To see the other types of publications on this topic, follow the link: Text Generation Using Neural Networks.
Journal articles on the topic 'Text Generation Using Neural Networks'
Create a spot-on reference in APA, MLA, Chicago, Harvard, and other styles
Consult the top 50 journal articles for your research on the topic 'Text Generation Using Neural Networks.'
Next to every source in the list of references, there is an 'Add to bibliography' button. Press on it, and we will generate automatically the bibliographic reference to the chosen work in the citation style you need: APA, MLA, Harvard, Chicago, Vancouver, etc.
You can also download the full text of the academic publication as pdf and read online its abstract whenever available in the metadata.
Browse journal articles on a wide variety of disciplines and organise your bibliography correctly.
1
Chakravarty, Aniv, and Jagadish S. Kallimani. "Unsupervised Multi-Document Abstractive Summarization Using Recursive Neural Network with Attention Mechanism." Journal of Computational and Theoretical Nanoscience 17, no. 9 (2020): 3867–72. http://dx.doi.org/10.1166/jctn.2020.8976.
Full textAbstract:
Text summarization is an active field of research with a goal to provide short and meaningful gists from large amount of text documents. Extractive text summarization methods have been extensively studied where text is extracted from the documents to build summaries. There are various type of multi document ranging from different formats to domains and topics. With the recent advancement in technology and use of neural networks for text generation, interest for research in abstractive text summarization has increased significantly. The use of graph based methods which handle semantic information has shown significant results. When given a set of documents of English text files, we make use of abstractive method and predicate argument structures to retrieve necessary text information and pass it through a neural network for text generation. Recurrent neural networks are a subtype of recursive neural networks which try to predict the next sequence based on the current state and considering the information from previous states. The use of neural networks allows generation of summaries for long text sentences as well. This paper implements a semantic based filtering approach using a similarity matrix while keeping all stop-words. The similarity is calculated using semantic concepts and Jiang–Conrath similarity and making use of a recurrent neural network with an attention mechanism to generate summary. ROUGE score is used for measuring accuracy, precision and recall scores.
2
Zatsepina, Aleksandra, Galina Bardina, and Polina Shindina. "Eco-sensitive site assessment: Integrating neural networks for environmentally conscious pre-project planning." E3S Web of Conferences 614 (2025): 05003. https://doi.org/10.1051/e3sconf/202561405003.
Full textAbstract:
The use of neural networks in the architectural design pre-phase is becoming increasingly prevalent among designers. This article presents a method of textual and graphical analysis of construction sites using neural networks. On the example of two projects, which won the architectural competition, the following are considered: qualitative characteristics analysis of the territory using Autodesk Forma software, cultural context analysis using ChatGPT, image generation using MidJourney and 3D-models generation using the neural network Meshy.ai. In regard to ChatGPT, the "risks" method is presented as a means of achieving optimal results. The possibility of factual errors in ChatGPT text generations is indicated.
3
Chary, Podakanti Satyajith. "Text Generation: Using Markov Model & LSTM Networks to Generate Realistic Text." International Journal for Research in Applied Science and Engineering Technology 11, no. 12 (2023): 1323–27. http://dx.doi.org/10.22214/ijraset.2023.57601.
Full textAbstract:
Abstract: Text generation plays a crucial role in various natural language processing applications, ranging from creative writing to chatbots. This research delves into the realm of text generation by exploring and comparing two distinct techniques: Markov models and Long Short-Term Memory (LSTM) networks. The study focuses on their ability to generate realistic text within specific styles or genres, providing valuable insights into their respective strengths and limitations. Markov models, rooted in probability theory, and LSTM networks, a type of recurrent neural network, represent contrasting approaches to text generation. The research employs these techniques on a carefully curated dataset, evaluating their performance based on coherence, style, and contextual relevance. The comparison aims to elucidate the nuanced differences in how these models capture dependencies within the data and their effectiveness in simulating authentic linguistic patterns. Through rigorous experimentation, this research investigates the intricacies of both Markov models and LSTM networks, shedding light on their individual contributions to the task of text generation. The examination extends beyond mere algorithmic efficacy, considering the impact of these techniques on the quality and diversity of the generated text. Additionally, the study explores the influence of hyperparameters, such as temperature in the context of LSTM networks, on the output's richness and variability.
4
Li, Zixiang, Qinfeng Wu, and Longjie Zhong. "Controllable text generation based on varied frameworks - rhyming lyrics generation technology." ITM Web of Conferences 73 (2025): 02012. https://doi.org/10.1051/itmconf/20257302012.
Full textAbstract:
Controllable Text Generation, as a cutting-edge technology in Natural Language Processing (NLP), has significantly improved the quality of text generation. Users can customize the generated content by setting specific attributes, formats, and emotional characteristics, thereby achieving the goal of conserving resources. However, despite notable progress in this field, several challenges remain, such as limited text diversity under multiple conditions and information disconnection during long-text generation. In light of this, this paper focuses on controllable text generation technology within a Chinese context, particularly emphasizing the key element of rhyming. The aim is to investigate an effective method for generating rhyming lyrics and poetry. By comparing the text generation performance under Recurrent Neural Networks (RNN), Bidirectional Recurrent Neural Networks (Bi-RNN), and Transformer frameworks, and evaluating the results using n-grams metrics, this study attempts to reveal which architecture is better suited for handling the specific controllable generation requirement of rhyming. This provides theoretical support and technical guidance for automatically creating Chinese poetry and lyrics.
5
Irina, A. Kiseleva, V. Borisova Yulia, and Yu. Maevskaya Anna. "Analysing the feasibility of using a neural network for generating English language assignments." Perspektivy Nauki i Obrazovania, no. 1 (February 28, 2025): 319–35. https://doi.org/10.32744/pse.2025.1.21.
Full textAbstract:
Introduction. Today neural networks are used in various fields, being able to improve the quality of work and significantly reduce the time spent on routine tasks. The application of neural networks in education is actively explored, but at the moment the research in the field of task generation for classroom teaching of foreign language students is scanty, which accounts for the relevance of this article. The aim of the work is to analyse the possibility of using neural networks as a tool for developing tasks for current control of students’ English-language knowledge. Methods and materials. The results were analysed using qualitative analysis, error analysis and performance analysis. The YandexGPT 2 neural network’s response to the queries for creating current control assignments was analysed, in particular, tasks for consolidating the studied vocabulary and test papers consisting of several parts. The survey involving 1st-year students was implemented at the Empress Catherine II Saint Petersburg Mining University. KEYWORDS Results. Based on the results of the analysis, the queries were modified to obtain an answer that would require minimal adjustment by the teacher. Although it was not possible to create queries that would consistently yield a good result, the neural network’s answers were successfully used for current control, making it possible to quickly prepare several versions of test papers aimed at consolidating the studied vocabulary and grammar. Thus, the survey explored the possibility of using neural networks for devising traditional English language tasks that are still in demand at the training sessions. Conclusion. The research showed that the use of neural networks for creating materials for current control significantly reduces the time for their development and allows the teacher to adapt the materials for each group of students, which improves their quality. The findings can be useful for the development of other types of teaching materials.
6
Al Aziz, Md Momin, Tanbir Ahmed, Tasnia Faequa, Xiaoqian Jiang, Yiyu Yao, and Noman Mohammed. "Differentially Private Medical Texts Generation Using Generative Neural Networks." ACM Transactions on Computing for Healthcare 3, no. 1 (2022): 1–27. http://dx.doi.org/10.1145/3469035.
