Academic literature on the topic 'Forward-backward search algorithm'

In recent years, lightweight cryptography has been a hot field in symmetric cryptography. One of the most crucial problems is to find low-latency implementations of linear layers. The current main heuristic search methods include the Boyar-Peralta (BP) algorithm with depth limit and the backward search. In this paper we firstly propose two improved BP algorithms with depth limit mainly by minimizing the Euclidean norm of the new distance vector instead of maximizing it in the tie-breaking process of the BP algorithm. They can significantly increase the potential for finding better results. Furthermore, we give a new framework that combines forward search with backward search to expand the search space of implementations, where the forward search is one of the two improved BP algorithms. In the new framework, we make a minor adjustment of the priority of rules in the backward search process to enable the exploration of a significantly larger search space. As results, we find better results for the most of matrices studied in previous works. For example, we find an implementation of AES MixColumns of depth 3 with 99 XOR gates, which represents a substantial reduction of 3 XOR gates compared to the existing record of 102 XOR gates.

In this paper, we introduce a new line search technique, then employ it to construct a novel accelerated forward–backward algorithm for solving convex minimization problems of the form of the summation of two convex functions in which one of these functions is smooth in a real Hilbert space. We establish a weak convergence to a solution of the proposed algorithm without the Lipschitz assumption on the gradient of the objective function. Furthermore, we analyze its performance by applying the proposed algorithm to solving classification problems on various data sets and compare with other line search algorithms. Based on the experiments, the proposed algorithm performs better than other line search algorithms.

Multi-agent forward search (MAFS) is a state-of-the-art privacy-preserving planning algorithm. We describe a new variant of MAFS, called multi-agent forward-backward search (MAFBS) that uses both forward and backward messages to reduce the number of messages sent and obtain new privacy properties. While MAFS requires agents to send a state s produced by an action a to all agents that can apply any action in s, MAFBS sends such messages forward only to agents that have an action that requires one of the effects of a. To achieve completeness, it sends messages backward to agents that can supply a missing precondition. This more focused message passing scheme reduces states exchanged, and requires that agents be aware only of other agents that they directly interact with, leading to agent privacy.

Although the heuristic search algorithm A* is well-known to be optimally efficient, this result explicitly assumes forward search. Bidirectional search has long held promise for surpassing A*'s efficiency, and many varieties have been proposed, but it has proven difficult to achieve robust performance across multiple domains in practice. We introduce a simple bidirectional search technique called Incremental KKAdd that judiciously performs backward search to improve the accuracy of the forward heuristic function for any search algorithm. We integrate this technique with A*, assess its theoretical properties, and empirically evaluate its performance across seven benchmark domains. In the best case, it yields a factor of six reduction in node expansions and CPU time compared to A*, and in the worst case, its overhead is provably bounded by a user-supplied parameter, such as 1%. Viewing performance across all domains, it also surpasses previously proposed bidirectional search algorithms. These results indicate that Incremental KKAdd is a robust way to leverage bidirectional search in practice.

We propose a variance-based modified backward-forward algorithm with a stochastic approximation version of Armijo’s line search, which is robust with respect to an unknown Lipschitz constant, for solving a class of stochastic variational inequality problems. A salient feature of the proposed algorithm is to compute only one projection and two independent queries of a stochastic oracle at each iteration. We analyze the proposed algorithm for its asymptotic convergence, sublinear convergence rate in terms of the mean natural residual function, and optimal oracle complexity under moderate conditions. We also discuss the linear convergence rate with finite computational budget for the proposed algorithm without strong monotonicity. Preliminary numerical experiments indicate that the proposed algorithm is competitive with some existing algorithms. Furthermore, we consider an application in dealing with an equilibrium problem in stochastic natural gas trading market.

The special steering characteristics and task complexity of autonomous articulated vehicle (AAV) make it often require multiple forward and backward movements during autonomous driving. In this paper, we present a simple yet effective method, named head correction with fixed wheel position (HC-FWP), for the demand of multiple forward and backward movements. The goal-directed rapid-exploring random tree (GDRRT) algorithm is first used to search for a feasible path in the obstacle map, and then, the farthest node search (FNS) algorithm is applied to obtain a series of key nodes, on which HC-FWP is used to correct AAV heading angles. Simulation experiments with Dynapac CC6200 articulated road roller parameters show that the proposed improved goal-directed rapid-exploring random tree (IGDRRT), consisting of GDRRT, FNS, and HC-FWP, can search a feasible path on maps that require the AAV to move back and forth.

