Actor-critic (AC) algorithms are model-free deep reinforcement learning techniques that have consistently demonstrated effectiveness across various domains. Enhancing exploration (action entropy) and exploitation (expected return) through more efficient sample utilization is pivotal to their success. A key strategy for a learning algorithm is to intelligently navigate the environment's state space, prioritizing the exploration of rarely visited states over frequently encountered ones. However, conventional approaches rarely quantify a novel state's utility for policy learning, which can lead to inefficient exploration. To address this, we propose an innovative approach to bolster exploration by employing an intrinsic reward based on a state's novelty and the potential benefits of exploring that state, which we term plausible novelty. Our method seamlessly integrates with off-policy AC algorithms. By incentivizing the exploration of plausibly novel states, AC algorithms can achieve substantial improvements in sample efficiency and overall training performance. Empirical results demonstrate 19% improvement in training return and 30% reduction in standard deviation, averaged across comparisons of three benchmark algorithm pairs in five different environments.

Artificial Intelligence Generated Content (AIGC) has rapidly emerged with the capability to generate different forms of content, including text, images, videos, and other modalities, which can achieve a quality similar to content created by humans. As a result, AIGC is now widely applied across various domains such as digital marketing, education, and public health, and has shown promising results by enhancing content creation efficiency and improving information delivery. However, there are few studies that explore the latest progress and emerging challenges of AIGC across different domains. To bridge this gap, this paper brings together 16 scholars from multiple disciplines to provide a cross-domain perspective on the trends and challenges of AIGC. Specifically, the contributions of this paper are threefold: (1) It first provides a broader overview of AIGC, spanning the training techniques of Generative AI, detection methods, and both the spread and use of AI-generated content across digital platforms. (2) It then introduces the societal impacts of AIGC across diverse domains, along with a review of existing methods employed in these contexts. (3) Finally, it discusses the key technical challenges and presents research propositions to guide future work. Through these contributions, this vision paper seeks to offer readers a cross-domain perspective on AIGC, providing insights into its current research trends, ongoing challenges, and future directions.

With the recent rapid expansion of the smart grid infrastructure paving the way for greater integration of computer and network technologies within the power grid, it has become well suited for the application of machine learning techniques. However, machine learning requires vast amounts of data, which within the smart grid setting can reveal great amounts of personal details of the individuals using the grid. This work considers the application of a variant of distributed machine learning, federated learning, which enhances data privacy. We propose a Smart Grid Hierarchical Federated Learning (SGHFL) framework, which is tuned to common smart grid architectures in the real world. We demonstrate how our SGHFL framework improves client dropout and poisoning robustness, using relatively lightweight models suitable for devices with limited computational capability. We provide theoretical justification underlying our design and have evaluated our algorithms and framework with three datasets/environments of progressively increasing practicality. We have also compared our framework with relevant works.

With the increasing tension between conflicting requirements of the availability of large amounts of data for effective machine learning-based analysis, and for ensuring their privacy, the paradigm of federated learning has emerged, a distributed machine learning setting where the clients provide only the machine learning model updates to the server rather than the actual data for decision making. However, the distributed nature of federated learning raises specific challenges related to fairness in a heterogeneous setting. This motivates the focus of our article, on the heterogeneity of client devices having different computational capabilities and their impact on fairness in federated learning. Furthermore, our aim is to achieve fairness in heterogeneity while ensuring privacy. As far as we are aware there are no existing works that address all three aspects of fairness, device heterogeneity, and privacy simultaneously in federated learning. In this article, we propose a novel federated learning algorithm with personalization in the context of heterogeneous devices while maintaining compatibility with the gradient privacy preservation techniques of secure aggregation. We analyze the proposed federated learning algorithm under different environments with different datasets and show that it achieves performance close to or greater than the state-of-the-art in heterogeneous device personalized federated learning. We also provide theoretical proofs for the fairness and convergence properties of our proposed algorithm.