Full textAbstract:
Technological advancements in data science have offered us affordable storage and efficient algorithms to query a large volume of data. Our health records are a significant part of this data, which is pivotal for healthcare providers and can be utilized in our well-being. The clinical note in electronic health records is one such category that collects a patient’s complete medical information during different timesteps of patient care available in the form of free-texts. Thus, these unstructured textual notes contain events from a patient’s admission to discharge, which can prove to be significant for future medical decisions. However, since these texts also contain sensitive information about the patient and the attending medical professionals, such notes cannot be shared publicly. This privacy issue has thwarted timely discoveries on this plethora of untapped information. Therefore, in this work, we intend to generate synthetic medical texts from a private or sanitized (de-identified) clinical text corpus and analyze their utility rigorously in different metrics and levels. Experimental results promote the applicability of our generated data as it achieves more than 80\% accuracy in different pragmatic classification problems and matches (or outperforms) the original text data.
7
Prof., Aarthy G., Shenoy K. Vibha, H. Thejashree, and S. Nithin. "Image based Spam Detection using Recurrent Neural Networks (RNN)." Recent Innovations in Wireless Network Security 5, no. 2 (2023): 20–31. https://doi.org/10.5281/zenodo.8141409.
Full textAbstract:
<em>Image-spam initially arose as a way of bypassing text-based spam filters. It is widely used to advertise products, mislead individuals into providing personal information, or transmit hazardous viruses. Image spam is harder to detect than text-based spam. Image-based encryption methods can be used to create image spam that is even more difficult to detect than what is often seen in reality. Image spam has evolved over time and may now overcome various kinds of classic anti-spam methods. Spammers can utilise pictures that just include text, sliced images, and randomly created images. Text-only images were used in the initial generation of image spam. Such images are practically empty, containing only pure text. Such text can be retrieved using optical character recognition (OCR), and then processed using normal text-based filters</em>
8
Naga, Sai Krishna Mohan Pitchikala, Kodakondla Saisuhas, and Ghosh Debargha. "Word Generation Using Recurrent Neural Network." Journal of Scientific and Engineering Research 7, no. 1 (2020): 309–15. https://doi.org/10.5281/zenodo.14058872.
Full textAbstract:
Computers have influenced the life of humans to a very great extent. Natural Processing language is a field of computer science which helps to exchange information very efficiently between humans and machines with less human requirement. Text generation techniques can be applied for improving language models, machine translation summarizing and captioning. In this project we train a Recurrent Neural Network so that it can generate new words related to the words that are fed to it.
9
Murad, Aylu. "Using Neural Network Principles for Learning the English Language." Eurasian Science Review An International peer-reviewed multidisciplinary journal 1, no. 3 (2025): 2262–74. https://doi.org/10.63034/esr-380.
Full textAbstract:
The scientific project on the topic "Using Neural Network Principles for Learning the English Language" is dedicated to the research of applying neural networks in educational technologies for studying the English language. The aim of the project is to analyze and develop methods that utilize neural networks to improve the effectiveness of English language learning. The project begins with a theoretical overview of neural networks, including their basic principles of operation and learning algorithms. Key components of neural networks, such as neurons, layers, and activation functions, as well as learning methods like the backpropagation algorithm, are discussed. Special attention is given to existing technologies and applications that use neural networks for language learning, such as machine translation, automatic error correction, and text generation. In conclusion, the project formulates key findings and recommendations for improving existing systems and developing new solutions. The project highlights the potential of neural networks in the educational field and opens up prospects for further research in the application of artificial intelligence in foreign language learning.
10
Orlando, Iparraguirre-Villanuev, Guevara-Ponce Victor, Ruiz-Alvarado Daniel, et al. "Text prediction recurrent neural networks using long shortterm memory-dropout." Text prediction recurrent neural networks using long shortterm memory-dropout 29, no. 3 (2023): 1758–68. https://doi.org/10.11591/ijeecs.v29.i3.pp1758-1768.
Full textAbstract:
Unit short-term memory (LSTM) is a type of recurrent neural network (RNN) whose sequence-based models are being used in text generation and/or prediction tasks, question answering, and classification systems due to their ability to learn long-term dependencies. The present research integrates the LSTM network and dropout technique to generate a text from a corpus as input, a model is developed to find the best way to extract the words from the context. For training the model, the poem "La Ciudad y los perros" which is composed of 128,600 words is used as input data. The poem was divided into two data sets, 38.88% for training and the remaining 61.12% for testing the model. The proposed model was tested in two variants: word importance and context. The results were evaluated in terms of the semantic proximity of the generated text to the given context.
11
He, Zhen, Shaobing Gao, Liang Xiao, Daxue Liu, and Hangen He. "Multimedia Data Modelling Using Multidimensional Recurrent Neural Networks." Symmetry 10, no. 9 (2018): 370. http://dx.doi.org/10.3390/sym10090370.
Full textAbstract:
Modelling the multimedia data such as text, images, or videos usually involves the analysis, prediction, or reconstruction of them. The recurrent neural network (RNN) is a powerful machine learning approach to modelling these data in a recursive way. As a variant, the long short-term memory (LSTM) extends the RNN with the ability to remember information for longer. Whilst one can increase the capacity of LSTM by widening or adding layers, additional parameters and runtime are usually required, which could make learning harder. We therefore propose a Tensor LSTM where the hidden states are tensorised as multidimensional arrays (tensors) and updated through a cross-layer convolution. As parameters are spatially shared within the tensor, we can efficiently widen the model without extra parameters by increasing the tensorised size; as deep computations of each time step are absorbed by temporal computations of the time series, we can implicitly deepen the model with little extra runtime by delaying the output. We show by experiments that our model is well-suited for various multimedia data modelling tasks, including text generation, text calculation, image classification, and video prediction.
12
Bilinska, Оksana, Khrystyna Kulchytska, and Yuriy Surovyi. "TRAINING A NEURAL NETWORK FOR IMAGE STYLING." Vìsnik Nacìonalʹnogo unìversitetu "Lʹvìvsʹka polìtehnìka". Serìâ Arhìtektura 2024, no. 1 (2024): 16–23. http://dx.doi.org/10.23939/sa2024.01.016.
Full textAbstract:
Improving the visualization of projects and portfolios of designers and architects can be achieved by enhancing the illustrativeness and stylization of images using artificial intelligence. The use of neural networks for content generation significantly speeds up the work of designers. Among all the neural networks for image generation, MidJourney shows the best results. After analyzing the licenses and subscription costs of services as well as the models they employ, the Stable Diffusion deep learning neural network was chosen. The Stable Diffusion neural network is open-source, unlike DALL-E and Midjourney, allowing for unlimited content generation. The stylized images were generated in Stable Diffusion using Dreambooth based on the Google Collab platform. The creation of a custom model was conducted in two stages. The first stage involved preparing images for training the Stable Diffusion neural network. The second stage was the direct training of the neural network based on the Google Collab platform. Kobzar's graphic drawings served as the training dataset. Initially, 77 drawings with the same theme were selected for model training. 30 of these were used to train the model after corrections in Adobe Photoshop and Topaz Photo AI. Adjustments included cropping, background removal, printing raster, noise reduction, sharpening, and scaling images. The originality of the work lies in the fact that the trained model was used to create stylized creative images, utilizing excerpts from the poet's poems describing nature and events in a very realistic way. The generated images have successfully passed the Turing test, indicating a realistic reproduction of the style of Taras Shevchenko's drawings and the utilization of the author's poetic text as a prompt. The use of neural networks for generating and styling images as virtual assistants for designers and architects speeds up the creative process and enables the creation of works of any complexity.
