In recent years, Deep Learning models have shown a great performance in complex optimization problems. They generally require large training datasets, which is a limitation in most practical cases. Transfer learning a...
详细信息
In recent years, Deep Learning models have shown a great performance in complex optimization problems. They generally require large training datasets, which is a limitation in most practical cases. Transfer learning allows importing the first layers of a pre-trained architecture and connecting them to fully-connected layers to adapt them to a new problem. Consequently, the configuration of the these layers becomes crucial for the performance of the model. Unfortunately, the optimization of these models is usually a computationally demanding task. One strategy to optimize Deep Learning models is the pruning scheme. Pruning methods are focused on reducing the complexity of the network, assuming an expected performance penalty of the model once pruned. However, the pruning could potentially be used to improve the performance, using an optimization algorithm to identify and eventually remove unnecessary connections among neurons. This work proposes EvoPruneDeepTL, an evolutionary pruning model for Transfer Learning based Deep Neural Networks which replaces the last fully-connected layers with sparse layers optimized by a genetic algorithm. Depending on its solution encoding strategy, our proposed model can either perform optimized pruning or feature selection over the densely connected part of the neural network. We carry out different experiments with several datasets to assess the benefits of our proposal. Results show the contribution of EvoPruneDeepTL and feature selection to the overall computational efficiency of the network as a result of the optimization process. In particular, the accuracy is improved, reducing at the same time the number of active neurons in the final layers. (c) 2022 Published by Elsevier Ltd.
Natural language processing is related to human-computer interaction, where several challenges involve natural language understanding. Word sense disambiguation problem consists in the computational assignment of a me...
详细信息
Natural language processing is related to human-computer interaction, where several challenges involve natural language understanding. Word sense disambiguation problem consists in the computational assignment of a meaning to a word according to a particular context in which it occurs. Many natural language processing applications, such as machine translation, information retrieval, and information extraction, require this task which occurs at the semantic level. evolutionary computation approaches can be effective to solve this problem since they have been successfully used for many real-world optimization problems. In this paper, we propose to solve the word sense disambiguation problem using genetic and memetic algorithms, and apply them to Modern Standard Arabic. We demonstrate the performance of several models of our algorithms by carrying out experiments on a large Arabic corpus, and comparing them against a naive Bayes classifier. Experimental results show that genetic algorithms can achieve more precise prediction than memetic algorithms and naive Bayes classifier, attaining 79%. (C) 2014 Elsevier Ltd. All rights reserved.
Population-based evolutionary algorithms are suitable for solving multi-objective optimization problems involving multiple conflicting objectives. This is because a set of well-distributed solutions can be obtained by...
详细信息
Population-based evolutionary algorithms are suitable for solving multi-objective optimization problems involving multiple conflicting objectives. This is because a set of well-distributed solutions can be obtained by a single run, which approximate the optimal tradeoff among the objectives. Over the past three decades, evolutionary multi-objective optimization has been intensively studied and used in various real-world applications. However, evolutionary multi-objective optimization faces various difficulties as the number of objectives increases. The simultaneous optimization of more than three objectives, which is called many-objective optimization, has attracted considerable research attention. This paper explains various difficulties in evolutionary many-objective optimization, reviews representative approaches, and discusses their effects and limitations. (c) 2023 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.
How to maintain a good balance between convergence and diversity is particularly important for the performance of the many-objective evolutionary algorithms. Especially, the many-objective optimization problem is a co...
详细信息
How to maintain a good balance between convergence and diversity is particularly important for the performance of the many-objective evolutionary algorithms. Especially, the many-objective optimization problem is a complicated Pareto front, the many-objective evolutionary algorithm can easily converge to a narrow of the Pareto front. An efficient environmental selection and normalization method are proposed to address this issue. The maximum angle selection method based on vector angle is used to enhance the diversity of the population. The maximum angle rule selects the solution as reference vector can work well on complicated Pareto front. A penalty-based adaptive vector distribution selection criterion is adopted to balance convergence and diversity of the solutions. As the evolution process progresses, the new normalization method dynamically adjusts the implementation of the normalization. The experimental results show that new algorithm obtains 30 best results out of 80 test problems compared with other five many-objective evolutionary algorithms. A large number of experiments show that the proposed algorithm has better performance, when dealing with numerous many-objective optimization problems with regular and irregular Pareto Fronts.
