMC 25-12-6-3 km grid. The regional AMM15SL2 configuration uses a two-tier SMC 3−1.5 km grid and is run operationally both as a standalone forced model (includes wave–current interactions) and as the wave component of a two-way ocean–wave coupled operational system. Model evaluation is focused on the global and regional baseline configurations. Results show evidence of resolution-dependent differences in wave growth, leading to slightly overestimated significant wave heights in coastal mid-range conditions by AMM15SL2 but an improved representation of extremes compared to GS512L4EUK. Additionally, although a positive impact of the surface currents is not always shown in the overall statistics of the significant wave height due to a larger spread in the observation–model differences, wave–current effects help to better capture the distribution of the energy in terms of frequency and direction near the coast (>20 % improvement), which has implications to beach safety, coastal overtopping risk and shoreline evolution. Future system developments such as the use of sea point wind forcing, the optimisation of the models in line with model resolution and the utilisation of SMC multi-grids are discussed.

Stakeholders’ interests on renewable and clean energy sources experienced a huge increase in the last decades, thanks to the remarkable benefits on climate, economic, and social issues. The integration of flexible energy storage systems represents a great chance for a further increased penetration as they support the mitigation of renewables’ main drawbacks (i.e., stochastic behavior) and guarantee the balance between energy supply and demand enabling non-simultaneous production and consumption. The increased focus on distributed generation and storage was also of interest to the research community which investigated both the economic and technical performances of the integrated systems. The operations management branch addressed this topic, since storage devices present many similarities with traditional inventory management applied to regular commodities. At a user level, the relation between energy production and storage can be studied by analogy with inventory models. Specifically, this study presents a multi-period newsvendor model for the management and optimal sizing of a battery energy storage system installed to increase the self-consumption rate by allowing loads shifting. This work aims to extend the traditional inventory management applying its concepts to energy systems operations in order to minimize the total energy cost. A numerical study is provided to show the behavior of the model.

Due to the complexity of scene interference and the variability of ship scale and position, automatic ship detection in remote sensing images makes for challenging research. The existing deep networks rarely design receptive fields that fit the target scale based on training data. Moreover, most of them ignore the effective retention of position information in the feature extraction process, which reduces the contribution of features to subsequent classification. To overcome these limitations, we propose a novel ship detection framework combining the dilated rate selection and attention-guided feature representation strategies, which can efficiently detect ships of different scales under the interference of complex environments such as clouds, sea clutter and mist. Specifically, we present a dilated convolution parameter search strategy to adaptively select the dilated rate for the multi-branch extraction architecture, adaptively obtaining context information of different receptive fields without sacrificing the image resolution. Moreover, to enhance the spatial position information of the feature maps, we calculate the correlation of spatial points from the vertical and horizontal directions and embed it into the channel compression coding process, thus generating the multi-dimensional feature descriptors which are sensitive to direction and position characteristics of ships. Experimental results on the Airbus dataset demonstrate that the proposed method achieves state-of-the-art performance compared with other detection models.

Abstract The last decade in particular has seen prosperity in global challenges. Traditional industry leaders have to compete with the challenges of the new generation in emerging markets. Although the European market is competitively mature, the prosperity in global challenges makes it necessary to defend the domestic market as well as to pursue growth opportunities in emerging markets. Analyzing the trends, a comparison of the US and Chinese innovation commercialization policies from a customer perspective with Europe was distinguished. The main difference is Europe's narrowly defined customer segment, companies do not adapt to customer micro-segments with non-matching product and service requirements, because individual attraction strategies must be implemented for each branch, which increase the difficulty to draw potential global clients. The aim of this work is to analyze the US and Chinese innovation commercialization process policies from a customer perspective to gain an insight into the improvement of European national policy models. A multi-criteria decision analysis was used to compare the US and European influencing factor criteria, which focuses on customer segmentation, to determine the most influential segment. Data were also collected in the analysis of the success of European commercialization. The focus on the US end-market approach is expected to increase the value of the proximity factor by improving the implementation of European innovation policy in practice.

Correctly anticipating PV electricity production may lessen stochastic fluctuations and incentivize energy consumption. To address the intermittent and unpredictable nature of photovoltaic power generation, this article presents an ensemble learning model (MVMD-CLES) based on the whale optimization algorithm (WOA), variational mode decomposition (VMD), convolutional neural network (CNN), long and short-term memory (LSTM), and extreme learning machine (ELM) stacking. Given the variances in the spatiotemporal distribution of photovoltaic data and meteorological features, a multi-branch character extraction iterative mixture learning model is proposed: we apply the MWOA algorithm to find the optimal decomposition times and VMD penalty factor, and then divide the PV power sequences into sub-modes with different frequencies using a two-layer algorithmic structure to reconstruct the obtained power components. The primary learner is CNN–BiLSTM, which is utilized to understand the temporal and spatial correlation of PV power from information about the weather and the output of photovoltaic cells, and the LSTM learns the periodicity and proximity correlation of the power data and obtains the corresponding component predictions. The second level is the secondary learner—the output of the first layer is learned again using the ELM to attenuate noise and achieve short-term prediction. In different case studies, regardless of weather changes, the proposed method is provided with the best group of consistency and constancy, with an average RMSE improvement of 12.08–39.14% over a single-step forecast compared to other models, the average forecast RMSE increased by 5.71–9.47% for the first two steps.

Abstract PISCA, originally introduced by Martinez et al. in 2018, represents a pivotal Bayesian phylogenetics tool for the modeling of tumor evolution using multi-region somatic chromosomal alteration (SCA) data. PISCA takes allele-specific copy number data, typically obtained from deep genome sequencing or SNP arrays, or absolute copy number data from low-pass genome sequencing methodologies. It extends the classic BEAST1 framework and inherits a rich repertoire of evolutionary models. Importantly, PISCA leverages longitudinal sampling to estimate SCA mutational clock rates, either employing strict clock models, where mutation rates remain constant throughout the evolutionary tree, or relaxed clocks, which allow each branch or subtree to possess its distinct mutation rate. This nuanced approach empowers PISCA to account for the heterogeneous rates of mutations, a pivotal consideration in understanding tumor evolution dynamics. However, PISCA has historically posed a formidable entry barrier for wider community adoption due to platform and Java dependencies for installation, as well as the need for manual or custom-scripted XML file generation. Here we showcase PISCA-box, a user-friendly interface designed to streamline the generation and testing of XML files. Our PISCA-box Docker image works on any desktop machine to create a locally hosted webpage, where users can input SCA data, sampling dates, select clock and demographic models, and set priors for key parameters - similar to the BEAST XML generator, BEAUTi. The Docker or Singularity installation can then be used for longer analyses using high-performance computing resources. To demonstrate its practical utility, we present novel colorectal cancer data. First, we examine data from a patient with a long-standing history of inflammatory bowel disorder (IBD), a known high-risk factor for colorectal cancer. This patient had undergone surveillance colonoscopies for many years, which provided 38 samples, and subsequently an additional 118 samples were collected from a total colectomy. All were analyzed with low-coverage whole genome sequencing. We find some samples from surveillance and colectomy form lineages with overlapping copy number events, that we used to construct a phylogeny, finding a cancer-adjacent clade that appears to evolve rapidly. A second dataset of multi-region samples from a cohort of exceptional survivors of oligometastatic colorectal cancer who lived >60 months from metastatic diagnosis with biopsies/resections across 3-10 time points is examined and temporal models are used to estimate the ages of metastatic clades. PISCA therefore harnesses the power of longitudinal SCA data to enable a comprehensive study of SCA dynamics. We are currently expanding this framework to include fluctuating methylation clocks which PISCA-box will soon include. By providing this accessible tool, we enable researchers to readily apply Bayesian phylogenetics to real-world clinical datasets to better understand tumor evolution. Citation Format: Heather E. Grant, Rachel Alcraft, Pablo Bousquets-Muñoz, Calum Gabbutt, Alison Berner, Mehmet Yalchin, Carlo C. Maley, Trevor A. Graham, Diego Mallo. PISCA-box: A user-friendly interface for Phylogenetic Inference using Somatic Chromosomal Alterations (PISCA) [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr B004.