13
Rabpreet, Singh Keer Rabpreet Singh Keer. "Handwriting generation using recurrent neural networks (LSTM)." International Journal of Scientific Development and Research 8, no. 9 (2023): 1085–109. https://doi.org/10.5281/zenodo.10446335.
Full textAbstract:
Handwriting is a skill developed by humans from a very early stage in order to represent his/her thoughts visually using letters and making meaningful words and sentences. Every person improves this skill by practicing and developing his/her own style of writing. Because of the distinctiveness of handwriting style, it is frequently used as a measure to identify a forgery. Even though the applications of synthesizing handwriting is less, this problem can be generalized and can be functionally applied to other more practical problems. Mimicking or imitating a specific handwriting style can have an extensive variety of applications like generating personalized handwritten documents, editing a handwritten document by using the similar handwriting style and also it is extended to compare handwriting styles to identify a forgery. All the training and test data is taken from IAM online handwriting database (IAMOnDB). IAM-OnDB consists of handwritten lines of data gathered from 223 various writers using an e-smart whiteboard.
14
Lee, Sujin, and Incheol Kim. "Multimodal Feature Learning for Video Captioning." Mathematical Problems in Engineering 2018 (2018): 1–8. http://dx.doi.org/10.1155/2018/3125879.
Full textAbstract:
Video captioning refers to the task of generating a natural language sentence that explains the content of the input video clips. This study proposes a deep neural network model for effective video captioning. Apart from visual features, the proposed model learns additionally semantic features that describe the video content effectively. In our model, visual features of the input video are extracted using convolutional neural networks such as C3D and ResNet, while semantic features are obtained using recurrent neural networks such as LSTM. In addition, our model includes an attention-based caption generation network to generate the correct natural language captions based on the multimodal video feature sequences. Various experiments, conducted with the two large benchmark datasets, Microsoft Video Description (MSVD) and Microsoft Research Video-to-Text (MSR-VTT), demonstrate the performance of the proposed model.
15
Pan, Zhong Liang, Ling Chen, and Guang Zhao Zhang. "Test Pattern Generation of VLSI Circuits Using Hopfield Neural Networks." Applied Mechanics and Materials 29-32 (August 2010): 1034–39. http://dx.doi.org/10.4028/www.scientific.net/amm.29-32.1034.
Full textAbstract:
A new test pattern generation method for the stuck-at faults in VLSI circuits is presented in this paper, the method uses Hopfield neural networks and chaotic simulated annealing. The Hopfield neural network corresponding to a digital circuit is built, the test patterns of faults in digital circuits are produced by computing the optima of the energy function. A chaotic simulated annealing algorithm is designed, which combines the features of chaotic systems and conventional simulated annealing, it is able to take the advantages of the stochastic properties and global search ability of chaotic system. The algorithm is used to compute the optima of the energy function of neural networks in order to produce the test patterns of faults. Experimental results show that the test pattern generation method proposed in this paper can produce the test patterns in short time for both single stuck-at faults and multiple stuck-at faults in digital circuits.
16
Guliyev, Shahriyar. "ARTIFICIAL TEXT GENERATION USING DEEP NEURAL NETWORKS: TRAINING OF SULEYMAN SANI AKHUNDOV’S PLAYS." SCIENTIFIC WORK 17, no. 6 (2023): 82–96. http://dx.doi.org/10.36719/2663-4619/91/82-96.
Full text
17
Sarwani, Mohammad Zoqi, Dian Ahkam Sani, and Fitria Chabsah Fakhrini. "Personality Classification through Social Media Using Probabilistic Neural Network Algorithms." International Journal of Artificial Intelligence & Robotics (IJAIR) 1, no. 1 (2019): 9. http://dx.doi.org/10.25139/ijair.v1i1.2025.
Full textAbstract:
Today the internet creates a new generation with modern culture that uses digital media. Social media is one of the popular digital media. Facebook is one of the social media that is quite liked by young people. They are accustomed to conveying their thoughts and expression through social media. Text mining analysis can be used to classify one's personality through social media with the probabilistic neural network algorithm. The text can be taken from the status that is on Facebook. In this study, there are three stages, namely text processing, weighting, and probabilistic neural networks for determining classification. Text processing consists of several processes, namely: tokenization, stopword, and steaming. The results of the text processing in the form of text are given a weight value to each word by using the Term Inverse Document Frequent (TF / IDF) method. In the final stage, the Probabilistic Neural Network Algorithm is used to classify personalities. This study uses 25 respondents, with 10 data as training data, and 15 data as testing data. The results of this study reached an accuracy of 60%.
18
Mahendiran, Ashwini. "Image Caption Generator Using Attention Based Neural Networks." International Journal for Research in Applied Science and Engineering Technology 11, no. 6 (2023): 1319–25. http://dx.doi.org/10.22214/ijraset.2023.53825.
Full textAbstract:
Abstract: Image caption generation is a method used to create sentences that describe the scene depicted in a given image. The process includes identifying objects within the image, carrying out various operations, and identifying the most important features of the image. Once the system has identified this information, it generates the most relevant and concise description of the image, which is both grammatically and semantically correct. With the progress in deep-learning techniques, algorithms are able to generate text in the form of natural sentences that can effectively describe an image. However, replicating the natural human ability to comprehend image content and produce descriptive text is a difficult task for machines. The uses of image captioning are vast and of great significance, as it involves creating succinct captions utilizing a variety of techniques such as Natural Language Processing (NLP), Computer Vision(CV), and Deep Learning (DL) techniques. The current study presents a system that employs an attention mechanism, in addition to an encoder and a decoder, to generate captions. It utilizes a pretrained CNN, Inception V3, to extract features from the image and a RNN, GRU, to produce a relevant caption. The attention mechanism used in this model is Bahdanau attention, and the Flickr-8Kdataset is utilized for training the model. The results demonstrate the model's capability to understand images and generate text in a reasonable manner.
19
Churi, Prof. Krutika H. "A Survey of Text Generation Models in NLP." International Journal of Advance and Applied Research 6, no. 25(A) (2025): 140–42. https://doi.org/10.5281/zenodo.15308095.
Full textAbstract:
<strong>Abstract:</strong> One of the main tasks of Natural Language Processing (NLP) is text creation, which is the process of turning an input into text that appears human. Rule-based systems have given way to data-driven deep learning models in text production over time. This survey offers a thorough analysis of text generation models, following their development, contrasting important approaches, and examining current issues and developments. The creation of transformer-based models, such as GPT, BERT, T5, and others, which have completely changed the field of text generating jobs, is given particular attention. Lastly, we go over the domain's unresolved issues and possible future study avenues.
20
Rubio-Manzano, Clemente, Alejandra Segura-Navarrete, Claudia Martinez-Araneda, and Christian Vidal-Castro. "Explainable Hopfield Neural Networks Using an Automatic Video-Generation System." Applied Sciences 11, no. 13 (2021): 5771. http://dx.doi.org/10.3390/app11135771.
Full textAbstract:
Hopfield Neural Networks (HNNs) are recurrent neural networks used to implement associative memory. They can be applied to pattern recognition, optimization, or image segmentation. However, sometimes it is not easy to provide the users with good explanations about the results obtained with them due to mainly the large number of changes in the state of neurons (and their weights) produced during a problem of machine learning. There are currently limited techniques to visualize, verbalize, or abstract HNNs. This paper outlines how we can construct automatic video-generation systems to explain its execution. This work constitutes a novel approach to obtain explainable artificial intelligence systems in general and HNNs in particular building on the theory of data-to-text systems and software visualization approaches. We present a complete methodology to build these kinds of systems. Software architecture is also designed, implemented, and tested. Technical details about the implementation are also detailed and explained. We apply our approach to creating a complete explainer video about the execution of HNNs on a small recognition problem. Finally, several aspects of the videos generated are evaluated (quality, content, motivation and design/presentation).