Soft actor-critic (SAC) is an off-policy actor-critic (AC) reinforcement learning (RL) algorithm, essentially based on entropy regularization. SAC trains a policy by maximizing the trade-off between expected return and entropy (randomness in the policy). It has achieved the state-of-the-art performance on a range of continuous control benchmark tasks, outperforming prior on-policy and off-policy methods. SAC works in an off-policy fashion where data are sampled uniformly from past experiences (stored in a buffer) using which the parameters of the policy and value function networks are updated. We propose certain crucial modifications for boosting the performance of SAC and making it more sample efficient. In our proposed improved SAC (ISAC), we first introduce a new prioritization scheme for selecting better samples from the experience replay (ER) buffer. Second we use a mixture of the prioritized off-policy data with the latest on-policy data for training the policy and value function networks. We compare our approach with the vanilla SAC and some recent variants of SAC and show that our approach outperforms the said algorithmic benchmarks. It is comparatively more stable and sample efficient when tested on a number of continuous control tasks in MuJoCo environments.

A continued fraction expansion represents a real number as an expression obtained by iteratively extracting the largest whole number from its fractional part and inverting the remainder.

The susceptibility of federated learning (FL) to attacks from untrustworthy endpoints has led to the design of several defense systems. FL defense systems enhance the federated optimization algorithm using anomaly detection, scaling the updates from endpoints depending on their anomalous behavior. However, the defense systems themselves may be exploited by the endpoints with more sophisticated attacks. First, this paper proposes three categories of attacks and shows that they can effectively deceive some well-known FL defense systems. In the first two categories, referred to as on-off attacks, the adversary toggles between being honest and engaging in attacks. We analyse two such on-off attacks, label flipping and free riding, and show their impact against existing FL defense systems. As a third category, we propose attacks based on “good mouthing” and “bad mouthing”, to boost or diminish influence of the victim endpoints on the global model. Secondly, we propose a new federated optimization algorithm, Viceroy, that can successfully mitigate all the proposed attacks. The proposed attacks and the mitigation strategy have been tested on a number of different experiments establishing their effectiveness in comparison with other contemporary methods. The proposed algorithm has also been made available as open source. Finally, in the appendices, we provide an induction proof for the on-off model poisoning attack, and the proof of convergence and adversarial tolerance for the new federated optimization algorithm.

In recent years, deep Convolutional Neural Networks (CNNs) have shown great potential in malware classification. CNNs, which are originally designed for image processing, identify malware binaries visualised as images. Despite offering promising performance, these human-designed networks are very large requiring more resources to train and deploy them. Evolutionary algorithms have been successfully used in designing deep neural networks automatically for different application domains. In this work, we use a Genetic Algorithm (GA) to optimise the CNN topology and hyperparameters for image-based malware classification. Computational experiments with two different malware datasets, Malimg and Microsoft Malware, show that the GA-evolved networks are very competitive to the networks designed by experts in classifying malware, yet they are also considerably smaller in size comparison.

Deep Neural Networks (DNN) require specifically tuned architectures and hyperparameters when being applied to any given task. Nature-inspired algorithms have been successfully applied for optimising various hyperparameters in different types of DNNs such as convolutional and recurrent for sentence classification. Hybrid networks, which contain multiple types of neural architectures have more recently been used for sentence classification in order to achieve better performance. However, the inclusion of hybrid architectures creates numerous possibilities of designing the network and those sub-networks also need fine-tuning. At present these hybrid networks are designed manually and various organisation attempts are noticed. In order to understand the benefit and the best design principle of such hybrid DNNs for sentence classification, in this work we used an Evolutionary Algorithm (EA) to optimise the topology and various hyperparameters in different types of layers within the network. In our experiments, the proposed EA designed the hybrid networks by using a single dataset and evaluated the evolved networks on multiple other datasets to validate their generalisation capability. We compared the EA-designed hybrid networks with human-designed hybrid networks in addition to other EA-optimised and expert-designed non-hybrid architectures.