Channel modeling is a first step towards the successful projecting of any wireless communication system. Hence, in this paper, we analyze the performance at the output of a multi-branch selection combining (SC) diversity receiver in a wireless environment that has been distracted by fading and co-channel interference (CCI), whereby the fading is modelled by newer Beaulieu-Xie (BX) distribution, and the CCI is modelled by the κ-µ distribution. The BX distribution provides the ability to include in consideration any number of line-of-sight (LOS) useful signal components and non-LOS (NLOS) useful signal components. This distribution contains characteristics of some other fading models thanks to its flexible fading parameters, which also applies to the κ-µ distribution. We derived here the expressions for the probability density function (PDF) and cumulative distribution function (CDF) for the output signal-to-co-channel interference ratio (SIR). After that, other performances are obtained, namely: outage probability (Pout), channel capacity (CC), moment-generating function (MGF), average bit error probability (ABEP), level crossing rate (LCR), and average fade duration (AFD). Numerical results are presented in several graphs versus the SIR for different values of fading and CCI parameters, as well as the number of input branches in the SC receiver. Then, the impact of parameters on all performance is checked. From our numerical results, it is possible to directly obtain the performance for all derived and displayed quantities for cases of previously known distributions of fading and CCI by inserting the appropriate parameter values. In the second part of the paper, a workflow for automated network experimentation relying on the synergy of Large Language Models (LLMs) and model-driven engineering (MDE) is presented, while the previously derived expressions are used for evaluation. Due to the aforementioned, the biggest value of the obtained results is the applicability to the cases of a large number of other distributions for fading and CCI by replacing the corresponding parameters in the formulas for the respective performances.

During the production of electronic panels, surface defects will inevitably appear. How to quickly and accurately detect these defects is very important to improve product quality. However, some problems such as high cost and low accuracy are still prominent when existing manual detection and traditional techniques are used to solve such problems. Therefore, more and more computer vision techniques are proposed to solve such problems, but the current application of deep learning-based object detection networks for surface defect detection of electronic panels is in a gap. The analysis found that there are two main reasons for this phenomenon. On the one hand, the surface defects of electronic panels have their unique characteristics such as multi-scale and irregular shape, and the current object detection networks cannot effectively solve these problems. On the other hand, the regression and classification tasks coupled in the current computational mechanism of each network are commonly found to cause the problem of conflict between them, which makes it more difficult to adapt these network models to the detection tasks in this scenario. Based on this, we design a supervised object detection network for electronic panel surface defect detection scenario for the first time. The computational mechanism of this network includes a prediction box generation strategy based on the double branch structure and a detection head design strategy that decouples the regression task from the classification task. In addition, we validated the designed network and the proposed method on our own collected dataset of surface defects in electronic panels. The final results of the comparative and ablation experiments show that our proposed method achieves an average accuracy of 78.897% for 64 surface defect categories, proving that its application to electronic panel surface defect detection scenarios can achieve better results.

Abstract. The South Atlantic rift basin evolved as branch of a large Jurassic-Cretaceous intraplate rift zone between the African and South American plates during the final breakup of western Gondwana. While the relative motions between South America and Africa for post-breakup times are well resolved, many issues pertaining to the fit reconstruction and particular the relation between kinematics and lithosphere dynamics during pre-breakup remain unclear in currently published plate models. We have compiled and assimilated data from these intraplated rifts and constructed a revised plate kinematic model for the pre-breakup evolution of the South Atlantic. Based on structural restoration of the conjugate South Atlantic margins and intracontinental rift basins in Africa and South America, we achieve a tight fit reconstruction which eliminates the need for previously inferred large intracontinental shear zones, in particular in Patagonian South America. By quantitatively accounting for crustal deformation in the Central and West African rift zone, we have been able to indirectly construct the kinematic history of the pre-breakup evolution of the conjugate West African-Brazilian margins. Our model suggests a causal link between changes in extension direction and velocity during continental extension and the generation of marginal structures such as the enigmatic Pre-salt sag basin and the São Paulo High. We model an initial E–W directed extension between South America and Africa (fixed in present-day position) at very low extensional velocities until Upper Hauterivian times (≈126 Ma) when rift activity along in the equatorial Atlantic domain started to increase significantly. During this initial ≈17 Myr-long stretching episode the Pre-salt basin width on the conjugate Brazilian and West African margins is generated. An intermediate stage between 126.57 Ma and Base Aptian is characterised by strain localisation, rapid lithospheric weakening in the equatorial Atlantic domain, resulting in both progressively increasing extensional velocities as well as a significant rotation of the extension direction to NE–SW. From Base Aptian onwards diachronous lithospheric breakup occurred along the central South Atlantic rift, first in the Sergipe-Alagoas/Rio Muni margin segment in the northernmost South Atlantic. Final breakup between South America and Africa occurred in the conjugate Santos–Benguela margin segment at around 113 Ma and in the Equatorial Atlantic domain between the Ghanaian Ridge and the Piauí-Ceará margin at 103 Ma. We conclude that such a multi-velocity, multi-directional rift history exerts primary control on the evolution of this conjugate passive margins systems and can explain the first order tectonic structures along the South Atlantic and possibly other passive margins.

Abstract. The South Atlantic rift basin evolved as a branch of a large Jurassic–Cretaceous intraplate rift zone between the African and South American plates during the final break-up of western Gondwana. While the relative motions between South America and Africa for post-break-up times are well resolved, many issues pertaining to the fit reconstruction and particularly the relation between kinematics and lithosphere dynamics during pre-break-up remain unclear in currently published plate models. We have compiled and assimilated data from these intraplated rifts and constructed a revised plate kinematic model for the pre-break-up evolution of the South Atlantic. Based on structural restoration of the conjugate South Atlantic margins and intracontinental rift basins in Africa and South America, we achieve a tight-fit reconstruction which eliminates the need for previously inferred large intracontinental shear zones, in particular in Patagonian South America. By quantitatively accounting for crustal deformation in the Central and West African Rift Zones, we have been able to indirectly construct the kinematic history of the pre-break-up evolution of the conjugate west African–Brazilian margins. Our model suggests a causal link between changes in extension direction and velocity during continental extension and the generation of marginal structures such as the enigmatic pre-salt sag basin and the São Paulo High. We model an initial E–W-directed extension between South America and Africa (fixed in present-day position) at very low extensional velocities from 140 Ma until late Hauterivian times (≈126 Ma) when rift activity along in the equatorial Atlantic domain started to increase significantly. During this initial ≈14 Myr-long stretching episode the pre-salt basin width on the conjugate Brazilian and west African margins is generated. An intermediate stage between ≈126 Ma and base Aptian is characterised by strain localisation, rapid lithospheric weakening in the equatorial Atlantic domain, resulting in both progressively increasing extensional velocities as well as a significant rotation of the extension direction to NE–SW. From base Aptian onwards diachronous lithospheric break-up occurred along the central South Atlantic rift, first in the Sergipe–Alagoas/Rio Muni margin segment in the northernmost South Atlantic. Final break-up between South America and Africa occurred in the conjugate Santos–Benguela margin segment at around 113 Ma and in the equatorial Atlantic domain between the Ghanaian Ridge and the Piauí-Ceará margin at 103 Ma. We conclude that such a multi-velocity, multi-directional rift history exerts primary control on the evolution of these conjugate passive-margin systems and can explain the first-order tectonic structures along the South Atlantic and possibly other passive margins.