We present our evolutionary Boss Improvement (EBI) approach, which receives partially complete bosses as input and generates fully equipped bosses that are complete. Additionally, the evolutionary algorithm and the ne...
详细信息
We present our evolutionary Boss Improvement (EBI) approach, which receives partially complete bosses as input and generates fully equipped bosses that are complete. Additionally, the evolutionary algorithm and the new genetic operations included in EBI favor genetic improvement, which affects the initial partial content of the incomplete bosses originally provided. We evaluate our approach using Kromaia, a commercial video game released on PlayStation 4 and PC. EBI uses an evolutionary algorithm to evolve a population of bosses guided by duels between the bosses being generated and a simulated player. Our approach evaluates the quality, in terms of game experience, of both the bosses generated and those included in Kromaia using six metrics (Completion, Duration, Uncertainty, Killer Moves, Permanence, and Lead Change) from the literature. The results show that the quality of the bosses created by EBI is comparable to the quality of the original bosses that were manually created by the developers of Kromaia. However, the EBI approach reduces the time required to build the bosses from five months (of elapsed time as opposed to dedicated time) to just 100 minutes of unattended run. EBI enables developers to accelerate the creation of content, such as bosses, which is essential to ensure player engagement.
In medical practice, all decisions, as for example the diagnosis based on the classification of images, must be made reliably and effectively. The possibility of having automatic tools helping doctors in performing th...
详细信息
In medical practice, all decisions, as for example the diagnosis based on the classification of images, must be made reliably and effectively. The possibility of having automatic tools helping doctors in performing these important decisions is highly welcome. Artificial Intelligence techniques, and in particular Deep Learning methods, have proven very effective on these tasks, with excellent performance in terms of classification accuracy. The problem with such methods is that they represent black boxes, so they do not provide users with an explanation of the reasons for their decisions. Confidence from medical experts in clinical decisions can increase if they receive from Artificial Intelligence tools interpretable output under the form of, e.g., explanations in natural language or visualized information. This way, the system outcome can be critically assessed by them, and they can evaluate the trustworthiness of the results. In this paper, we propose a new general-purpose method that relies on interpretability ideas. The approach is based on two successive steps, the former being a filtering scheme typically used in Content-Based Image Retrieval, whereas the latter is an evolutionary algorithm able to classify and, at the same time, automatically extract explicit knowledge under the form of a set of IF-THEN rules. This approach is tested on a set of chest X-ray images aiming at assessing the presence of COVID-19.
The paper concerns the use of evolutionary algorithms to solve the problem of multiobjective optimization and learning of fuzzy cognitive maps (FCMs) on the basis of multidimensional medical data related to diabetes. ...
详细信息
The paper concerns the use of evolutionary algorithms to solve the problem of multiobjective optimization and learning of fuzzy cognitive maps (FCMs) on the basis of multidimensional medical data related to diabetes. The analyzed approach consists of two stages. The first stage is to group multidimensional medical data using k-means clustering. The second stage is automatic construction of the FCM model for each group of data based on various criteria depending on the structure and forecasting capabilities. The simulation analysis was performed with the use of the developed multiobjective Individually Directional evolutionary Algorithm. Experiments show that the collection of fuzzy cognitive maps, in which each element is built on the basis of data for the particular group of patients, allows us to receive higher forecasting accuracy compared to the standard approaches.
A key issue in tackling multimodal multi-objective optimization problems (MMOPs) is achieving the balance between objective space diversity and decision space diversity to obtain multiple Pareto sets (PSs) while guara...