The performance that Deep Neural Networks can achieve on a specific task is impacted significantly by the hyperparameters selected. In order to compare the performance of various Deep Neural Network architectures on sentence classification, an optimised set of hyperparameters needs to be found for each architecture. In this work we use a simple Genetic Algorithm to optimise the hyperparameters of three different architectures and we evaluate their performance on a suite of sentence classification benchmarks. We found that a single Genetic Algorithm is capable of optimising a variety of different architectures and that the evolved configurations found can compete with those chosen by experts while using fewer overall trainable parameters. The three architectures tested are a recurrent neural network and two types of convolutional networks with a large difference in complexity. Of the three architectures, optimised for sentence classification, a simple Convolutional Neural Network was the overall best performer consistently achieving good performance while using very few trainable parameters.

Robust optimization of real-world problems is essential to reduce the significant negative impact of uncertainties and noises present in the environment. Uncertainties in the decision variables are often handled using explicit or implicit averaging methods, in which the fitness of a solution isjudged based on the objective values of neighbouring solutions. Explicit averaging methods are highly reliable but require additional objective function evaluation, which can significantly increases the overall computational cost of an optimization process. On the other hand, implicit averaging techniques are computationally cheap, yet they suffer from low reliability since they use the history of search in a population-based optimization algorithm. This work proposes a conditional Pareto optimal dominance to improve the reliability of robust optimization methods that use implicit averaging methods. The proposed method is applied to Multi-Objective Particle Swarm optimisation. Empirical study with a benchmark suite shows the benefit of the proposed conditional Pareto optimal dominance in locating robust solutions in multi-objective problems.

Convolutional Neural Networks (CNNs) are currently the most prominent deep neural network models and have been used with great success for image classification and other applications. The performance of CNNs depends on their architecture and hyperparameter settings. Early CNN models like LeNet and AlexNet were manually designed by experienced researchers. The empirical design and optimisation of a new CNN architecture require a lot of expertise and can be very time-consuming. In this paper, we propose a genetic algorithm that can, for a given image processing task, efficiently explore a defined space of potentially suitable CNN architectures and simultaneously optimise their hyperparameters. We named this fast automatic optimisation model fast-CNN and employed it to find competitive CNN architectures for image classification on CIFAR10. In a series of comparative simulation experiments we could demonstrate that the network designed by fast-CNN achieved nearly as good accuracy as some of the other best network models available but fast-CNN took significantly less time to evolve. The trained fast-CNN network model also generalised well to CIFAR100.

The signalling pathways of the glioblastoma (GBM) microenvironment play critical roles in the origin and progress of the disease. Recently, it has been shown in in silico experiments that combination therapy targeting multiple key cytokines of the GBM microenvironment can significantly improve the therapeutic response. The study also revealed the inter-patient heterogeneity in response to the same combination therapy and emphasized the need for personalized treatment. This work proposes an evolutionary algorithm with a heuristic for creating combination therapies tailored to a patient's molecular profile. We investigated the effectiveness of the proposed approach using a sophisticated model of the GBM microenvironment and several virtual patients. We found that the proposed method was more successful, compared to the existing method, in designing a more effective combination therapy. Our results also underscored the importance of personalized treatment or patient stratification in designing combination therapy for GBM.

Abstract We propose a multivariate feature selection method that uses proximity graphs for assessing the quality of feature subsets. Initially, a complete graph is built, where nodes are the samples, and edge weights are calculated considering only the selected features. Next, a proximity graph is constructed on the basis of these weights and different fitness functions, calculated over the proximity graph, to evaluate the quality of the selected feature set. We propose an iterative methodology on the basis of a memetic algorithm for exploring the space of possible feature subsets aimed at maximizing a quality score. We designed multiple local search strategies, and we used an adaptive strategy for automatic balancing between the global and local search components of the memetic algorithm. The computational experiments were carried out using four well‐known data sets. We investigate the suitability of three different proximity graphs (minimum spanning tree, k‐nearest neighbors, and relative neighborhood graph) for the proposed approach. The selected features have been evaluated using a total of 49 classification methods from an open‐source data mining and machine learning package (WEKA). The computational results show that the proposed adaptive memetic algorithm can perform better than traditional genetic algorithms in finding more useful feature sets. Finally, we establish the competitiveness of our approach by comparing it with other well‐known feature selection methods.