Abstract Background: Early T-cell precursor acute lymphoblastic leukemia (ETP-ALL) is an uncommon childhood leukemia that has been associated with very poor clinical outcomes in some studies. ETP-ALL cells arrest at a more immature differentiation stage than other T-lymphoblasts, and are hypothesized to retain multi-lineage differentiation potential, which may contribute to chemoresistance with standard lymphoid-directed therapy. Based on the recent clinical success of chimeric antigen receptor (CAR)-modified T-cells in children with B-ALL, we sought to identify potential surface protein targets on ETP lymphoblasts using differential gene expression analysis combined with a bioinformatic algorithm to predict surface expression. Methods: Cell-surface targets on ETP-ALL were predicted by identifying overexpressed transcripts based on gene expression and a bioinformatic algorithm to predict surface expression. Using several gene expression platforms and reference databases, (Oncogenomics website-Pediatric Oncology Branch, NCI, Gene Expression Omnibus, Gene Ontology, Human Protein References Database) ETP-ALL samples were compared to peripheral blood mononuclear cell (PBMC) controls on an individual transcript basis. A list of the top 25 transcripts was generated based on cell surface proteins, and the resultant list ordered by the degree of difference from PBMC controls. We next used human leukemia cells from six established ETP-ALL patient-derived xenograft (PDX) models using flow cytometry to evaluate for cell surface expression of proteins encoded by the overexpressed transcripts. Additionally, since CD7 and CD33 expression on ETP-ALL patient samples is universal with minimal normal tissue distribution, we developed two new second-generation anti-CD7 or anti-CD33 CAR constructs using a 41-BB/CD3ζ backbone. Results: Multiple gene transcripts encoding cell surface proteins potentially amenable to CAR T-cell targeting were overexpressed in ETP-ALL cells in comparison to PBMC controls. Many of these proteins are involved in cell signaling, cell adhesion, and metastasis, and thus potentially important for leukemic cell survival. TSPAN7 (also known as TALLA-1) was the strongest differentially expressed transcript. Despite identification of several transcripts, we did not detect increased surface expression of multiple antigens that were identified as top 25 transcripts, including TALLA-1, MCAM, EPHB6, or TSLPR. Interestingly, TALLA-1 was expressed on the more mature T-cell ALL lines, JURKAT and HPB-AU, suggesting that the surface expression of TALLA protein may be developmentally regulated. Although a new target could not be identified, given the universal expression of CD7 and CD33 on ETP-ALL, we proceeded with development of CARs targeting these antigens. CD33 CAR T-cells had excellent in vitro activity in human AML cell line MOLM-14 with minimal anti-leukemia activity in six tested ETP-ALL PDX models, perhaps due to their lower CD33 expression. We next tested T-cells transduced with a bicistronic CD7-redirected CAR with a truncated EGFR (EGFRt) to facilitate measurement of transduction efficiency and to provide a CAR deletion method. Despite high EGFRt surface expression in transduced T-cells, these CD7 CAR T-cells did not demonstrate in vitro activity against ETP-ALL or mature T-ALL samples despite high CD7 surface expression on all leukemia cell lines. We postulated that abnormal CAR distribution within the T-cell itself could be a potential factor in the observed lack of CD7 CAR T-cell activity. Using fluorescent-labeling to assess CAR surface membrane distribution, we detected high intracellular expression of the CD7 CAR, and noted that it did not traffic to the cell surface. Conclusions: We applied multimodal techniques to evaluate for cell surface expression on ETP-ALL that could serve as a target for immunotherapy. Although novel targets could not be identified, we were able to design an active anti-CD33 CAR. Further studies are in progress to evaluate what degree of antigen expression is needed to be amenable to targeted therapy. Additionally, ongoing studies are assessing whether optimization of CAR design can enhance cell surface trafficking and thereby potentially improve the anti-leukemia efficacy of CD7 CAR T-cells. Disclosures Orentas: Lentigen Technology, Inc.: Employment. Maude:Novartis: Consultancy. Teachey:Novartis: Research Funding.