21
Iparraguirre-Villanueva, Orlando, Victor Guevara-Ponce, Daniel Ruiz-Alvarado, et al. "Text prediction recurrent neural networks using long short-term memory-dropout." Indonesian Journal of Electrical Engineering and Computer Science 29, no. 3 (2023): 1758. http://dx.doi.org/10.11591/ijeecs.v29.i3.pp1758-1768.
Full textAbstract:
<span lang="EN-US">Unit short-term memory (LSTM) is a type of recurrent neural network (RNN) whose sequence-based models are being used in text generation and/or prediction tasks, question answering, and classification systems due to their ability to learn long-term dependencies. The present research integrates the LSTM network and dropout technique to generate a text from a corpus as input, a model is developed to find the best way to extract the words from the context. For training the model, the poem "</span><em><span lang="EN-US">La Ciudad y los perros</span></em><span lang="EN-US">" which is composed of 128,600 words is used as input data. The poem was divided into two data sets, 38.88% for training and the remaining 61.12% for testing the model. The proposed model was tested in two variants: word importance and context. The results were evaluated in terms of the semantic proximity of the generated text to the given context.</span>
22
Guo, Zhijiang, Yan Zhang, Zhiyang Teng, and Wei Lu. "Densely Connected Graph Convolutional Networks for Graph-to-Sequence Learning." Transactions of the Association for Computational Linguistics 7 (November 2019): 297–312. http://dx.doi.org/10.1162/tacl_a_00269.
Full textAbstract:
We focus on graph-to-sequence learning, which can be framed as transducing graph structures to sequences for text generation. To capture structural information associated with graphs, we investigate the problem of encoding graphs using graph convolutional networks (GCNs). Unlike various existing approaches where shallow architectures were used for capturing local structural information only, we introduce a dense connection strategy, proposing a novel Densely Connected Graph Convolutional Network (DCGCN). Such a deep architecture is able to integrate both local and non-local features to learn a better structural representation of a graph. Our model outperforms the state-of-the-art neural models significantly on AMR-to-text generation and syntax-based neural machine translation.
23
Pogorilyy, S. D., A. A. Kramov, and P. V. Biletskyi. "METHOD FOR COHERECE EVALUATION OF UKRAINIAN TEXTS USING CONVO-LUTIONAL NEURAL NETWORK." Collection of scientific works of the Military Institute of Kyiv National Taras Shevchenko University, no. 65 (2019): 64–71. http://dx.doi.org/10.17721/2519-481x/2019/65-08.
Full textAbstract:
The estimation of text coherence is one of the most actual tasks of computer linguistics. Analysis of text coherence is widely used for writing and selection of documents. It allows clearly conveying the idea of an author to a reader. The importance of this task can be confirmed by the availability of actual works that are dedicated to solving it. Different automated methods for the estimation of text coherence are based on the methodology of machine learning. Corresponding methods are based on of formal text representation and following detection of regularities for the generation of an output result. The purpose of this work is to perform the analytic review of different automated methods for the estimation of text coherence; to justify method selection and adapt it due to the features of the Ukrainian language; to perform the experimental verification of the effectiveness of the suggested method for a Ukrainian corpus. In this paper, the comparative analysis of the methods for the estimation of coherence of English texts basing on a machine learning methodology has been performed. The expediency of application of methods that are based on trained universal models for the formalized representation of text components has been justified. The following models using neural networks with different architecture can be considered: recurrent and convolutional networks. These types of networks are widely used for text processing because they allow processing input data with an unfixed structure like sentences or words. Despite the ability of recurrent neural networks to take into account previous data (this behavior is similar to text perception by the reader), the convolutional neural network for conducting experimental research has been chosen. Such choice has been made due to the ability of convolutional neural networks to detect relations between entities regardless of the distance between them. In this paper, the principle of the method basing on the convolutional neural network and the corresponding architecture has been described. Program application for the verification of the suggested method effectiveness has been created. Formalized representation of text elements has been performed using a previously trained model for the semantic representation of words; the training process of this model has been implemented on the corpus of Ukrainian scientific abstracts. The training of the formed networks using pre-trained model has been performed. Experimental verification of method effectiveness for solving of document discrimination task and insert task has been made on the set of scientific articles. The results obtained may indicate that the method using convolutional neural networks can be used for further estimation of coherence of Ukrainian texts.
24
Rahbar Alam, Md, Prakash Anand, Rishika Barve, Abhrajit Chakraborty, and Madhu D Naik. "Multilingual Text to Video Generation of Press Information Bureau Press Release." INTERANTIONAL JOURNAL OF SCIENTIFIC RESEARCH IN ENGINEERING AND MANAGEMENT 09, no. 03 (2025): 1–9. https://doi.org/10.55041/ijsrem42380.
Full textAbstract:
This paper introduces a novel AI-powered framework that turns the Press Information Bureau (PIB) press releases into interactive, multilingual video content. The proposed system automates the summarization of text from press releases into 13 regional languages, along with creating video from that summary by using Generative Adversarial Networks and Large language model. This makes it accessible, culturally relevant, and allows better communication and interaction. Preliminary results show higher outreach and interaction metrics, thereby making the solution presented here quite promising for an inclusive public communication approach. Keywords— Artificial Intelligence, Generative Adversarial Network, Press Information Bureau (PIB), Large Language Model, Convolutional Neural Network, Summarization.
25
Komarskiy, O. C., and А. Yu Doroshenko. "Recurrent neural network model for music generation." PROBLEMS IN PROGRAMMING, no. 1 (March 2022): 087–93. http://dx.doi.org/10.15407/pp2022.01.087.
Full textAbstract:
The paper considers the possibility of generating musical compositions using recurrent neural networks. Two approaches to the generation of musical works are proposed and considered, namely using the method of notes and the method of chords. The research of both methods was carried out, and their advantages and disadvantages were formulated. As a result it was decided to use the method of notes as the main one for music generation. The process of searching and processing data for learning a music neural network is described in detail, the algorithm for converting data from MIDI format to your own text for use in a neural network is considered in detail. The learning process of the neural network was also described, and the learning speed was compared using GPUs and CPUs, as a result of which it was determined that learning takes place faster using a graphics processor, in some cases 5.5 times. As a result of testing the operation of the neural network, it was determined that the optimal characteristics of the recurrent neural network for music generation is a network consisting of 4 LSTM layers, each with a dimension of 600 neurons. As music generation cannot be assessed by objective characteristics, a special focus group survey was conducted to assess quality. It shows that music generated by a neural network received almost the same marks as music. written by a man. It should be considered as a great result. It was also determined that it was difficult for the survey participants to correctly identify the author of a musical work, since they correctly identified the authors in only 58% of cases. The proposed solution allows to easily generate musical compositions without human intervention.
26
Machueva, D. A., D. R. Baraev, and T. M.-A. Bechurkaev. "VARIOUS ASPECTS OF THE WORK OF GENERATIVE INTELLIGENCE IN THE FINE ARTS USING THE DALL-E NEURAL NETWORK AS AN EXAMPE." Vestnik komp'iuternykh i informatsionnykh tekhnologii, no. 243 (September 2024): 20–25. https://doi.org/10.14489/vkit.2024.09.pp.020-025.