The classification performance of a weighted voting ensemble of classifiers largely depends on the proper weight chosen for each base classifier's vote. In this paper, we propose the use of Differential Evolution algorithm for adjustment of voting-weights of base classifiers used in a heterogeneous ensemble of classifiers (HEoC). We used the average Matthews Correlation Coefficient (MCC), calculated over 10-fold cross-validation, as the quality measure of an ensemble. We applied the vanilla DE algorithm to maximise the average MCC score over the training dataset. The algorithm optimises the base classifiers' voting weights in order to attain better generalisation performance of the ensemble on testing datasets. Experiments were performed using 10 binary-class datasets taken from UCI-Machine Learning Repository. The results show consistent and superior generalisation performance of the constructed ensembles when compared with the base classifiers and other well-known ensemble of classifiers.

In this paper, we introduce our approach for evolving reaction networks. It is an efficient derivative of the neuroevolution of augmenting topologies algorithm directed at the evolution of biochemical systems or molecular programs. Our method addresses the problem of meaningful crossovers between two chemical reaction networks of different topologies. It also builds on features such as speciation to speed up the search, to the point where it can deal with complete, realistic mathematical models of the biochemical processes. We demonstrate this framework by evolving credible biochemical answers to challenging autonomous molecular problems: in vitro batch oscillatory networks that match specific oscillation shapes. Our experimental results suggest that the search space is efficiently covered and that, by using crossover and preserving topological innovations, significant improvements in performance can be obtained for the automatic design of molecular programs.

Evolutionary algorithms (EAs) have been successfully used in many studies for evolving both the structure and parameters of biological networks including gene regulatory networks that demonstrate different functionalities. However, most of these studies have used only mutation as the genetic operator in the evolutionary framework, perhaps due to the difficulty of implementing the crossover operation that generates the feasible network models. Nevertheless, crossover is considered to be the most powerful operator of EA which preserves the building blocks and promote quick convergence to a global optima. In this work we propose to use a Messy Genetic Algorithm (MGA) for evolving biological reaction networks that can calculate mathematical functions. The tactful encoding of MGA for reaction networks using a variable length chromosome, allows the use of crossover as well as mutation for the problem in hand that results in a fully functional EA. Earlier MGA has been used for solving many complex problems for which solution encoding is difficult. We used the proposed MGA for evolving different types of mathematical function calculating networks and the success was very encouraging. The evolved networks were able to calculate the target functions for mutually exclusive test data sets satisfactorily. Comparing with some other existing method based on Asexual Evolution (AE), the proposed method was superior in terms of different functions it could successfully evolve and the accuracy at which it could calculate those functions.

The unimodal Gaussian has been the distribution of choice for many extensions in Estimation of Distribution Algorithms (EDA). Some groups have used clustering algorithms, like k-means, to use multimodal distributions in different modifications of EDA. Most proposals use a fixed number of groups or clusters, and other works use heuristic approaches to find the right number of clusters in the search space without any previous information. The heuristic methods, however, lack the mathematical rigor required in the inference of a probability distribution's parameters. In this work, we propose the use of the Nonparametric Bayesian Model known as Dirichlet Process to fit the number of clusters given the data in a modified Population Based Incremental Learning (PBIL) model. We compare our approach with similar techniques that also use multimodal probability distributions to enhance the quality of the search in other EDA approaches. Our approach shows improvements by reducing the number of generations needed to find good results that are comparable to the state of the art in clustered EDA.

Proteins are composed by amino acids, which are created by genes. To understand how different genes interact to create different proteins, we need to model the gene regulatory networks (GRNs) of different organisms. There are different models that attempt to model GRNs. In this paper, we use the popular S-System to model small networks. This model has been solved with different evolutionary computation techniques, which have obtained good results; yet, there are no models that achieve a perfect reconstruction of the network. We implement a variation of particle swarm optimization (PSO), called dissipative PSO (DPSO), to optimize the model; we also research the use of an L1 regularizer and compare it with other evolutionary computing approaches. To the best of our knowledge, neither the DPSO nor L1 optimizer has been jointly used to solve the S-System. We find that the combination of S-System and DPSO offers advantages over previously used methods, and presents promising results for inferencing larger and more complex networks.