The paper analyses the migration and ethnocultural processes in the Kosinj Valley in western Lika, which peaked during and immediately after the end of the Great Turkish War at the end of the 17th century and the start of the 18th century. During that time, the area became the intersection of the primary early modern migration flows in this part of the imperial multiple borderlands – southeast and northwest – with their diverse religious (Roman Catholics, Orthodox Christians), linguistic (neo-Shtokavian Ijekavian and Ikavian, Chakavian, and transitional Chakavian- Kajkavian), traditional (Dinaric, Adriatic, Pannonian, Eastern Alpine), and socioeconomic characteristics (frontiersmen/peasants, pastoral/agrarian). Each ethnocultural component was internally heterogeneous, characterised by distinct layers shaped by migrations across the turbulent imperial borderlands from the 16th to the 18th centuries. This led to various interactions and intermingling of cultural traditions, ethnic elements, languages, and dialects. It is concluded that the state military-provincial and Chamber authorities played a dominant role in these processes, directly or indirectly encouraging and directing migration flows for geopolitical and economic reasons. Therefore, from the outset, these migratory movements were characterised by planned military and agrarian colonisation. They favoured Catholic elements originating from Habsburg territories. Phenomenologically, they closely parallel the Vlach colonization of the Kosinj Valley a century earlier, which was conducted and directed by Ottoman authorities. Migrations in the turbulent border area of conflicting empires, with the Kosinj Valley at its centre for over 160 years, exhibited distinct dynamics. It was a region marked by intense emigration and immigration of defectors, including internal, external, or cross-border, and return migrations. The 16th and 17th centuries witnessed intense migration that began to subside only at the beginning of the 18th century. Namely, very few families in the 17th century had the chance to live in one place for three generations. The sons and grandsons of the Croats who fled to Carniola returned to Croatian territory in Gorski Kotar, only for the second generation of returnees to move to Lika. Similar return migrations are also observed among the Croats, Vlachs, and Bunjevci from Ogulin. To understand the settlement process, certain internal aspects of differentiated rural society also need to be considered. It is no coincidence that the elders, who were also leaders of migrations, often acquired the largest plots of land and formed the most numerous families in their new homeland. The same applies to Vlach elder families who, on both sides of the old border, retained their prestigious role as village leaders. In that period, the basic outlines of the modern ethnocultural structure of the Kosinj region were formed. The region became an intersection of primary migration flows in the Early Modern Period: the Vlach-Bunjevac, originating from the deep southeastern Dinaric interior, and the Croatian-Carniolan from the northwest, which also included elements originating in the eastern Alpine region. Their protagonists were early modern ethnic groups such as Vlachs, Croats, Carniolans and Bunjevci, who differed from each other along several criteria: socio-economic status, livelihoods, religious affiliation, dialects, family models and cultural areas. They all had in common that each ethnic group was internally heterogeneous and multi-layered. For instance, among the Vlachs, we observe traces of the old Balkan layer, Croats from Ogulin display Carniolan and Uskok-Vlach elements, Carniolan Croats exhibit Slovenian and German influences, and the Bunjevci include Orthodox Vlach converts. Vlachs from Kosinj historically migrated from East Herzegovina. They were part of the “Glamoč” and “Dinaric” migration flows, much like the majority of the Orthodox Vlach population that settled in Lika during the 16th century. Only a small fraction of the newly arrived population was strategically settled by Ottoman authorities across vast areas along their northwestern border, spanning from the Adriatic hinterland to the Drava River between 1550 and 1560. Serb Orthodoxy, neo-Shtokavian Ijekavian (also known as the East Herzegovina-Krajina dialect), seasonal transhumance, and the pastoral-patriarchal culture of the Dinaric area are elements that made them distinctive and distinguishable from their surroundings. After breaking away from their origins in southeastern Herzegovina during the early migrations of the 15th and 16th centuries, they followed a unique development trajectory that shaped a distinct “krajiški” (borderland) type within this historical ethnocultural group in their new habitats across northwestern Ottoman, and later Habsburg and Venetian territories. In 1689, Croatian Krajišnici (frontiersmen) from the Ogulin captaincy settled in Lower Kosinj, situated on the northern edge of a valley that today comprises the hamlets of Sveti Ivan, Draškovići, Selište, Rudine, Klobučari, and Goljak. They were part of the Croatian Chakavian population, specifically peasant soldiers who congregated around borderland fortresses in Ogulin, Oštarije, and Modruš during the 17th century. Some of their Chakavian linguistic features have been preserved to this day in the hamlets of Goljak, Rudinka, and Selište. The ethno-demographic structure of these Croats, which emerged during the first half of the 17th century, shows a tripartite composition. It comprised natives of Modruš, newcomers and/or returnees from Vinodol and Carniola, along with various branches of the Uskok-Vlach migration flows, predominantly migrants from Senj. In 1689, approximately 40 settler families from Gorski Kotar or the borders of Carniola settled the land along the Bakovac stream, extending all the way to its confluence with the Lika River. This area encompasses present-day Upper Kosinj, including the hamlets of Sušanj, Podjelar, and Poljanka, as well as the region of Bakovac and its hamlets Ribnik and Ruja. In most cases, these were descendants of the Croatian population who fled to Carniola in the 16th century. Therefore, it is more appropri¬ate to refer to this segment of the Catholic population in Kosinj in the pre-modern sense as “Carniolan Croats” despite the presence of some Slovenian and German elements among them. Secondly, it is necessary to differentiate between narrower and broader contexts of the usage of the Carniolan ethnicity within Lika itself. Carniolans, in the narrower sense, are represented by these settlers from Gorski Kotar, who were dependent peasants settled in Lika by the Inner Austrian Court Chamber, which introduced its chamber system there. In a broader sense, this ethnic term became a general label for Catholics in the Karlovac Generalate, particularly among the Orthodox population of Lika and Kordun. Bunjevci, also known as “Catholic Vlachs”, who settled in the Kosinj Valley, belonged to the so-called “Krmpote branch”. They were Catholic Vlachs originating from West Herzegovina who gradually migrated northwest, following the Dinaric migration flow along the Dinara and Velebit mountains during the 16th and 17th centuries. Due to their Herzegovian roots, history of Ottoman subjugation, Vlach social structure, pastoral and patriarchal cultural traditions of the Dinaric area (the culture of “dark cloth and gusle”), and the linguistic features of neo-Shtokavian Ikavian dialect, they exhibited cultural affinities with schismatic Vlachs in Lika. Special attention is given to the historical anthroponymy of the inhabitants of Kosinj, which verifies the aforementioned dynamics. Archival materials, including various documents, lists, and registers, provide valuable insights into their origins, movements, and distribution. Surnames among the Catholic population were documented considerably earlier, allowing for continuous tracing, especially among Croats from Ogulin and Carniola, dating back to the 15th and 16th centuries. The surnames of Orthodox families in Kosinj, and to a lesser extent Bunjevci families, can only be traced from their appearance in the Karlovac Generalate records, which means at best from the 17th century. The structural formation and meaning of surnames in Kosinj largely align with the dialectal and sociocultural characteristics of ethnic groups. Essentially, we can identify two anthroponymic origins: Croatian-Carniolan and Vlach-Bunjevac. It should be noted that surnames became established significantly earlier among Catholics, starting from the decisions of the Council of Trent in 1563, which introduced the practice of church registers for baptisms (Liber baptizatorum), marriages (Liber copulatorum), and deaths (Liber mortuorum). The Roman Ritual of 1614 mandated and required the keeping of parish family books known as Status animarum (“State of Souls”) in each parish. Combining it with other sources makes it significantly easier to reconstruct the migration patterns and dispersion of certain Catholic families of Croats and Carniolans, unlike the Bunjevci and especially the Vlachs, who were under Ottoman rule for a long time. By incorporating certain Croatian families that were once part of the middle and lower nobility, anthroponymic continuity is fully established. The Orthodox Church only began to introduce records of its congregation in the 18th century, prompted by the Habsburg Monarchy. First, through parish Domovni protokoli (Household protocols) in the mid-18th century, which listed Orthodox families in specific parishes, but it was only with the Regulament (Regulation) of 1770 that the introduction of parish registers began, mirroring the practice of the Catholic Church. However, in most Orthodox parishes in the Eparchy of Upper Karlovac, this practice was not adopted until the first half of the 19th century. Therefore, it’s not possible to trace most of Kosinj Vlachs’ early movements or pinpoint their origins before the 18th century. Family traditions and the custom of Slava are not always reliable indicators. For the establishment of surnames among Vlach Orthodox families, and to a lesser extent among Bunjevci, secular military-frontier structures were far more influential than church structures. For military record-keeping purposes, frontier conscriptions were conducted, involving lists of military conscripts, which led to Vlach surnames becoming permanent and hereditary. This probably also explains the higher occurrence of colourful and satirical nicknames, or simply nicknames of differentiated family branches, in their anthroponymic repertoire.

Existing Magnetic resonance imaging (MRI) translation models rely on Generative Adversarial Networks, primarily employing simple convolutional neural networks. Unfortunately, these networks struggle to capture global representations and contextual relationships within MRI images. While the advent of Transformers enables capturing long-range feature dependencies, they often compromise the preservation of local feature details. To address these limitations and enhance both local and global representations, we introduce a novel Dual-Branch Generative Adversarial Network (DBGAN). In this framework, the Transformer branch comprises sparse attention blocks and dense self-attention blocks, allowing for a wider receptive field while simultaneously capturing local and global information. The CNN branch, built with integrated residual convolutional layers, enhances local modeling capabilities. Additionally, we propose a fusion module that cleverly integrates features extracted from both branches. Extensive experimentation on two public datasets and one clinical dataset validates significant performance improvements with DBGAN. On Brats2018, it achieves a 10%improvement in MAE, 3.2% in PSNR, and 4.8% in SSIM for image generation tasks compablack to RegGAN. Notably, the generated MRIs receive positive feedback from radiologists, underscoring the potential of our proposed method as a valuable tool in clinical settings.

Statistical Shape Models (SSMs) are well-established tools for assessing the variability of 3D geometry and for broadening a limited set of shapes. They are widely used in medical imaging due to their ability to model complex geometries and their high efficiency as generative models. The principal step behind these techniques is a registration phase, which, in the case of complex geometries, can be a critical issue due to the correspondence problem, as it necessitates the development of correspondence mapping between shapes. The thoracic aorta, with its high level of morphological complexity, poses a multi-scale deformation problem due to the presence of several branch vessels with varying diameters. Moreover, branch vessels exhibit significant variability in shape, making the correspondence optimization even more challenging. Consequently, existing studies have focused on developing SSMs based only on the main body of the aorta, excluding the supra-aortic vessels from the analysis. In this work, we present a novel non-rigid registration algorithm based on optimizing a differentiable distance function through a modified gradient descent approach. This strategy enables the inclusion of custom, domain-specific constraints in the objective function, which act as landmarks during the registration phase. The algorithm’s registration performance was tested and compared to an alternative Statistical Shape modeling framework, and subsequently used for the development of a comprehensive SSM of the thoracic aorta, including the supra-aortic vessels. The developed SSM was further evaluated against the alternative framework in terms of generalisation, specificity, and compactness to assess its effectiveness.