Full textAbstract:
Neural creativity, or Artificial Intelligence Art, is becoming increasingly widespread, competing with traditional works of art created by hand. The study is aimed at analyzing the operating principles and capabilities of neural networks that generate images, as well as assessing their advantages and limitations in various fields of application using the DALL-E neural network as an example. The analysis of domestic and foreign literature in the field of artificial intelligence (AI) technologies and law, structuring and systematization of the received information, is included. The principles of operation of the DALL-E neural network, its training process and the architecture of GANs (Generative Adversarial Networks) are considered. The generation stages from processing the query text to image generation are shown, and the concept of a diffuse model is described. Generative neural networks can serve as an independent tool for obtaining the final result. However, most often the technical capabilities of AI are used in combination with human creative abilities and ideas, complementing, expanding and enriching them, offering new creative ideas, significantly speeding up work and increasing its efficiency. It is important to ensure responsible use of technology and develop appropriate regulatory standards to minimize possible risks. The current issues of authorship of generated works are considered. Existing copyright legislation, both Russian and foreign, allows for some legal uncertainty in the area of establishing and protecting rights to works of neural networks. Based on the analysis of the principle of interaction with neural networks, possible solutions to problems are given.
27
Voronov, R., and O. Donets. "AUTOMATIC CONTROL OF THE TECHNOLOGICAL PROCESS USING NEURAL NETWORKS TO DETERMINE THE PARAMETERS OF THE PRODUCTION PROCESS." Municipal economy of cities 3, no. 170 (2022): 7–11. http://dx.doi.org/10.33042/2522-1809-2022-3-170-7-11.
Full textAbstract:
In multifactorial systems using textual and graphical information in matrix factorization to facilitate the problem of separate data processing. Recently, in some studies, the study of neural networks to understand the content of text and graphic elements more deeply and to achieve efficacy by creating more accurate patterns of recognition of elements. However, the open question remains about how to effectively use graphic data from the thermal imager in matrix factorization. In this paper, we proposed a double-regularized matrix factorization with deep neural networks (DRMF) to solve this problem. DRMF applies a multilayered neural network model by stacking a convolutional neural network and a secured repetitive neural network to create independent distributed views of user content and objects. Then representations serve to regularize the generation of hidden models for both users and for elements of matrix factorization. So the proposed new model of the neural network works better than a model with a single convergent neural network. In this paper, we propose double - regularized matrix factorization with deep neural networks (DRMF) to solve this problem. DRMF uses a multi-layered neural network model by enclosing a convoluted neural network and a secure repeating neural network to create independent distributed representations of user content and objects. Then the representations are used to regularize the generation of hidden models for both users and elements of matrix factorization. Thus, the proposed new neural network model works better than the model with a single converging neural network. In traditional SF methods, only a feedback matrix is used, which contains explicit (eg, estimates) or implicit feedback to train and predict the life of the motor. As a rule, the feedback matrix is liquid, which means that most users encounter several elements. Based on this was presented in Proc. BigData Congress. However, this view has been significantly expanded using a new deep neural network model and adding new experimental attachments compared to the conference publication.
28
Stoll, Stephanie, Necati Cihan Camgoz, Simon Hadfield, and Richard Bowden. "Text2Sign: Towards Sign Language Production Using Neural Machine Translation and Generative Adversarial Networks." International Journal of Computer Vision 128, no. 4 (2020): 891–908. http://dx.doi.org/10.1007/s11263-019-01281-2.
Full textAbstract:
AbstractWe present a novel approach to automatic Sign Language Production using recent developments in Neural Machine Translation (NMT), Generative Adversarial Networks, and motion generation. Our system is capable of producing sign videos from spoken language sentences. Contrary to current approaches that are dependent on heavily annotated data, our approach requires minimal gloss and skeletal level annotations for training. We achieve this by breaking down the task into dedicated sub-processes. We first translate spoken language sentences into sign pose sequences by combining an NMT network with a Motion Graph. The resulting pose information is then used to condition a generative model that produces photo realistic sign language video sequences. This is the first approach to continuous sign video generation that does not use a classical graphical avatar. We evaluate the translation abilities of our approach on the PHOENIX14T Sign Language Translation dataset. We set a baseline for text-to-gloss translation, reporting a BLEU-4 score of 16.34/15.26 on dev/test sets. We further demonstrate the video generation capabilities of our approach for both multi-signer and high-definition settings qualitatively and quantitatively using broadcast quality assessment metrics.
29
Silva, Kanishka, Burcu Can, Raheem Sarwar, Frederic Blain, and Ruslan Mitkov. "Text Data Augmentation Using Generative Adversarial Networks – A Systematic Review." Journal of Computational and Applied Linguistics 1 (July 18, 2023): 6–38. https://doi.org/10.33919/jcal.23.1.1.
Full textAbstract:
Insufficient data is one of the main drawbacks in natural language processing tasks, and the most prevalent solution is to collect a decent amount of data that will be enough for the optimisation of the model. However, recent research directions are strategically moving towards increasing training examples due to the nature of the data-hungry neural models. Data augmentation is an emerging area that aims to ensure the diversity of data without attempting to collect new data exclusively to boost a model’s performance. Limitations in data augmentation, especially for textual data, are mainly due to the nature of language data, which is precisely discrete. Generative Adversarial Networks (GANs) were initially introduced for computer vision applications, aiming to generate highly realistic images by learning the image representations. Recent research has focused on using GANs for text generation and augmentation. This systematic review aims to present the theoretical background of GANs and their use for text augmentation alongside a systematic review of recent textual data augmentation applications such as sentiment analysis, low resource language generation, hate speech detection and fraud review analysis. Further, a notion of challenges in current research and future directions of GAN-based text augmentation are discussed in this paper to pave the way for researchers especially working on low-text resources.
30
Sushanth, P. "Automatic Text Summarization using Long Short-Term Memory (LSTM)." International Journal for Research in Applied Science and Engineering Technology 13, no. 5 (2025): 6237–42. https://doi.org/10.22214/ijraset.2025.71560.
Full textAbstract:
Text summarization is the process of automatically generating a shorter version of a given text while retaining its important information. Long Short-Term Memory (LSTM) is a type of recurrent neural network that is commonly used in natural language processing tasks such as text summarization. LSTM networks have a memory component that allows them to remember important information from the input text, which enables them to generate a more concise and relevant summary of the original text. LSTM networks can be trained on a large corpus of text data, and they can be fine-tuned for specific applications such as summarization. Overall, LSTM networks are a powerful tool for text summarization, as they can effectively capture the long-term dependencies in natural language data and produce high-quality summaries. Long Short-Term Memory (LSTM) is a type of recurrent neural network (RNN) that is able to effectively capture long-term dependencies in sequential data. LSTMs are composed of memory cells, input gates, forget gates, and output gates, which allow the network to selectively remember and forget information over time. This makes LSTMs well-suited for tasks such as language modeling and time series prediction. Despite their ability to handle complex sequential data, LSTMs are still subject to the vanishing gradient problem, which can limit their performance on longer sequences. However, recent advancements in LSTM architecture have helped to alleviate this issue.
31
Krishnaveni, G. "Automated Image Caption Generator Using Deep Learning." International Journal for Research in Applied Science and Engineering Technology 13, no. 5 (2025): 279–84. https://doi.org/10.22214/ijraset.2025.70145.
