Professor Ferrante Neri

Neural architecture search has attracted great attention in the research community and has been successfully applied in the industry recently. Differentiable architecture search (DARTS) is an efficient architecture search method. However, the networks searched by DARTS are often unstable due to the large gap in the architecture depth between the search phase and the verification phase. In addition, due to unfair exclusive competition between different candidate operations, DARTS is prone to skip connection aggregation, which may cause performance collapse. In this article, we propose progressive partial channel connections based on channel attention for differentiable architecture search (PA-DARTS) to solve the above problems. In the early stage of searching, we only select a few key channels for convolution using channel attention and reserve all candidate operations. As the search progresses, we gradually increase the number of channels and eliminate unpromising candidate operations to ensure that the search phase and verification phase are all carried out on 20 cells. Due to the existence of the partial channel connections based on channel attention, we can eliminate the unfair competition between operations and increase the stability of PA-DARTS. Experimental results showed that PA-DARTS could achieve 97.59% and 83.61% classification accuracy on CIFAR-10 and CIFAR-100, respectively. On ImageNet, our algorithm achieved 75.3% classification accuracy.

Multi-objective optimisation is a prominent subfield of optimisation with high relevance in real-world problems, such as engineering design. Over the past 2 decades, a multitude of heuristic algorithms for multi-objective optimisation have been introduced and some of them have become extremely popular. Some of the most promising and versatile algorithms have been implemented in software platforms. This article experimentally investigates the process of interpreting and implementing algorithms by examining multiple popular implementations of three well-known algorithms for multi-objective optimisation. We observed that official and broadly employed software platforms interpreted and thus implemented the same heuristic search algorithm differently. These different interpretations affect the algorithmic structure as well as the software implementation. Numerical results show that these differences cause statistically significant differences in performance.

Pattern Search is a family of gradient-free direct search methods for numerical optimisation problems. The characterising feature of pattern search methods is the use of multiple directions spanning the problem domain to sample new candidate solutions. These directions compose a matrix of potential search moves, that is the pattern. Although some fundamental studies theoretically indicate that various directions can be used, the selection of the search directions remains an unaddressed problem. The present article proposes a procedure for selecting the directions that guarantee high convergence/high performance of pattern search. The proposed procedure consists of a fitness landscape analysis to characterise the geometry of the problem by sampling points and selecting those whose objective function values are below a threshold. The eigenvectors of the covariance matrix of this distribution are then used as search directions for the pattern search. Numerical results show that the proposed method systematically outperforms its standard counterpart and is competitive with modern complex direct search and metaheuristic methods.

Mathematics, despite being the foundation of computer science, is nowadays often considered a totally separate subject. The fact that many jobs in computer science do not explicitly require any specific mathematical knowledge posed questions about the importance of mathematics within computer science undergraduate curricula. In many educational systems, a prior high school knowledge of mathematics is often not a mandatory requirement to be enrolled into a degree of computer science. On the other hand, several studies report that mathematics is important to computer scientists since it provides essential analytical and critical skills and since many professional and research tasks in computer science require an in-depth understanding of mathematical concepts. From this assumption, this article proposes an analysis of the cohort of computer science’ students, with a specific reference to British Universities, and identifies some challenges that lecturers of mathematical subjects normally face. On the basis of this analysis this article proposes two teaching techniques to promote effective learning. The proposed techniques aim at addressing the diversity of cohorts in terms of mathematical background and skepticism from part of the cohort of students to consider mathematics as an essential element of their education. Numerical results indicate the validity and effectiveness of the proposed teaching techniques.

Fitness landscape analysis for optimisation is a technique that involves analysing black-box optimisation problems to extract pieces of information about the problem, which can beneficially inform the design of the optimiser. Thus, the design of the algorithm aims to address the specific features detected during the analysis of the problem. Similarly, the designer aims to understand the behaviour of the algorithm, even though the problem is unknown and the optimisation is performed via a metaheuristic method. Thus, the algorithmic design made using fitness landscape analysis can be seen as an example of explainable AI in the optimisation domain. The present paper proposes a framework that performs fitness landscape analysis and designs a Pattern Search (PS) algorithm on the basis of the results of the analysis. The algorithm is implemented in a restarting fashion: at each restart, the fitness landscape analysis refines the analysis of the problem and updates the pattern matrix used by PS. A computationally efficient implementation is also presented in this study. Numerical results show that the proposed framework clearly outperforms standard PS and another PS implementation based on fitness landscape analysis. Furthermore, the two instances of the proposed framework considered in this study are competitive with popular algorithms present in the literature.

The wide deployment of cloud-assisted electronic health (eHealth) systems has already shown great benefits in managing electronic health records (EHRs) for both medical institutions and patients. However, it also causes critical security concerns. Since once a medical institution generates and outsources the patients’ EHRs to cloud servers, patients would not physically own their EHRs but the medical institution can access the EHRs as needed for diagnosing, it makes the EHRs integrity protection a formidable task, especially in the case that a medical malpractice occurs, where the medical institution may collude with the cloud server to tamper with the outsourced EHRs to hide the medical malpractice. Traditional cryptographic primitives for the purpose of data integrity protection cannot be directly adopted because they cannot ensure the security in the case of collusion between the cloud server and medical institution. In this paper, a secure cloud-assisted eHealth system is proposed to protect outsourced EHRs from illegal modification by using the blockchain technology (blockchain-based currencies, e.g., Ethereum). The key idea is that the EHRs only can be outsourced by authenticated participants and each operation on outsourcing EHRs is integrated into the public blockchain as a transaction. Since the blockchain-based currencies provide a tamper-proofing way to conduct transactions without a central authority, the EHRs cannot be modified after the corresponding transaction is recorded into the blockchain. Therefore, given outsourced EHRs, any participant can check their integrity by checking the corresponding transaction. Security analysis and performance evaluation demonstrate that the proposed system can provide a strong security guarantee with a high efficiency.