Recently, we have witnessed substantial success using the deep neural network in many tasks. While there still exists concerns about the explainability of decision-making, it is beneficial for users to discern the defects in the deployed deep models. Existing explainable models either provide the image-level visualization of attention weights or generate textual descriptions as post-hoc justifications. Different from existing models, in this paper, we propose a new interpretation method that explains the image similarity models by salience maps and attribute words. Our interpretation model contains visual salience maps generation and the counterfactual explanation generation. The former has two branches: global identity relevant region discovery and multi-attribute semantic region discovery. Branch one aims to capture the visual evidence supporting the similarity score, which is achieved by computing counterfactual feature maps. Branch two aims to discover semantic regions supporting different attributes, which helps to understand which attributes in an image might change the similarity score. Then, by fusing visual evidence from two branches, we can obtain the salience maps indicating important response evidence. The latter will generate the attribute words that best explain the similarity using the proposed erasing model. The effectiveness of our model is evaluated on the classical face verification task. Experiments conducted on two benchmarks VGG-Face2 and Celeb-A demonstrate that our model can provide convincing interpretable explanations for the similarity. Moreover, our algorithm can be applied to evidential learning cases, e.g. finding the most characteristic attributes in a set of face images and we verify its effectiveness on the VGGFace2 dataset.

Most tracing algorithms have used variants of proportionality based tracing algorithm. However, proportional sharing is not an inherent characteristic of the system. It is a rule enforced to arrive at a solution in a simple and fair manner. We propose a multi-commodity network flow optimization approach for optimal tracing of power flows. The resulting optimization formulation is shown to be sparse linear programming (LP) problem. The paper is devoted to the constraint modeling of this optimization approach. A unified formulation that models lossy MW flow network and provides consistent results for generation and load tracing is presented. Optimal tracing provides complete power audit information like share of loads in generation, generators' contribution in loads, loss allocation to generators and loads, and decomposition of branch flows into generation and load components. Illustrative examples and results on IEEE 30 bus system show the workability of the proposed approach.

Resume. Global social shifts are requiring qualitative growth of staffing education at all levels. In the period of changes, the system of certification training and postgraduate education of managerial, pedagogical and scientific-pedagogical workers of the branch identified as a priority in implementing innovative approaches to the formation of a new generation of educators, development of new strategies, forms, the content of their professional growth. The study proposes to consider diversification of advanced training for education workers at all levels using the consortium resource as a new institutional form of an educational organization. The effectiveness of the proposed model in ensuring the systematic coordination of inter-institutional cooperation in the fields of academic mobility, research, technological development, and innovation, enhancing the competitiveness of national education at the international level had been proved. Following the example of the activities of the Consortium of Postgraduate Education Institutions and the Ukrainian Open University of Postgraduate Education new opportunities for the qualitative development of postgraduate education have been opened. These opportunities are sharing open educational technologies, resources, and repositories, creating new models of collaboration to develop and disseminate educational resources, the purpose of which is to ensure the quality of education and to implement educational activities based on new technology, didactics and the organization of training. The basic directions and models of innovative approach are analysed for education workers training at all education levels in the context of sustainable development goals, the need for diversification of professional development of pedagogical, scientifically pedagogical and managerial education workers in the context of multi-vector interaction of educational institutions through a consortium is investigated and substantiated. It is proved that the introduction of diversified models of institutional organization of education in the digital society is a crucial component of the continuity of professional development of management, pedagogical and scientific-pedagogical workers and education as a whole.

Abstract Aims Post-procedure conduction abnormalities (CA) remain a common complication of transcatheter aortic valve implantation (TAVI), highlighting the need for personalized prediction models. We used machine learning (ML), integrating statistical and mechanistic modelling to provide a patient-specific estimation of the probability of developing CA after TAVI. Methods and results The cohort consisted of 151 patients with normal conduction and no pacemaker at baseline who underwent TAVI in nine European centres. Devices included CoreValve, Evolut R, Evolut PRO, and Lotus. Preoperative multi-slice computed tomography was performed. Virtual valve implantation with patient-specific computer modelling and simulation (CM&S) allowed calculation of valve-induced contact pressure on the anatomy. The primary composite outcome was new onset left or right bundle branch block or permanent pacemaker implantation (PPI) before discharge. A supervised ML approach was applied with eight models predicting CA based on anatomical, procedural and mechanistic data. CA occurred in 59% of patients (n = 89), more often after mechanical than first or second generation self-expanding valves (68% vs. 60% vs. 41%). CM&S revealed significantly higher contact pressure and contact pressure index in patients with CA. The best model achieved 83% accuracy (area under the curve 0.84) and sensitivity, specificity, positive predictive value, negative predictive value, and F1-score of 100%, 62%, 76%, 100%, and 82%. Conclusion ML, integrating statistical and mechanistic modelling, achieved an accurate prediction of CA after TAVI. This study demonstrates the potential of a synergetic approach for personalizing procedure planning, allowing selection of the optimal device and implantation strategy, avoiding new CA and/or PPI.

The intermediate neutron capture process (i-process) can develop during proton ingestion events (PIE), potentially during the early stages of low-mass low-metallicity asymptotic giant branch (AGB) stars. We examine the impact of overshoot mixing on the triggering and development of i-process nucleosynthesis in AGB stars of various initial masses and metallicities. We computed AGB stellar models, with initial masses of 1, 2, 3, and 4 and metallicities in the $-2.5 Fe/H 0$ range, using the stellar evolution code STAREVOL with a network of 1160 nuclei coupled to the transport equations. We considered different overshooting profiles below and above the thermal pulses, and below the convective envelope. The occurrence of PIEs is found to be primarily governed by the amount of overshooting at the top of pulse ($f_ top $) and to increase with rising $f_ top $. For $f_ top = 0$, 0.02, 0.04, and 0.1, we find that 0 <!PCT!>, 6 <!PCT!>, 24 <!PCT!>, and 86 <!PCT!> of our 21 AGB models with $-2<$ Fe/H $<0$ experience a PIE, respectively. Variations of the overshooting parameters during a PIE leads to a scatter on abundances of $0.5 - 1$ dex on elements, with $36<Z<56;$ however, this barely impacts the production of elements with $56<Z<80$, which therefore appear to be a reliable prediction of our models. Actinides are only produced if the overshooting at the top of pulse is small enough. We also find that PIEs leave a 13C-pocket at the bottom of the pulse that can give rise to an additional radiative s-process nucleosynthesis. In the case of the 2 models with Fe/H $=-1$ and $-0.5$, it produces a noticeable mixed i+s chemical signature at the surface. Finally, the chemical abundance patterns of 22 observed r/s-stars candidates (18 dwarfs or giants and 4 post-AGB) with $-2<$ Fe/H $<-1$ are found to be in reasonable agreement with our AGB model predictions. The binary status of the dwarfs/giants being unclear, we suggest that these stars have acquired their chemical pattern either from the mass transfer of a now-extinct AGB companion or from an early generation AGB star that polluted the natal cloud. The occurrence of PIEs and the development of i-process nucleosynthesis in AGB stars remains sensitive to the overshooting parametrization. A high (yet realistic) $f_ top $ value triggers PIEs at (almost) all metallicities. The existence of r/s-stars at Fe/H -1$ is in favour of an i-process operating in AGB stars up to this metallicity. Stricter constraints from multi-dimensional hydrodynamical models on overshoot coefficients could deliver new insights into the contribution of AGB stars to heavy elements in the Universe.