Full textAbstract:
One of the most important tasks in computer vision and natural language processing is the automatic creation of image captions.. This paper presents an approach to automatically generate descriptive captions for images by combining Convolutional Neural Networks (CNNs) and Inception V3 architecture. The proposed system utilizes a pre-trained Inception V3 model to extract high-level features from input images. These extracted features are then passed to a Recurrent Neural Network (RNN), specifically an LSTM (Long Short-Term Memory) network, to generate coherent and contextually relevant captions. Inception V3, a deep convolutional neural network designed for large-scale image classification, serves as the feature extractor. It helps capture rich spatial hierarchies within the images, making it highly effective for understanding complex visual information. The LSTM network, on the other hand, is used to model the sequence of words in the caption, ensuring grammatical correctness and semantic accuracy. The system is trained on a large dataset of images paired with humangenerated captions, such as the MS-COCO dataset, to ensure robust learning. The proposed method is evaluated based on its performance in generating captions that are semantically and syntactically appropriate. The model’s performance is compared to other existing image captioning methods, demonstrating its effectiveness in generating descriptive and accurate captions for unseen images. This work highlights the synergy between CNNs for visual feature extraction and LSTM networks for sequence generation, offering a promising solution for tasks requiring image-to-text conversion, including image retrieval, content-based indexing, and accessibility applications.
32
Abbas Ali Alkhazraji, Ali, Baheeja Khudair, and Asia Mahdi Naser Alzubaidi. "Ancient Textual Restoration Using Deep Neural Networks." BIO Web of Conferences 97 (2024): 00009. http://dx.doi.org/10.1051/bioconf/20249700009.
Full textAbstract:
Ancient text restoration represents a critical area in computer science because it reflects an imagination about human life in early eras. Deep leaning plays a crucial role in AI last few years, specifically Generative Adversarial Networks (GANs), to regenerate and acclimatize old manuscripts that have suffered from the time effects, degradation, or deterioration. This work used Codex Sinaiticus dataset that preprocessed by encoding the dataset after that number and special character have been removed, new line symbol has been removed, tokenization process has been used to separate each word as an instance. Class target has been generated by removing character making it as a target and replacing it with special character. Using produces Generative Adversarial Networks (GANs), which consist of generator and discriminator inside in one learning framework. The generator part responsible for generating the missing text while the discriminator evaluates the generated text. But using an iteratively procedure these networks together collaboratively to provide a very sensitive reconstruction operations with the same format of ancient manuscripts, inscriptions and documents. Three prediction models used as proposed techniques for retrieving missing ancient texts are LSTM, RNN, and GAN and the results was validation accuracy 86%,92% and 98% respectively.
33
Lu, Shida, Kai Huang, Talha Meraj, and Hafiz Tayyab Rauf. "A novel CAPTCHA solver framework using deep skipping Convolutional Neural Networks." PeerJ Computer Science 8 (April 6, 2022): e879. http://dx.doi.org/10.7717/peerj-cs.879.
Full textAbstract:
A Completely Automated Public Turing Test to tell Computers and Humans Apart (CAPTCHA) is used in web systems to secure authentication purposes; it may break using Optical Character Recognition (OCR) type methods. CAPTCHA breakers make web systems highly insecure. However, several techniques to break CAPTCHA suggest CAPTCHA designers about their designed CAPTCHA’s need improvement to prevent computer vision-based malicious attacks. This research primarily used deep learning methods to break state-of-the-art CAPTCHA codes; however, the validation scheme and conventional Convolutional Neural Network (CNN) design still need more confident validation and multi-aspect covering feature schemes. Several public datasets are available of text-based CAPTCHa, including Kaggle and other dataset repositories where self-generation of CAPTCHA datasets are available. The previous studies are dataset-specific only and cannot perform well on other CAPTCHA’s. Therefore, the proposed study uses two publicly available datasets of 4- and 5-character text-based CAPTCHA images to propose a CAPTCHA solver. Furthermore, the proposed study used a skip-connection-based CNN model to solve a CAPTCHA. The proposed research employed 5-folds on data that delivers 10 different CNN models on two datasets with promising results compared to the other studies.
34
Ivaschenko, Anton, Arkadiy Krivosheev, Anastasia Stolbova, and Oleg Golovnin. "Hybridization of Intelligent Solutions Architecture for Text Understanding and Text Generation." Applied Sciences 11, no. 11 (2021): 5179. http://dx.doi.org/10.3390/app11115179.
Full textAbstract:
This study proposes a new logical model for intelligent software architecture devoted to improving the efficiency of automated text understanding and text generation in industrial applications. The presented approach introduces a few patterns that provide a possibility to build adaptable and extensible solutions using machine learning technologies. The main idea is formalized by the concept of expounder hybridization. It summarizes an experience of document analysis and generation solutions development and social media analysis based on artificial neural networks’ practical use. The results of solving the task by the best expounder were improved using the method of aggregating multiple expounders. The quality of expounders’ combination can be further improved by introducing the pro-active competition between them on the basis of, e.g., auctioning algorithm, using several parameters including precision, solution performance and score. Analysis of the proposed approach was carried out using a dataset of legal documents including joint-stock company decision record sheets and protocols. The solution is implemented in an enterprise content management system and illustrated by an example of processing of legal documentation.
35
Henshaw, Christopher, Jacob Dennis, Jonathan Nadzam, and Alan J. Michaels. "Number Recognition Through Color Distortion Using Convolutional Neural Networks." Computers 14, no. 2 (2025): 34. https://doi.org/10.3390/computers14020034.
Full textAbstract:
Machine learning applied to image-based number recognition has made significant strides in recent years. Recent use of Large Language Models (LLMs) in natural language search and generation of text have improved performance for general images, yet performance limitations still exist for data subsets related to color blindness. In this paper, we replicated the training of six distinct neural networks (MNIST, LeNet5, VGG16, AlexNet, and two AlexNet modifications) using deep learning techniques with the MNIST dataset and the Ishihara-Like MNIST dataset. While many prior works have dealt with MNIST, the Ishihara adaption addresses red-green combinations of color blindness, allowing for further research in color distortion. Through this research, we applied pre-processing to accentuate the effects of red-green and monochrome colorblindness and hyper-parameterized the existing architectures, ultimately achieving better overall performance than currently published in known works.
36
.., Pallavi, and Sarika Chaudhary. "Maximizing Anomaly Detection Performance in Next-Generation Networks." Journal of Cybersecurity and Information Management 12, no. 2 (2023): 36–51. http://dx.doi.org/10.54216/jcim.120203.
Full textAbstract:
The paper discusses major components of the proposed intrusion detection system as well as associated ideas. Dimensionality reduction solutions are highly valued for their potential to improve the efficiency of anomaly detection. Furthermore, feature selection and fusion methods are applied to optimise the system's capabilities. The following summary of network control, management, and cloud-based network processing aspects highlights operations managers, cloud resources, network function virtualization (NFV), and hardware and software components. We discuss prospective Deep Autoencoders (DAEs) applications, such as their use in the dimensionality reduction module, training methodologies, and benefits. Data transformation utilising coded representations is also graphically displayed and described in the text using an encoder and decoder system. The role of the anomaly detection via virtual network function in the suggested technique is also investigated. This component leverages a deep neural network (DNN) to identify anomalies in the 5G network's peripherals. DNN design issues, optimisation methodologies, and the trade-off between model complexity and detection efficacy are also discussed. Overall, the passage provides an overview of the proposed intrusion detection scheme, its components, and the techniques employed, underscoring their contributions to improving efficiency, accuracy, and security in Next Generation Networks.