This work combines recent advances in AI planning under memory limitation, namely bitvector and symbolic search. Bitvector search assumes a bijective mapping between state and memory addresses, while symbolic search compactly represents state sets. The memory requirements vary with the structure of the problem to be solved. The integration of the two algorithms into one hybrid algorithm for strongly solving general games initiates a BDD-based solving algorithm, which consists of a forward computation of the reachable state set, possibly followed by a layered backward retrograde analysis. If the main memory becomes exhausted, it switches to explicit-state two-bit retrograde search. We use the classical game of Connect Four as a case study, and solve some instances of the problem space-efficiently with the proposed hybrid search algorithm.

For the past few decades, various algorithms have been proposed to solve convex minimization problems in the form of the sum of two lower semicontinuous and convex functions. The convergence of these algorithms was guaranteed under the L-Lipschitz condition on the gradient of the objective function. In recent years, an inertial technique has been widely used to accelerate the convergence behavior of an algorithm. In this work, we introduce a new forward–backward splitting algorithm using a new line search and inertial technique to solve convex minimization problems in the form of the sum of two lower semicontinuous and convex functions. A weak convergence of our proposed method is established without assuming the L-Lipschitz continuity of the gradient of the objective function. Moreover, a complexity theorem is also given. As applications, we employed our algorithm to solve data classification and image restoration by conducting some experiments on these problems. The performance of our algorithm was evaluated using various evaluation tools. Furthermore, we compared its performance with other algorithms. Based on the experiments, we found that the proposed algorithm performed better than other algorithms mentioned in the literature.

In this work, we aim to investigate the convex minimization problem of the sum of two objective functions. This optimization problem includes, in particular, image reconstruction and signal recovery. We then propose a new modified forward-backward splitting method without the assumption of the Lipschitz continuity of the gradient of functions by using the line search procedures. It is shown that the sequence generated by the proposed algorithm weakly converges to minimizers of the sum of two convex functions. We also provide some applications of the proposed method to compressed sensing in the frequency domain. The numerical reports show that our method has a better convergence behavior than other methods in terms of the number of iterations and CPU time. Moreover, the numerical results of the comparative analysis are also discussed to show the optimal choice of parameters in the line search.

This paper aims to explore the influence of uneven distribution of online education resources on the development of students in backward areas, and puts forward the strategy of optimizing resource allocation through algorithms to improve this situation. Firstly, the study analyzes the current situation of online education resource allocation, and points out the uneven distribution of resources caused by geography, economic conditions and students' needs. Subsequently, an improved algorithm based on multi-objective optimization and heuristic search (FERA algorithm) is designed and implemented, aiming at distributing online education resources to students in backward areas more fairly and effectively. The experiment selects representative schools and students in backward areas as samples, collects relevant data, and uses FERA algorithm to optimize resource allocation. The experimental results show that FERA algorithm significantly improves the fairness and efficiency of resource allocation, meets the diverse needs of students, and has a positive impact on students' academic performance and comprehensive quality. This paper calls on the government, schools and all walks of life to make joint efforts to popularize algorithm optimization technology, strengthen infrastructure construction and formulate differentiated policies in order to realize the fair distribution of educational resources.