Abstract Background and Aims Peritoneal dialysis (PD) is a renal replacement therapy that enables metabolic waste products and excess fluids to be eliminated through the peritoneal membrane. Exposure to conventional dialysates at high glucose concentrations is considered critical for the pathogenesis of peritoneal fibrosis, angiogenesis, and epithelial-mesenchymal transition (EMT). These events largely contribute to peritoneal membrane aging, resulting in ultrafiltration and clearance failure. Pre-clinical research in this field suffers from the lack of valid and reproducible in vivo systems and so far has been limited to in vitro systems of mesothelial or endothelial cell lines. In vitro studies on mesothelial cell monolayers have validated the significant protective effects of dialysates with alternative osmo-metabolic agents. With their better biocompatibility these novel formulations could provide a valid substitute for conventional solutions (1,2). The method proposed, based on multi-photon microscopy, aims to study the physiology of the peritoneal membrane during dialysis exchange and to validate the effects of biocompatible dialysates in animal models of fibrosis as observed in the course of dialysis treatment. Method We have implemented a surgical procedure to optimize the stability of a flap of parietal peritoneum to allow direct microscope observation. The innovative technology of multi-photon microscopy enables one to observe the vessels of microcirculation in vivo and provides a 3D rendering of the peritoneal membrane, giving an overview of all the single layers without the use of specific markers (Figure 1). One may also assess specific phenomena induced by the fibrotic process, such as the thickening of the sub-mesothelial interstitium, exploiting the autofluorescence signal from excited collagen fibers caused by the physical phenomenon of Second Harmonic generation. In vivo microscopy evaluation of peritoneal membrane senescence parameters was conducted in rats subjected to one daily intraperitoneal injection (10 mL/day) for 15 days with one of the following PD solutions: a commercially available glucose-based neutral-pH, low-GDP, PD solution (Physioneal 3.86%, Baxter Healthcare, Italy), and a new single-chamber low glucose PD solution containing osmo-metabolic agents such as L-carnitine and xylitol (XyloCore HS, Galenica Senese, Italy). XyloCore HS is a lactate-buffered PD solution with glucose (1.5%), L-carnitine (0.02%) and xylitol (2.0%). The osmotic strength of the two PD solutions tested was comparable. Results Treatment with XyloCore HS was associated with a significant reduction in thickness of the sub-mesothelial interstitium (p = 0.013), the density of collagen fibers (p = 0.012) and the vascular composition (p = 0.006), as well as the number of branch points (p = 0.0335), when compared to rats treated with a commercial glucose-based PD solution, Physioneal. Figure 2 shows the different thicknesses of sub-mesothelial stroma upon treatment with the PD solutions under investigation. Conclusion Previous in vitro studies have shown that XyloCore (1,2), a novel glucose-sparing PD solution currently in Phase III clinical development (ELIXIR - NCT03994471), is able to counteract the glucotoxic effects on the peritoneal membrane induced by conventional dialysates. Our pre-clinical in vivo methodology not only confirms our previous in vitro findings, but also suggests that long-term protective effects may be achieved with XyloCore. Further studies are in progress in a diabetic animal model to extend the favorable peritoneal in vivo findings of XyloCore treatment to a systemic level, possibly by improving glycemic control.