37
Ashwini Mandale-Jadhav. "Text Summarization Using Natural Language Processing." Journal of Electrical Systems 20, no. 11s (2025): 3410–17. https://doi.org/10.52783/jes.8095.
Full textAbstract:
Text summarization is a crucial task in natural language processing (NLP) that aims to condense large volumes of text into concise and informative summaries. This paper presents a comprehensive study of text summarization techniques using advanced NLP methods. The research focuses on extractive summarization, where key sentences or phrases are extracted from the original text to form a coherent summary. Various approaches such as graph-based algorithms, deep learning models, and hybrid methods combining linguistic features and neural networks are explored and evaluated. The paper also investigates the impact of domain-specific summarization techniques for specialized content areas. Experimental results on benchmark datasets demonstrate the effectiveness and scalability of the proposed methods compared to baseline summarization techniques. The findings contribute to advancing the state-of-the-art in text summarization, with implications for applications in information retrieval, document analysis, and automated content generation
38
Mahar, Shahid Ali, Muhammad Imran Mushtaque, Mashooque Ali Mahar, Javed Ahmed Mahar, and Aurangzeb Magsi. "Sindhi Text-Based Students Sentiment Analysis Using Convolutional Neural Network." VAWKUM Transactions on Computer Sciences 12, no. 2 (2024): 149–64. https://doi.org/10.21015/vtcs.v12i2.1943.
Full textAbstract:
Current generation especially the teenager students are using Social Media (SM) platforms at an extreme level even the sentimental angles are too discussed there. In the province Sindh, students mostly prefer to text the message in origin of their mother tongue i.e. Sindhi lexicon for sharing their views regarded politics, religions, sports, education etc.All these sentimental conveys are important for enhancing the academic capabilities.In this research paper, approach is broken down into multiple phases comprising of number of WhatsApp chat, lexicon generation, dataset tokenization, Convolutional Neural Network (CNN); all based on respective sentiments.To validate the experimentation process at standard level. 100 WhatsApp data chats were collected from different levels of students and divided into four categories.The CNN Model is used for sentimental classification. Accuracy, Precision, Recall and F-Score are the four parameters used for model evaluation. The model provides 0.874% accuracy, 0.883% recall, 0.863% precision and 0.745% F-Score.
39
Parab, Amogh, Ananya Malik, Arish Damania, Arnav Parekhji, and Pranit Bari. "Successive Image Generation from a Single Sentence." ITM Web of Conferences 40 (2021): 03017. http://dx.doi.org/10.1051/itmconf/20214003017.
Full textAbstract:
Through various examples in history such as the early man’s carving on caves, dependence on diagrammatic representations, the immense popularity of comic books we have seen that vision has a higher reach in communication than written words. In this paper, we analyse and propose a new task of transfer of information from text to image synthesis. Through this paper we aim to generate a story from a single sentence and convert our generated story into a sequence of images. We plan to use state of the art technology to implement this task. With the advent of Generative Adversarial Networks text to image synthesis have found a new awakening. We plan to take this task a step further, in order to automate the entire process. Our system generates a multi-lined story given a single sentence using a deep neural network. This story is then fed into our networks of multiple stage GANs inorder to produce a photorealistic image sequence.
40
Sandeep, Biradar Veeresh, and Nase Gururaj. "Creating Synthetic Pictures from Text utilizing RNN and CNN." Advancement in Image Processing and Pattern Recognition 8, no. 1 (2024): 1–4. https://doi.org/10.5281/zenodo.13772404.
Full textAbstract:
<em>A fresh machine learning task is to synthesize textual descriptions into visual representations. Using a mix of Recurrent Neural Networks (RNNs) and Convolutional Neural Networks (CNNs), one such method creates artificial images from text. RNNs, particularly those with Long Short-Term Memory (LSTM) units, absorb and comprehend the sequential structure of textual input. These networks collect contextual information and provide descriptive embedding’s. These embedding’s are used by a convolutional neural network (CNN) picture generating model to produce coherent and detailed pictures based on the abstract text attributes. Utilizing CNNs' capacity to produce contextually correct and high-resolution visual output, this hybrid technique enables advancements in automated design, virtual reality, and content production. Synthetic visuals that visually correspond with the given verbal descriptions are the end result. </em>
41
Kurematsu, Yasuo, Takashi Murai, Takuji Maeda, and Shinzo Kitamura. "Learning Control System of Biped Locomotive Robot Using Neural Networks." Journal of Robotics and Mechatronics 5, no. 6 (1993): 542–47. http://dx.doi.org/10.20965/jrm.1993.p0542.
Full textAbstract:
The authors are studying the autonomous walking trajectory generation of a biped locomotive robot using the system consisting of an inverted pendulum equation and neural networks. This paper uses the trajectory generation system to simulate and to verify how the robot reacts to a change in its initial posture or the initial weight coefficient of a multi-layered neural network or an addition of disturbances during walking. The simulation test showed that the initial posture of the robot mainly determined a success in walking as well as a gait and that some disturbances did not prevent the robot from walking.
42
Zahoor, Saniya, and Shabir A.Sofi. "Automatic Podcast Generation." Journal of University of Shanghai for Science and Technology 23, no. 10 (2021): 22–28. http://dx.doi.org/10.51201/jusst/21/09700.
Full textAbstract:
A massive leap forward in the field of Human Computer Interaction in living memory has been achieved by the Google Duo system to sustain a natural sounding and coherent phone call with a human being without them being able to tell the difference. The computer system capitalized on recent developments in the field of Synthetic voice generation along with real time processing and response generation. The aim of this work is to replicate the success of that presentation as well as to build upon that body of work and generate useful content summaries which can be converted into high quality podcasts. In particular, our approach first comprises of extracting text data from web pages using various Natural Language Processing (NLP) tools as well as deep neural networks. After that it summarises text into byte sized chunks using extractive summarisation. Then, in the end it generates clear, high quality audio podcasts from the produced summaries using recently developed text to speech engines.
43
Zatsepina, Aleksandra, Liudmila Mikhaylovskaya, and Polina Shindina. "Neural network-assisted eco-architectural designs: Conceptualizing sustainable construction solutions." E3S Web of Conferences 614 (2025): 05002. https://doi.org/10.1051/e3sconf/202561405002.
Full textAbstract:
Nowadays, neural networks are increasingly being used to generate sketchy architectural solutions. In this article, two projects that won the architectural competition are used as examples to demonstrate how to use the MidJorney neural network. Examples are given for generating text from the source image, generating images based on text queries and making corrections, combining images, images, and text, and altering an image fragment. A roadmap has been drawn up—an algorithm for using the MidJorney neural network to work on an architectural concept.
44
Vlasov, Konstantin. "Neural Cryptographic Information Security System of Recurrent Convergent Neural Networks." Voprosy kiberbezopasnosti, no. 4(38) (2020): 44–55. http://dx.doi.org/10.21681/2311-3456-2020-04-44-55.
Full textAbstract:
Abstract. The purpose: to construct an algorithm for information transformation by recurrent convergent neural networks with a given set of local minima of the energy functional for its subsequent application in the field of information security. Method: system analysis of the existing neural network paradigms that can be used for classification of images. Neural cryptographic system synthesis is with analogy methods, recurrent convergent neural networks, noise- resistant encoding and block ciphers algorithms. The result: a promising neural cryptographic system is proposed that can be used to develop an algorithm for noise-resistant coding, symmetric or stream data encryption based on the generation of various variants of the distorted image representing the sequence of bits to mask the original message. An algorithm for block symmetric data encryption based on Hopfield-type neural networks has been created. Key information includes information on the selected (using radial basic functions) structural characteristics of the potential with a given set of energy minima, which determines the dynamics of the neural network as a potential dynamic system, whose attractors are symbols (several symbols) of the alphabet of the input text. The size of the key depends on the power of the alphabet of the original message and the form of representation of the energy functional. The presented neural cryptographic system can also be used in the authentication system.