Dans le contexte actuel des missions aériennes, les pilotes et opérateurs à bord d'une plateforme aéroportée sont confrontés à des situations tactiques de plus en plus complexes.Outre la trajectoire de l'appareil, ils doivent aussi utiliser au mieux plusieurs capteurs pour effectuer de multiples tâches essentielles à une bonne représentation de la situation tactique — allant de la surveillance et de la poursuite des cibles à l'identification des cibles, ainsi que la conduite de tir.Ils doivent également faire face à de nouveaux types de cibles plus difficiles à détecter, connectées en réseau et capables de se coordonner de plus en plus efficacement.Lors d'une mission à haute intensité, il peut être difficile d'exécuter toutes les tâches simultanément, car des prises de décision complexes en un très court laps de temps sont nécessaires.Les travaux présentés dans ce manuscrit sont motivés par un besoin de l'amélioration et l'évaluation de la planification pour la surveillance air-air avec un radar embarqué.Une part importante de cette thèse est consacrée à l'adaptation d'un modèle général pour la recherche de cible, ainsi que l'algorithme Forward And Backward (FAB) pour la tâche de surveillance air-air avec un radar. De nouveaux algorithmes sont également proposés.L'étude s'étend à l'intégration progressive des caractéristiques de la technologie radar au modèle de recherche de cible. Par exemple, l'effort de recherche est ici alloué à des cônes d'observation afin de représenter la forme d'un faisceau radar. Des cônes d'observations disjoints sont d'abord considérés, puis le modèle est complexifié en considérant des cônes d'observation recouvrants afin d'augmenter la qualité des plans calculés. Le modèle de détection est lui aussi graduellement amélioré afin de refléter le fonctionnement du radar.Un cadre robuste pour l'évaluation des stratégies de pointage radar est proposé. L'outil d'évaluation des algorithmes permet d'une part la comparaison des différents algorithmes sur différents critères clairement définis, et d'autre part de mesurer leur optimalité avec des bornes théoriques sur les performances atteignables.Grâce à ce cadre expérimental, nous validons la supériorité des algorithmes proposés par rapport à une heuristiques de la littérature ouverte utilisée dans l'industrie, et proposons ainsi un nouveau point de comparaison à l'état de l'art de la surveillance air-air.À travers des expérimentations, l'efficacité de ces algorithmes est validée, notamment en explorant les compromis entre la qualité de la solution et le temps de calcul afin de considérer des contraintes d'exécution en temps réel.En définitive, cette recherche représente une avancée dans l'optimisation de la surveillance air-air avec un radar embarqué. En effet les algorithmes proposés démontrent une performance supérieure face à l'heuristique existante, et un cadre d'évaluation robuste est introduit pour une comparaison méthodique. Ces contributions forment la base pour des études ultérieures dans des scénarios plus complexes, envisageant l'utilisation de multiples capteurs embarqués sur plusieurs plateformes, coordonnés pour exécuter simultanément des tâches diverses. Ce travail s'inscrit dans une démarche plus large visant à développer des outils qui permettront d'alléger la charge mentale des opérateurs, leur permettant ainsi de se concentrer sur les aspects opérationnels cruciaux de leurs missions<br>In the current context of aerial missions, pilots and operators aboard airborne platforms are facing increasingly complex tactical situations. Apart from managing the aircraft's trajectory, they must also use multiple sensors to complete various essential tasks for a comprehensive representation of the tactical situation. These tasks range from surveillance and tracking of targets to target identification and fire control. They must also deal with new, harder-to-detect targets that are networked and capable of more efficient coordination. During high-intensity missions, performing all these tasks simultaneously can be challenging due to the need for complex decision-making within very short timeframes.The work presented in this manuscript is driven by the need to enhance and evaluate air-to-air surveillance planning with an onboard radar. A significant portion of this thesis is dedicated to adapting a general target search model and the Forward And Backward (FAB) algorithm for the specific task of air-to-air surveillance using radar. New algorithms are also introduced. The study extends to the gradual integration of radar technology features into the target search model. For instance, research efforts are allocated to observation cones to represent radar beam shape. Initially, disjoint observation cones are considered, and the model is further enriched by incorporating overlapping observation cones to enhance the quality of computed plans. The detection model is also progressively refined to accurately reflect radar operation.A robust framework for evaluating radar pointing strategies is proposed. The algorithm evaluation tool allows for both the comparison of different algorithms based on clearly defined criteria and the measurement of their optimality against theoretical performance bounds. Through this experimental framework, we validate the superiority of the proposed algorithms over a heuristic from the open literature used in the industry, thus providing a new benchmark in the field of air-to-air surveillance. Through experimentation, the effectiveness of these algorithms is confirmed, particularly by exploring the trade-offs between solution quality and computation time to accommodate real-time execution constraints.Ultimately, this research represents a step forward in optimizing air-to-air surveillance with onboard radar. The proposed algorithms demonstrate superior performance compared to existing heuristics, and a robust evaluation framework is introduced for systematic comparison. These contributions serve as the basis for future studies in more complex scenarios, envisioning the use of multiple onboard sensors across various platforms, coordinated to perform diverse tasks simultaneously. This work aligns with a broader objective of developing tools to reduce the cognitive load on operators, allowing them to focus on the critical operational aspects of their missions