Fig 1: Image from a secret Rangoon fashion shoot. Photograph: Myanmar Photo Archive / Lukas Birk. Introduction NOTE: Rangoon, Burma has been known as Yangon, Myanmar, since 2006. I use Rangoon and Burma for the period prior to 2006 and Yangon and Myanmar for the period thereafter. In addition, I have removed the name of any activist currently in Myanmar due to the recent policy of executing political prisoners. On 1 February 2021, Myanmar was again plunged into political turmoil when the military illegally overthrew the country’s democratically elected government. This is the third time Myanmar, formally known as Burma, has been subject to a coup d’état; violent seizures of power took place in 1962 and in 1988-90. While those two earlier military governments met with opposition spearheaded by students and student organisations, in 2021 the military faced organised resistance through a mass Civil Disobedience Movement (CDM) initiated by government healthcare workers who refused to come to work. They were joined by private sector “strikes” and, perhaps most visible of all to western viewers, mass street demonstrations “led” by “Gen Z” activists—young people who had come of age during Myanmar’s brief decade of democracy. There is little doubt that the success of the CDM and associated protests is due to the widespread coverage and reach of social media as well as the creative communications skills of the country’s first “generation of digital natives”, who are sufficiently familiar and comfortable with social platforms to “participate and shape their identities in communication and dialogue with global digital media content” (Jordt et al. 12 ). The leveraging of global culture, including the use of English in protest signs, was notable in garnering international media coverage and so keeping Myanmar’s political plight front-of-mind with governments around the world. Yet this is not the whole story behind the effectiveness of these campaigns. As Lisa Brooten argues, contemporary networks are built on “decades of behind-the-scenes activism to build a multi-ethnic civil society” (East Asia Forum). The leading democracy activist, Min Ko Naing, aligned “veteran activists from previous generations with novice Gen Z activists”, declaring “this revolution represents a combination of Generations X, Y and Z in fighting against the military dictatorship’” (Jordt et al. 18). Similarly, the creative strategies used by 2021’s digital campaigners also build on protests by earlier generations of young, creative people. This paper looks at two creative protest across the generations. The first is “secret” fashion photography of the late 1970s collected in Lukas Birk’s Yangon Fashion 1979 – Fashion=Resistance. The second is the contemporary #100projectors campaign, a “projection project for Myanmar democracy movement against the military dictatorship” (in the interest of full disclosure, I took part in the #100projectors project). Drawing from the contemporary advertising principle of “segmentation”, the communications practice where potential consumers are divided into “subgroups … based on specific characteristics and needs” (WARC 1), as well as contemporary thinking on the “aesthetics” of “cosmopolitanism”, (Papastergiadis, Featherstone, and Christensen), I argue that contemporary creative strategies can be traced back to the creative tactics of resistance employed by earlier generations of protesters and their re-imagining of “national space and its politics” (Christensen 556) in the interstices of cosmopolitan Rangoon, Burma, and Yangon, Myanmar. #100projectors Myanmar experienced two distinct periods of military rule, the Socialist era between 1962 and 1988 under General Ne Win and the era under the State Law and Order Restoration Council – State Peace and Development Council between 1988 and 2011. These were followed by a semi-civilian era from 2011 to 2021 (Carlson 117). The coup in 2021 marks a return to extreme forms of control, censorship, and surveillance. Ne Win’s era of military rule saw a push for Burmanisation enforced through “significant cultural restrictions”, ostensibly to protect national culture and unity, but more likely to “limit opportunities for internal dissent” (Carlson 117). Cultural restrictions applied to art, literature, film, television, as well as dress. Despite these prohibitions, in the 1970s Rangoon's young people smuggled in illegal western fashion magazines, such as Cosmopolitan and Vogue, and commissioned local tailors to make up the clothes they saw there. Bell-bottoms, mini-skirts, western-style suits were worn in “secret” fashion shoots, with the models posing for portraits at Rangoon photographic studios such as the Sino-Burmese owned Har Si Yone in Chinatown. Some of the wealthier fashionistas even came for weekly shoots. Demand was so high, a second branch devoted to these photographic sessions was opened with its own stock of costumes and accessories. Copies of these head to toe fashion portraits, printed on 12 x 4 cm paper, were shared with friends and family; keeping portrait albums was a popular practice in Burma and had been since the 1920s and 30s (Birk, Burmese Photographers 113). The photos that survive this era are collected in Lukas Birk’s Yangon Fashion 1979 – Fashion=Resistance. #100projectors was launched in February 2021 by a group of young visual and video artists with the aim of resisting the coup and demanding the return of democracy. Initially a small group of projectionists or “projector fighters”, as the title suggests they plan to amplify their voices by growing their national and international network to 100. #100projectors is one of many campaigns, movements, and fundraisers devised by artists and creatives to protest the coup and advocate for revolution in Myanmar. Other notable examples, all run by Gen Z activists, include the Easter Egg, Watermelon, Flash, and Marching Shoes strikes. The Marching Shoe Strike, which featured images of flowers in shoes, representing those who had died in protests, achieved a reach of 65.2 million in country with 1.4 million interactions across digital channels (VERO, 64) and all of these campaigns were covered by the international press, including The Guardian, Reuters, The Straits Times, and VOA East Asia Pacific Session, as well as arts magazines around the world (for example Hyperallergic, published in Brooklyn). #100projectors material has been projected in Finland, Scotland, and Australia. The campaign was written about in various art magazines and their Video #7 was screened at the Bangkok Art and Culture Centre in February 2022 as part of Defiant Art: A Year of Resistance to the Myanmar Coup. At first glance, these two examples seem distant in both their aims and achievements. Fashion photos, taken in secret and shared privately, could be more accurately described as a grassroots social practice rather than a political movement. While Birk describes the act of taking these images as “a rebellion” and “an escape” in a political climate when “a pair of flowers and a pair of sunglasses might just start a revolution”, the fashionistas’ photographs seem “ephemeral” at best, or what Mina Roces describes as the subtlest form of resistance or ‘weapons of the weak’ (Scott in Roces 7). By contrast, #100projectors has all the hallmarks of a polished communications campaign. They have a logo and slogans: “We fight for light” and “The revolution must win”. There is a media plan, which includes the use of digital channels, encrypted messaging, live broadcasts, as well as in-situ projections. Finally, there is a carefully “targeted” audience of potential projectionists. It is this process of defining a target audience, based on segmentation, that is particularly astute and sophisticated. Traditionally, segmentation defined audiences based on demographics, geodemographics, and self-identification. However, in the online era segments are more likely to be based on behaviour and activities revealed in search data as well as shares, depending on preferences for privacy and permission. Put another way, as a digital subject, “you are what you choose to share” (WARC 1). The audience for #100projectors includes artists and creative people around the world who choose to share political video art. They are connected through digital platforms including Facebook as well as encrypted messaging. Yet this contemporary description of digital subjectivity, “you are what you choose to share”, also neatly describes the Yangon fashionistas and the ways in which they resist the political status quo. Photographic portraits have always been popular in Burma and so this collection does not look especially radical. Initially, the portraits seem to speak only about status, taste, and modernity. Several subjects within the collection are shown in national or ethnic dress, in keeping with the governments edict that Burma consisted of 135 ethnicities and 8 official races. In addition, there is a portrait of a soldier in full uniform. But the majority of the images are of men and women in “modern” western gear typical of the 1970s. With their wide smiles and careful poses, these men and women look like they’re performing sophisticated worldliness as well as showing off their wealth. They are cosmopolitan adepts taking part in international culture. Status is implicit in the accessories, from sunglasses to jewellery. One portrait is shot at mid-range so that it clearly features a landline phone. In 1970s Burma, this was an object out of reach for most. Landlines were both prohibitively expensive and reserved for the true elites. To make a phone call, most people had to line up at special market stalls. To be photographed with a phone, in western clothes (to be photographed at all), seems more about aspiration than anarchy. In the context of Ne Win’s Burma, however, the portraits clearly capture a form of political agency. Burma had strict edicts for dress and comportment: kissing in public was banned and Burmese citizens were obliged to wear Burmese dress, with western styles considered degenerate. Long hair, despite being what Burmese men traditionally wore prior to colonisation, was also deemed too western and consequently “outlawed” (Edwards 133). Dress was not only proscribed but hierarchised and heavily gendered; only military men had “the right to wear trousers” (Edwards 133). Public disrespect of the all-powerful, paranoid, and vindictive military (known as “sit tat” for military or army versus “Tatmadaw” for the good Myanmar army) was dangerous bordering on the suicidal. Consequently, wearing shoulder-length hair, wide bell bottoms, western-style suits, and “risqué” mini-skirts could all be considered acts of at least daring and definitely defiance. Not only are these photographs a challenge to gender constructions in a country ruled by a hyper-masculine army, but these images also question the nature of what it meant to be Burmese at a time when Burmeseness itself was rigidly codified. Recording such acts on film and then sharing the images entailed further risk. Thus, these models are, as Mina Roces puts it, “express[ing] their agency through sartorial change” (Roces 5). Fig. 2: Image from a secret Rangoon fashion shoot – illicit dress and hair. Photograph: Myanmar Photo Archive / Lukas Birk. Fig. 3: Image from a secret Rangoon fashion shoot. Photograph: Myanmar Photo Archive / Lukas Birk. Roces also notes the “challenge” of making protest visible in spaces “severely limited” under authoritarian regimes (Roces 