45
Jahan, Ayesha, Sanobar Shadan, Yasmeen Fatima, and Naheed Sultana. "Image Orator - Image to Speech Using CNN, LSTM and GTTS." International Journal for Research in Applied Science and Engineering Technology 11, no. 6 (2023): 4473–81. http://dx.doi.org/10.22214/ijraset.2023.54470.
Full textAbstract:
Abstract: This report presents an image to audio system that utilizes a combination of Long Short-Term Memory (LSTM) and Convolutional Neural Networks (CNN) for image captioning and Google Text-to-Speech (GTTS) for generating audio output. The aim of the project is to create an accessible system that converts images into descriptive audio signals for visually impaired individuals. The proposed system has the potential to provide meaningful context and information about the image through descriptive audio output, making it easier for visually impaired individuals to engage with visual content. In conclusion, the proposed image to audio system, which combines LSTM and CNN for image captioning and GTTS for audio generation, is a promising approach to making visual content more accessible to individuals with visual impairments. Future work may involve exploring different neural network architectures, optimising the system for real-time performance, and incorporating additional audio features to enhance the overall user experience.
46
Yu, Hao, and Xinfu Li. "MLRNet: A Meta-Loss Reweighting Network for Biased Data on Text Classification." Applied Sciences 14, no. 1 (2023): 164. http://dx.doi.org/10.3390/app14010164.
Full textAbstract:
Artificially generated datasets often exhibit biases, leading conventional deep neural networks to overfit. Typically, a weighted function adjusts sample impact during model updates using weighted loss. Meta-neural networks, trained with meta-learning principles, generalize well across tasks, acquiring generalized weights. This enables the self-generation of tailored weighted functions for data biases. However, datasets may simultaneously exhibit imbalanced classes and corrupted labels, posing a challenge for current meta-models. To address this, this paper presents Meta-Loss Reweighting Network (MLRNet) with fusion attention features. MLRNet continually evolves sample loss values, integrating them with sample features from self-attention layers in a semantic space. This enhances discriminative power for biased samples. By employing minimal unbiased meta-data for guidance, mutual optimization between the classifier and the meta-model is conducted, endowing biased samples with more reasonable weights. Experiments on English and Chinese benchmark datasets including artificial and real-world biased data show MLRNet’s superior performance under biased data conditions.
47
Piven, Artem. "Analysis of Financial Reports in Companies Using Machine Learning." Financial Markets, Institutions and Risks 7, no. 4 (2023): 135–54. http://dx.doi.org/10.61093/fmir.7(4).135-154.2023.
Full textAbstract:
The article aims to develop new algorithms for the automated analysis of financial reports based on machine learning algorithms, which increase the efficiency and accuracy of converting financial information into a text form. In this context, special attention is paid to deep learning methods and neural networks that contribute to automating and analyzing financial reports and their further interpretation. The article examines the problems of generating text data from financial statements, describes the general characteristics of this process, and systematizes the technologies used to solve the task of developing text data and available methods of machine learning. Specific technologies of text generation using neural networks were analyzed, and the potential and prospects of machine learning in the creation of text data based on the analysis of financial reports were investigated. The process of developing a module intended for automated analysis of financial statements is described in detail, a technical task is created, which is necessary to solve the given task, and the structure and functionality of the developed module in the automated system are described. The result is a developed module for automated analysis of financial reports. Given that the module is created using Python, it can be easily integrated into different systems or function as an independent system, for example, a website or an application for a personal computer. The results of the developed automated module are demonstrated in the example of the analysis of financial reports of the companies Microsoft, Alphabet, and Apple.
48
Gurunath, R., and Debabrata Samanta. "Insights into Artificial Neural Network techniques, and its Application in Steganography." Journal of Physics: Conference Series 2089, no. 1 (2021): 012043. http://dx.doi.org/10.1088/1742-6596/2089/1/012043.
Full textAbstract:
Abstract Deep Steganography is a data concealment technology that uses artificial intelligence (AI) to automate the process of hiding and extracting information through layers of training. It enables for the automated generation of a cover depending on the concealed message. Previously, the technique depended on the existing cover to hide data, which limited the number of Steganographic characteristics available. Artificial intelligence and deep learning techniques have been used to steganography recently and the results are satisfactory. Although neural networks have demonstrated their ability to imitate human talents, it is still too early to draw comparisons between people and them. To improve their capabilities, neural networks are being employed in a number of disciplines, including steganography. Recurrent Neural Networks (RNN) is a widely used technology that automatically creates Stego-text regardless of payload volume. The features are extracted using a convolution neural network (CNN) based on the image. Perceptron, Multi-Layer Perceptron (MLP), Feed Forward Neural Network, Long Short Term Memory (LSTM) networks, and others are examples of this. In this research, we looked at all of the neural network approaches for Steganographic purposes in depth. This article also discusses the problems that each technology faces, as well as potential solutions.
49
Chandrashekar, Dr A. M. "Generating Diverse Synthetic Morse Code Datasets for Neural-Network Based Classification." INTERNATIONAL JOURNAL OF SCIENTIFIC RESEARCH IN ENGINEERING AND MANAGEMENT 09, no. 06 (2025): 1–9. https://doi.org/10.55041/ijsrem49350.
Full textAbstract:
Abstract - This work presents a system for translating between Morse code and human-readable text, with applications in secure and efficient communication. A novel algorithm is proposed to generate one-dimensional synthetic datasets for classifying Morse code using supervised learning, especially neural networks. Despite minimal features, these datasets carry high information density and include added noise to test model robustness. An automated decoder converts Morse signals into audio, offering potential use in secure transmissions by intelligence agencies. The open- source algorithm and datasets encourage further research in classification and communication systems. Keywords—Morse code, neural networks, machine learning, secure communication, signal decoding, dataset generation.
50
Pears, Russel, and Ashwini Kumar Sharma. "Generating Human-Interpretable Rules from Convolutional Neural Networks." Information 16, no. 3 (2025): 230. https://doi.org/10.3390/info16030230.
Full textAbstract:
Advancements in the field of artificial intelligence have been rapid in recent years and have revolutionized various industries. Various deep neural network architectures capable of handling both text and images, covering code generation from natural language as well as producing machine translation and text summaries, have been proposed. For example, convolutional neural networks or CNNs perform image classification at a level equivalent to that of humans on many image datasets. These state-of-the-art networks have reached unprecedented levels of success by using complex architectures with billions of parameters, numerous kernel configurations, weight initialization, and regularization methods. Unfortunately to reach this level of success, the models that CNNs use are essentially black box in nature, with little or no human-interpretable information on the decision-making process. This lack of transparency in decision making gave rise to concerns amongst some sectors of the user community such as healthcare, finance, justice, and defense, among others. This challenge motivated our research, where we successfully produced human-interpretable influential features from CNNs for image classification and captured the interactions between these features by producing a concise decision tree making that makes classification decisions. The proposed methodology makes use of a pretrained VGG-16 with fine-tuning to extract feature maps produced by learnt filters. On the CelebA image benchmark dataset, we successfully produced human-interpretable rules that captured the main facial landmarks responsible for segmenting men from women with 89.6% accuracy, while on the more challenging Cats vs. Dogs dataset, the decision tree achieved 87.6% accuracy.