详细信息
A key issue in tackling multimodal multi-objective optimization problems (MMOPs) is achieving the balance between objective space diversity and decision space diversity to obtain multiple Pareto sets (PSs) while guaranteeing convergence to the Pareto front (PF). However, most of the existing methods for MMOPs face the following two shortages: (i) they put insufficient emphasis on improving decision space diversity, resulting in missing some PSs or PS segments;and (ii) they lack the utilization of promising historical individuals which may help search the PSs. To alleviate these limitations, this paper proposes a novel multi-stage evolutionary algorithm with two improved optimization strategies. Specifically, the proposed method decomposes solving MMOP into two tasks, i.e., the Exploration task and the Exploitation task. The Exploration task first aims to explore the decision space to detect the multiple PSs, then, the Exploitation task aims to enhance the diversities on both objective and decision spaces (i.e., exploiting the PF and PSs). To better search PSs, historical individuals that are well-distributed in the decision space are stored as the evolutionary experience, and then used to generate offspring individuals. Moreover, a new differential evolution is designed to force crowded individuals to move to sparse and undetected regions on the PSs to enhance the diversity of PSs. Extensive experimental studies compare the proposed method with five state-of-the-art methods tailored for MMOPs on two benchmark test suites. The results demonstrate that the proposed method can outperform others in terms of three performance indicators.
In the traffic light scheduling problem, the evaluation of candidate solutions requires the simulation of a process under various (traffic) scenarios. Thus, good solutions should not only achieve good objective functi...
详细信息
In the traffic light scheduling problem, the evaluation of candidate solutions requires the simulation of a process under various (traffic) scenarios. Thus, good solutions should not only achieve good objective function values, but they must be robust (low variance) across all different scenarios. Previous work has shown that combining IRACE with evolutionary operators is effective for this task due to the power of evolutionary operators in numerical optimization. In this article, we further explore the hybridization of evolutionary operators and the elitist iterated racing of IRACE for the simulation-optimization of traffic light programs. We review previous works from the literature to find the evolutionary operators performing the best when facing this problem to propose new hybrid algorithms. We evaluate our approach over a realistic case study derived from the traffic network of Malaga (Spain) with 275 traffic lights that should be scheduled optimally. The experimental analysis reveals that the hybrid algorithm comprising IRACE plus differential evolution offers statistically better results than the other algorithms when the budget of simulations is low. In contrast, IRACE performs better than the hybrids for a high simulations budget, although the optimization time is much longer.
Highly constrained multiobjective optimization problems (HCMOPs) refer to constrained multiobjective optimization problems (CMOPs) with complex constraints and small feasible regions, which are commonly encountered in...
详细信息
Highly constrained multiobjective optimization problems (HCMOPs) refer to constrained multiobjective optimization problems (CMOPs) with complex constraints and small feasible regions, which are commonly encountered in many real-world applications. Current constraint-handling techniques will face two difficulties when dealing with HCMOPs: 1) feasible solution is hard to be found and too much search effort is spent in locating the feasible region and 2) since the total feasible region of an HCMOP can consist of several disconnected subregions, the search process might be stuck in the comparatively larger feasible subregion, which does not contain the whole Pareto front (PF). To address these two issues, an evolutionary algorithm with constraint relaxation strategy based on differential evolution algorithm, that is, CRS-DE, is proposed in this article. In each generation, the CRS-DE relaxes the constraints by dividing the infeasible solutions into two subpopulations based on total constraint violation, that is, the "semifeasible" subpopulation (SF) and "infeasible" subpopulation (IF), respectively. The SF provides information on the promising regions of finding the feasible solution and is the driving force for convergence toward the PF, while the IF focuses on global exploration for new promising regions. Corresponding reproduction and selection strategies are devised for the SF, IF, and feasible subpopulations, which create a clear division of labor with cooperation to facilitate the search for feasible solutions. To leverage the influence of CRS and prevent the population from premature convergence, a mobility restriction mechanism is developed to restrict the individuals in the SF and IF from entering the feasible subpopulation and enhance the diversity of the whole population. Comprehensive experiments on a series of benchmark test problems and a real-world CMOP demonstrate the competitiveness of our method compared with other representative algorithms in terms of
暂无评论