10). Burma under the Socialist government was a particularly difficult place in which to mount any form of resistance. Consequences included imprisonment or even execution, as in the case of the student leader Tin Maung Oo. Ma Thida, a writer and human rights advocate herself jailed for her work, explains the use of creative tools such as metaphor in a famous story about a crab by the writer and journalist Hanthawaddy U Win Tin: The crab, being hard-shelled, was well protected and could not be harmed. However, the mosquito, despite being a far smaller animal, could bite the eyes of the crab, leading to the crab’s eventual death. ... Readers drew the conclusion that the socialist government of Ne Win was the crab that could be destabilized if a weakness could be found. (Thida 317) If the metaphor of a crab defeated by a mosquito held political meaning, then being photographed in prohibited fashions was a more overt way of making defiance and resistant “visible”. While that visibility seems ephemeral, the fashionistas also found a way not only to be seen by the camera in their rebellious clothing, but also by a “public” or audience of those with whom they shared their images. The act of exchanging portraits, what Birk describes as “old-school Instagram”, anticipates not only the shared selfie, but also the basis of successful contemporary social campaigns, which relied in part on networks sharing posts to amplify their message (Birk, Yangon Fashion 17). What the fashionistas also demonstrate is that an act of rebellion can also be a means of testing the limits of conformity, of the need for beauty, of the human desire to look beautiful. Acts of rebellion are also acts of celebration and so, solidarity. Fig. 4: Image from a secret Rangoon fashion shoot – illicit dress length. Photograph: Myanmar Photo Archive / Lukas Birk. Fig. 5: Image from a secret Rangoon fashion shoot – illicit trousers. Photograph: Myanmar Photo Archive / Lukas Birk. As the art critic and cultural theorist Nikos Papastergiadis writes, “the cosmopolitan imagination in contemporary art could be defined as an aesthetic of openness that engenders a global sense of inter-connectedness” (207). Inter-connectedness and its possibilities and limits shape the aesthetic imaginary of both the secret fashion shoots of 1970s Rangoon and the artists and videographers of 2021. In the videos of the #100projectors project and the fashion portraits of stylish Rangoonites, interconnection comes as a form of aesthetic blending, a conversation that transcends the border. The sitter posing in illicit western clothes in a photo studio in the heart of Rangoon, then Burma’s capital and seat of power, cannot help but point out that borders are permeable, and that national identity is temporally-based, transitory, and full of slippages. In this spot, 40-odd years earlier, Burmese nationalists used dress as a means of publicly supporting the nationalist cause (Edwards, Roces). Like the portraits, the #100projector videos blend global and local perspectives on Myanmar. Combining paintings, drawings, graphics, performance art recordings, as well as photography, the work shares the ‘instagrammable’ quality of the Easter Egg, Watermelon, and Marching Shoes strikes with their bright colours and focus on people—or the conspicuous lack of people and the example of the Silent Strike. Graphics are in Burmese as well as English. Video #6 was linked to International Women’s Day. Other graphics reference American artists such as Shepherd Fairey and his Hope poster, which was adapted to feature Aung San Suu Kyi’s face during then-President Obama’s visit in 2012. The videos also include direct messages related to political entities such as Video #3, which voiced support for the Committee Representing Pyidaungsu Hlutaw (CRPH), a group of 15 elected MPs who represented the ideals of Gen Z youth (Jordt et al., viii). This would not necessarily be understood by an international viewer. Also of note is the prevalence of the colour red, associated with Aung San Suu Kyi’s NLD. Red is one of the three “political” colours formerly banned from paintings under SLORC. The other two were white, associated with the flowers Aung Sang Suu Kyi wore in her hair, and black, symbolic of negative feelings towards the regime (Carlson, 145). The Burmese master Aung Myint chose to paint exclusively in the banned colours as an ongoing act of defiance, and these videos reflect that history. The videos and portraits may propose that culturally, the world is interconnected. But implicit in this position is also the failure of “interconnectedness”. The question that arises with every viewing of a video or Instagram post or Facebook plea or groovy portrait is: what can these protesters, despite the risks they are prepared to take, realistically expect from the rest of the world in terms of help to remove the unwanted military government? Interconnected or not, political misfortune is the most effective form of national border. Perhaps the most powerful imaginative association with both the #100projectors video projections and fashionistas portraits is the promise of transformation, in particular the transformations possible in a city like Rangoon / Yangon. In his discussion of the cosmopolitan space of the city, Christensen notes that although “digital transformations touch vast swathes of political, economic and everyday life”, it is the city that retains supreme significance as a space not easily reducible to an entity beneath the national, regional, or global (556). The city is dynamic, “governed by the structural forces of politics and economy as well as moralities and solidarities of both conservative and liberal sorts”, where “othered voices and imaginaries find presence” in a mix that leads to “contestations” (556). Both the fashionistas and the video artists of the #100projectors use their creative work to contest the ‘national’ space from the interstices of the city. In the studio these transformations of the bodies of Burmese subjects into international “citizens of the world” contest Ne Win’s Burma and reimagine the idea of nation. They take place in the Chinatown, a relic of the old, colonial Rangoon, a plural city and one of the world’s largest migrant ports, where "mobility, foreignness and cross-cultural hybridity" were essential to its make-up (Aung Thin 778). In their instructions on how to project their ideas as a form of public art to gain audience, the #100projectors artists suggest projectors get “full on creative with other ways: projecting on people, outdoor cinema, gallery projection” (#100projectors). It is this idea projection as an overlay, a doubling of the everyday that evokes the possibility of transformation. The #100projector videos screen on Rangoon bridges, reconfiguring the city, albeit temporarily. Meanwhile, Rangoon is doubled onto other cities, towns, villages, communities, projected onto screens but also walls, fences, the sides of buildings in Finland, Scotland, Australia, and elsewhere. Conclusion In this article I have compared the recent #100projectors creative campaign of resistance against the 2021 coup d’état in Myanmar with the “fashionistas” of 1970 and their “secret” photo shoots. While the #100projectors is a contemporary digital campaign, some of the creative tactics employed, such as dissemination and identifying audiences, can be traced back to the practices of Rangoon’s fashionistas of the 1970s. ­­Creative resistance begins with an act of imagination. The creative strategies of resistance examined here share certain imaginative qualities of connection, a privileging of the ‘cosmopolitan’ and ‘interconnectedness’ as well as the transformativity of actual space, with the streets of Rangoon, itself a cosmopolitan city. References @100projectors Instagram account. <https://www.instagram.com/100projectors/>. @Artphy_1 Instagram account. <https://www.instagram.com/artphy_1/>. Aung Thin, Michelle. “Sensations of Rootedness’ in Cosmopolitan Rangoon or How the Politics of Authenticity Shaped Colonial Imaginings of Home.” Journal of Intercultural Studies 41.6 (2020): 778-792. Birk, Lukas. Yangon Fashion 1979 – Fashion=Resistance. France: Fraglich Publishing, 2020. ———. Burmese Photographers. Myanmar: Goethe-Institut Myanmar, 2018. Brooten, Lisa. “Power Grab in a Pandemic: Media, Lawfare and Policy in Myanmar.” Journal of Digital Media & Policy 13.1 (2022): 9-24. ———. “Myanmar’s Civil Disobedience Movement Is Built on Decades of Struggle.” East Asia Forum, 29 Mar. 2021. 29 July 2022 <https://www.eastasiaforum.org/2021/03/29/myanmars-civil-disobedience-movement-is-built-on-decades-of-struggle/>. Carlson, Melissa. “Painting as Cipher: Censorship of the Visual Arts in Post-1988 Myanmar.” Sojourner: Journal of Social Issues in Southeast Asia 31.1 (2016): 116-72. Christensen, Miyase. “Postnormative Cosmopolitanism: Voice, Space and Politics.” The International Communication Gazette 79.6–7 (2017): 555–563. Edwards, Penny. “Dressed in a Little Brief Authority: Clothing the Body Politic in Burma.” In Mina Roces & Louise Edwards (eds), The Politics of Dress in Asia and the Americas. Brighton: Sussex Academic Press, 121–138. France24. “‘Longyi Revolution’: Why Myanmar Protesters Are Using Women’s Clothes as Protection.” 10 Mar. 2021. <https://youtu.be/ebh1A0xOkDw>. Ferguson, Jane. “Who’s Counting? Ethnicity, Belonging, and the National Census in Burma/Myanmar.” Bijdragen tot de Taal-, Land- en Volkenkunde 171 (2015): 1–28. Htun Khaing. “Salai Tin Maung Oo, Defiant at the End.” Frontier, 24 July 2017. 1 Aug. 2022 <https://www.frontiermyanmar.net/en/salai-tin-maung-oo-defiant-to-the-end>. Htun, Pwin, and Paula Bock. “Op-Ed: How Women Are Defying Myanmar’s Junta with Sarongs and Cellphones.” Los Angeles Times, 16 Mar. 2021. <https://www.latimes.com/opinion/story/2021-03-16/myanmar-military-women-longyi-protests>. Jordt, Ingrid, Tharaphi Than, and Sue Ye Lin. How Generation Z Galvanized a Revolutionary Movement against Myanmar’s 2021 Military Coup. Singapore: Trends in Southeast Asia ISEAS – Yusof Ishak Institute, 2021. Ma Thida. “A ‘Fierce’ Fear: Literature and Loathing after the Junta.” In Myanmar Media in Transition: Legacies, Challenges and Change. Eds. Lisa Brooten, Jane Madlyn McElhone, and Gayathry Venkiteswaran. Singapore: ISEAS - Yusof Ishak Institute, 2019. 315-323. Myanmar Poster Campaign (@myanmarpostercampaign). “Silent Strike on Feb 1, 2022. We do not forget Feb 1, 2021. We do not forget about the coup. And we do not forgive.” Instagram. <https://www.instagram.com/p/CZJ5gg6vxZw/>. Papastergiadias, Nikos. “Aesthetic Cosmopolitanism.” In Routledge International Handbook of Cosmopolitanism Studies. Ed. Gerard Delanty. London: Routledge, 2018. 198-210. Roces, Mina. “Dress as Symbolic Resistance in Asia.” International Quarterly for Asian Studies 53.1 (2022): 5-14. Smith, Emiline. “In Myanmar, Protests Harness Creativity and Humor.” Hyperallergic, 12 Apr. 2021. 29 July 2022 <https://hyperallergic.com/637088/myanmar-protests-harness-creativity-and-humor/>. Thin Zar (@Thinzar_313). “Easter Egg Strike.” Instagram. <https://www.instagram.com/p/CNPfvtAMSom/>. VERO. “Myanmar Communication Landscape”. 10 Feb. 2021. <https://vero-asean.com/a-briefing-about-the-current-situation-in-myanmar-for-our-clients-partners-and-friends/>. World Advertising Research Centre (WARC). “What We Know about Segmentation.” WARC Best Practice, May 2021. <https://www-warc-com.ezproxy.lib.rmit.edu.au/content/article/bestprac/what-we-know-about-segmentation/110142>.