The performance of Differential Evolution (DE) algorithm is significantly affected by its parameter setting. But the choice of parameters is heavily dependent on the problem characteristics. Therefore, recently a couple of adaptation schemes that automatically adjust DE parameters have been proposed. The current work presents another adaptation scheme for DE parameters namely amplification factor and crossover rate. We systematically analyze the effectiveness of the proposed adaptation scheme for DE parameters using a standard benchmark suite consisting of ten functions. The undertaken empirical study shows that the proposed adaptive DE (aDE) algorithm exhibits an overall better performance compared to other prominent adaptive DE algorithms as well as canonical DE.

This paper proposes cellular differential evolution (cDE) algorithm for solving dynamic economic dispatch (DED) problems with valve-point effects. DEDs are high dimensional optimization problems with many equality and inequality constraints. The problem of premature convergence in solving high dimensional optimization problems using evolutionary algorithms (EAs) could be fought using population structuring. This work investigates the suitability a structured DE algorithm, called cDE, in solving these large dimensional optimization tasks. The suitability and effectiveness of the proposed algorithm is validated using two test systems consisting of 10 and 13 thermal units respectively. Numerical results clearly show that the proposed method outperforms existing methods in terms of solution quality and robustness.

Learning a ranking function is important for numerous tasks such as information retrieval (IR), question answering, and product recommendation. For example, in information retrieval, a Web search engine is required to rank and return a set of documents relevant to a query issued by a user. We propose RankDE, a ranking method that uses differential evolution (DE) to learn a ranking function to rank a list of documents retrieved by a Web search engine. To the best of our knowledge, the proposed method is the first DE-based approach to learn a ranking function for IR. We evaluate the proposed method using LETOR dataset, a standard benchmark dataset for training and evaluating ranking functions for IR. In our experiments, the proposed method significantly outperforms previously proposed rank learning methods that use evolutionary computation algorithms such as Particle Swam Optimization (PSO) and Genetic Programming (GP), achieving a statistically significant mean average precision (MAP) of 0.339 on TD2003 dataset and 0.430 on the TD2004 dataset. Moreover, the proposed method shows comparable results to the state-of-the-art non-evolutionary computational approaches on this benchmark dataset. We analyze the feature weights learnt by the proposed method to better understand the salient features for the task of learning to rank for information retrieval.

The performance of a memetic algorithm (MA) largely depends on the synergy between its global and local search counterparts. The amount of global exploration and local exploitation to be carried out, for optimal performance, varies with problem type. Therefore, an algorithm should intelligently allocate its computational efforts between genetic search and local search. In this work we propose an adaptive local search method that adjusts the effort for local tuning of individuals, taking feedback from the search. We implemented an MA hybridizing this adaptive local search method with differential evolution algorithm. Experimenting with a standard benchmark suite it was found that the proposed MA can utilize its global and local search components adaptively. The proposed algorithm also exhibited very competitive performance with other existing algorithms.

Recent study has introduced the powerful selection scheme in cellular genetic algorithm that can produce all ranges of selective pressure. The parameters used in that study, however, are empirically estimated by numbers of experiments. In this study, we propose the idea of performing a parameter estimation from a theoretical perspective. In the concept of maximizing the probability to find the new best solution together with hill-climbing optimization, enabling search for an optimal parameter in each generation. The selection scheme with the optimal parameter yields the numbers of mating that maximizes the probability of finding better solutions. This optimal parameter changes during run and it is adaptive to the behavior of a particular evolution. In order to confirm the capability of this parameter estimation method, we have conducted experiments to compare the manually tuned static parameter and the estimated dynamic parameter obtained from this method. Result from the experiment shows that the algorithm with estimated parameter performed better than the former method, even with the best tuned parameter. Therefore, by applying this parameter estimation to the selection scheme stated at the beginning, we would be able to create a new universal adaptive paradigm for the cellular evolutionary algorithm.