Artificial Intelligence (AI) has emerged as a defining force in the current era, shaping the contours of technology and deeply permeating our everyday lives. From autonomous vehicles to predictive analytics and personalized recommendations, AI continues to revolutionize various facets of human existence, progressively becoming the invisible hand guiding our decisions. Simultaneously, its growing influence necessitates the need for a nuanced understanding of AI, thereby providing the impetus for this book, “Introduction to Artificial Intelligence and Neural Networks.” This book aims to equip its readers with a comprehensive understanding of AI and its subsets, machine learning and deep learning, with a particular emphasis on neural networks. It is designed for novices venturing into the field, as well as experienced learners who desire to solidify their knowledge base or delve deeper into advanced topics. In Chapter 1, we provide a thorough introduction to the world of AI, exploring its definition, historical trajectory, and categories. We delve into the applications of AI, and underscore the ethical implications associated with its proliferation. Chapter 2 introduces machine learning, elucidating its types and basic algorithms. We examine the practical applications of machine learning and delve into challenges such as overfitting, underfitting, and model validation. Deep learning and neural networks, an integral part of AI, form the crux of Chapter 3. We provide a lucid introduction to deep learning, describe the structure of neural networks, and explore forward and backward propagation. This chapter also delves into the specifics of Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs). In Chapter 4, we outline the steps to train neural networks, including data preprocessing, cost functions, gradient descent, and various optimizers. We also delve into regularization techniques and methods for evaluating a neural network model. Chapter 5 focuses on specialized topics in neural networks such as autoencoders, Generative Adversarial Networks (GANs), Long Short-Term Memory Networks (LSTMs), and Neural Architecture Search (NAS). In Chapter 6, we illustrate the practical applications of neural networks, examining their role in computer vision, natural language processing, predictive analytics, autonomous vehicles, and the healthcare industry. Chapter 7 gazes into the future of AI and neural networks. It discusses the current challenges in these fields, emerging trends, and future ethical considerations. It also examines the potential impacts of AI and neural networks on society. Finally, Chapter 8 concludes the book with a recap of key learnings, implications for readers, and resources for further study. This book aims not only to provide a robust theoretical foundation but also to kindle a sense of curiosity and excitement about the endless possibilities AI and neural networks offer. The journ

AbstractEfficient methods for the simulation of quantum circuits on classic computers are crucial for their analysis due to the exponential growth of the problem size with the number of qubits. Here we study lumping methods based on bisimulation, an established class of techniques that has been proven successful for (classic) stochastic and deterministic systems such as Markov chains and ordinary differential equations. Forward constrained bisimulation yields a lower-dimensional model which exactly preserves quantum measurements projected on a linear subspace of interest. Backward constrained bisimulation gives a reduction that is valid on a subspace containing the circuit input, from which the circuit result can be fully recovered. We provide an algorithm to compute the constraint bisimulations yielding coarsest reductions in both cases, using a duality result relating the two notions. As applications, we provide theoretical bounds on the size of the reduced state space for well-known quantum algorithms for search, optimization, and factorization. Using a prototype implementation, we report significant reductions on a set of benchmarks. Furthermore, we show that constraint bisimulation complements state-of-the-art methods for the simulation of quantum circuits based on decision diagrams.

In the paper, we propose the distributed detection and identification multi-agent system (DDIMAS) framework that is the first attempt to apply in solving distributed denial of service (DDoS) problem. It includes three stages which are information heuristic rule, meta-heuristic algorithm and backward and forward search (BFS) rule, respectively. Moreover, the framework is a flexible architecture that can incorporate into other algorithms or rules to improve the overall performance. From the evaluation design, the experiment results show that our method is with higher detection rate and better accuracy than standard repositories. The proposed framework resolves issues in other swarm optimization algorithms and reveals that the performance of DDIMAS is better than existing methods and the adaptive meta-heuristic algorithm framework outperforms other methods for detecting DDoS attacks.

A challenging inverse problem is to identify the smooth function and the differential equation it represents from uncertain data. This paper extends the procedure previously developed for smooth data. The approach involves two steps. In the first step the data is smoothed using a recursive Bezier filter. For smooth data a single application of the filter is sufficient. The final set of data points provides a smooth estimate of the solution. More importantly, it will also identify smooth derivatives of the function away from the edges of the domain. In the second step the values of the function and its derivatives are used to establish a specific form of the differential equation from a particular class of the same. Since the function and its derivatives are known, the only unknowns are parameters describing the structure of the differential equations. These parameters are of two kinds: the exponents of the derivatives and the coefficients of the terms in the differential equations. These parameters can be determined by defining an optimization problem based on the residuals in a reduced domain. To avoid the trivial solution a discrete global search is used to identify these parameters. An example involving a third order constant coefficient linear differential equation is presented. A basic simulated annealing algorithm is used for the global search. Once the differential form is established, the unknown initial and boundary conditions can be obtained by backward and forward numerical integration from the reduced region.