BACKGROUND: Most of the existing in silico phosphorylation site prediction systems use machine learning approach that requires preparing a good set of classification data in order to build the classification knowledge. Furthermore, phosphorylation is catalyzed by kinase enzymes and hence the kinase information of the phosphorylated sites has been used as major classification data in most of the existing systems. Since the number of kinase annotations in protein sequences is far less than that of the proteins being sequenced to date, the prediction systems that use the information found from the small clique of kinase annotated proteins can not be considered as completely perfect for predicting outside the clique. Hence the systems are certainly not generalized. In this paper, a novel generalized prediction system, PPRED (Phosphorylation PREDictor) is proposed that ignores the kinase information and only uses the evolutionary information of proteins for classifying phosphorylation sites. RESULTS: Experimental results based on cross validations and an independent benchmark reveal the significance of using the evolutionary information alone to classify phosphorylation sites from protein sequences. The prediction performance of the proposed system is better than those of the existing prediction systems that also do not incorporate kinase information. The system is also comparable to systems that incorporate kinase information in predicting such sites. CONCLUSIONS: The approach presented in this paper provides an efficient way to identify phosphorylation sites in a given protein primary sequence that would be a valuable information for the molecular biologists working on protein phosphorylation sites and for bioinformaticians developing generalized prediction systems for the post translational modifications like phosphorylation or glycosylation. PPRED is publicly available at the URL http://www.cse.univdhaka.edu/~ashis/ppred/index.php.

BACKGROUND: Gene regulatory network is an abstract mapping of gene regulations in living cells that can help to predict the system behavior of living organisms. Such prediction capability can potentially lead to the development of improved diagnostic tests and therapeutics. DNA microarrays, which measure the expression level of thousands of genes in parallel, constitute the numeric seed for the inference of gene regulatory networks. In this paper, we have proposed a new approach for inferring gene regulatory networks from time-series gene expression data using linear time-variant model. Here, Self-Adaptive Differential Evolution, a versatile and robust Evolutionary Algorithm, is used as the learning paradigm. RESULTS: To assess the potency of the proposed work, a well known nonlinear synthetic network has been used. The reconstruction method has inferred this synthetic network topology and the associated regulatory parameters with high accuracy from both the noise-free and noisy time-series data. For validation purposes, the proposed approach is also applied to the simulated expression dataset of cAMP oscillations in Dictyostelium discoideum and has proved it's strength in finding the correct regulations. The strength of this work has also been verified by analyzing the real expression dataset of SOS DNA repair system in Escherichia coli and it has succeeded in finding more correct and reasonable regulations as compared to various existing works. CONCLUSION: By the proposed approach, the gene interaction networks have been inferred in an efficient manner from both the synthetic, simulated cAMP oscillation expression data and real expression data. The computational time of this approach is also considerably smaller, which makes it to be more suitable for larger network reconstruction. Thus the proposed approach can serve as an initiate for the future researches regarding the associated area.

We propose a crossover-based adaptive local search (LS) operation for enhancing the performance of standard differential evolution (DE) algorithm. Incorporating LS heuristics is often very useful in designing an effective evolutionary algorithm for global optimization. However, determining a single LS length that can serve for a wide range of problems is a critical issue. We present a LS technique to solve this problem by adaptively adjusting the length of the search, using a hill-climbing heuristic. The emphasis of this paper is to demonstrate how this LS scheme can improve the performance of DE. Experimenting with a wide range of benchmark functions, we show that the proposed new version of DE, with the adaptive LS, performs better, or at least comparably, to classic DE algorithm. Performance comparisons with other LS heuristics and with some other well-known evolutionary algorithms from literature are also presented.

We present a memetic algorithm for evolving the structure of biomolecular interactions and inferring the effective kinetic parameters from the time series data of gene expression using the decoupled Ssystem formalism. We propose an Information Criteria based fitness evaluation for gene network model selection instead of the conventional Mean Squared Error (MSE) based fitness evaluation. A hill-climbing local-search method has been incorporated in our evolutionary algorithm for efficiently attaining the skeletal architecture which is most frequently observed in biological networks. The suitability of the method is tested in gene circuit reconstruction experiments, varying the network dimension and/or characteristics, the amount of gene expression data used for inference and the noise level present in expression profiles. The reconstruction method inferred the network topology and the regulatory parameters with high accuracy. Nevertheless, the performance is limited to the amount of expression data used and the noise level present in the data. The proposed fitness function has been found more suitable for identifying correct network topology and for estimating the accurate parameter values compared to the existing ones. Finally, we applied the methodology for analyzing the cell-cycle gene expression data of budding yeast and reconstructed the network of some key regulators.

Noise is inevitable in microarray data. The real challenge lies in identifying the biomolecular interactions in spite of the significant noise level present in the expression profiles that current technology offers. In this paper, we study the usefulness of an evolutionary approach in reverse engineering the biomolecular connections in a gene circuit from observed system dynamics that is contaminated with noise. The method uses an Information Criteria based fitness evaluation for selecting models, represented in decoupled S-system formalism, instead of the conventional mean squared error (MSE) based fitness evaluation. The suitability of the method is tested in experiments of reconstructing an artificial network from gene expression profiles with varying noise levels. The proposed fitness function has been found more suitable for identifying correct network topology and for estimating the accurate parameter values compared to the existing one.

In this paper we present an evolutionary approach for inferring the structure and dynamics in gene circuits from observed expression kinetics. For representing the regulatory interactions in a genetic network the decoupled S-system formalism has been used. We proposed an Information Criteria based fitness evaluation for model selection instead of the traditional Mean Squared Error (MSE) based fitness evaluation. A hill climbing local search method has been incorporated in our evolutionary algorithm for attaining the skeletal architecture which is most frequently observed in biological networks. Using small and medium-scale artificial networks we verified the implementation. The reconstruction method identified the correct network topology and predicted the kinetic parameters with high accuracy.

We present a modified version of Differential Evolution (DE) for locating the global minimum at a higher convergence velocity. The proposed model differs from conventional DE by applying selection both for reproduction and survival, whereas the original model applies exclusively "knock-out" selection mechanism for survival. Because of its one-to-one reproduction strategy DE often consumes too many fitness evaluations to locate the global optimum. In this work we show that selecting parents for breeding and offspring for survival, DE's search capability can be further accelerated, which will be particularly useful for expensive function optimizations. Computational results using many benchmark functions are reported which show significant improvements in the convergence characteristics of the proposed algorithm over the original one.

In this work we present an improved evolutionary method for inferring S-system model of genetic networks from the time series data of gene expression. We employed Differential Evolution (DE) for optimizing the network parameters to capture the dynamics in gene expression data. In a preliminary investigation we ascertain the suitability of DE for a multimodal and strongly non-linear problem like gene network estimation. An extension of the fitness function for attaining the sparse structure of biological networks has been proposed. For estimating the parameter values more accurately an enhancement of the optimization procedure has been also suggested. The effectiveness of the proposed method was justified performing experiments on a genetic network using different numbers of artificially created time series data.

In this paper, we proposed Fittest Individual Refinement (FIR), a crossover based local search method for Differential Evolution (DE). The FIR scheme accelerates DE by enhancing its search capability through exploration of the neighborhood of the best solution in successive generations. The proposed memetic version of DE (augmented by FIR) is expected to obtain an acceptable solution with a lower number of evaluations particularly for higher dimensional functions. Using two different implementations DEfirDE and DEfirSPX we showed that proposed FIR increases the convergence velocity of DE for well known benchmark functions as well as improves the robustness of DE against variation of population. Experiments using multimodal landscape generator showed our proposed algorithms consistently outperformed their parent algorithms. A performance comparison with reported results of well known real coded memetic algorithms is also presented.

A very effective means to study the gene networks is visualization. With rapid increase of the size of gene networks, it has become more realistic to identify the collaborating genes in the network, which will facilitate the behavioral study of the groups and the network as a whole. In our previous paper, we presented a layered approach for visualizing gene regulatory networks. In this paper, we present a 3D layout model for visualizing gene networks, which clusters the correlated genes depending on their causal relationships. To demonstrate the effectiveness of the approach, we visualize real gene networks of different sizes. The experimental results show the superiority and usefulness of the new model when compared with previous results.