Deep convolutional neural networks (CNNs) are widely used for image classification. Deep CNNs often require a large memory and abundant computation resources, limiting their usability in embedded or mobile devices. To overcome this limitation, several pruning methods have been proposed. However, most of the existing methods focus on pruning parameters and cannot efficiently address the computation costs of deep CNNs. Additionally, these methods ignore the connections between the feature maps of different layers. This paper proposes a multi-objective pruning based on feature map selection (MOP-FMS). Unlike previous studies, we use the number of floating point operations (FLOPs) as a pruning objective in addition to the accuracy of the pruned network. First, we propose an encoding method based on feature map selection with a compact and efficient search space. Second, novel domain-specific crossover and mutation operators with reparation are designed to generate new individuals and make them meet the constraint rules. Then, decoding and pruning methods are proposed to prune networks based on the results of feature map selection. Finally, multi-objective optimisation is used for evaluation and individual selection. Our method has been tested with commonly used network structures. Numerical results demonstrate that the proposed method achieves better results than other state-of-the-art methods in terms of pruning rate without decreasing the accuracy rate to a high degree. •Considering the relation between the feature maps from different layers, the pruning problem is formulated as a bi-objective optimisation problem with feature map selection, and the accuracy rate and computation cost are simultaneously optimised.•A novel feature map-based encoding method and a unique decoding method are proposed for pruning common structures or networks with additive aggregation.•Special initialisation, crossover and mutation operators are designed with the quick reparation method to satisfy the encoding constraints of this specific problem.

Membrane computing models are parallel and distributed natural computing models. These models are often referred to as P systems. This paper proposes a novel multi-behaviors co-ordination controller model using enzymatic numerical P systems for autonomous mobile robots navigation in unknown environments. An environment classifier is constructed to identify different environment patterns in the maze-like environment and the multi-behavior co-ordination controller is constructed to coordinate the behaviors of the robots in different environments. Eleven sensory prototypes of local environments are presented to design the environment classifier, which needs to memorize only rough information, for solving the problems of poor obstacle clearance and sensor noise. A switching control strategy and multi-behaviors coordinator are developed without detailed environmental knowledge and heavy computation burden, for avoiding the local minimum traps or oscillation problems and adapt to the unknown environments. Also, a serial behaviors control law is constructed on the basis of Lyapunov stability theory aiming at the specialized environment, for realizing stable navigation and avoiding actuator saturation. Moreover, both environment classifier and multi-behavior coordination controller are amenable to the addition of new environment models or new behaviors due to the modularity of the hierarchical architecture of P systems. The simulation of wheeled mobile robots shows the effectiveness of this approach.

Several variants of spiking neural P systems (SNPS) have been presented in the literature to perform arithmetic operations. However, each of these variants was designed only for one specific arithmetic operation. In this paper, a complete arithmetic calculator implemented by SNPS is proposed. An application of the proposed calculator to information fusion is also proposed. The information fusion is implemented by integrating the following three elements: (1) an addition and subtraction SNPS already reported in the literature; (2) a modified multiplication and division SNPS; (3) a novel storage SNPS, i.e. a method based on SNPS is introduced to calculate basic probability assignment of an event. This is the first attempt to apply arithmetic operation SNPS to fuse multiple information. The effectiveness of the presented general arithmetic SNPS calculator is verified by means of several examples.

This article proposes a simplistic algorithmic framework, namely hyperSPAM, composed of three search algorithms for addressing continuous optimisation problems. The Covariance Matrix Adaptation Evolution Strategy (CMAES) is activated at the beginning of the optimisation process as a preprocessing component for a limited budget. Subsequently, the produced solution is fed to the other two single-solution search algorithms. The first performs moves along the axes while the second makes use of a matrix orthogonalization to perform diagonal moves. Four coordination strategies, in the fashion of hyperheuristics, have been used to coordinate the two single-solution algorithms. One of them is a simple randomized criterion while the other three are based on a success based reward mechanism. The four implementations of the hyperSPAM framework have been tested and compared against each other and modern metaheuristics on an extensive set of problems including theoretical functions and real-world engineering problems. Numerical results show that the different versions of the framework display broadly a similar performance. One of the reward schemes appears to be marginally better than the others. The simplistic random coordination also displays a very good performance. All the implementations of hyperSPAM significantly outperform the other algorithms used for comparison.

In recent years, incremental sampling-based motion planning algorithms have been widely used to solve robot motion planning problems in high-dimensional configuration spaces. In particular, the Rapidly-exploring Random Tree (RRT) algorithm and its asymptotically-optimal counterpart called RRT* are popular algorithms used in real-life applications due to its desirable properties. Such algorithms are inherently iterative, but certain modules such as the collision-checking procedure can be parallelized providing significant speedup with respect to sequential implementations. In this paper, the RRT and RRT* algorithms have been adapted to a bioinspired computational framework called Membrane Computing whose models of computation, a.k.a. P systems, run in a non-deterministic and massively parallel way. A large number of robotic applications are currently using a variant of P systems called Enzymatic Numerical P systems (ENPS) for reactive controlling, but there is a lack of solutions for motion planning in the framework. The novel models in this work have been designed using the ENPS framework. In order to test and validate the ENPS models for RRT and RRT*, we present two ad-hoc implementations able to emulate the computation of the models using OpenMP and CUDA. Finally, we show the speedup of our solutions with respect to sequential baseline implementations. The results show a speedup up to 6x using OpenMP with 8 cores against the sequential implementation and up to 24x using CUDA against the best multi-threading configuration.

The performance of most metaheuristic algorithms depends on parameters whose settings essentially serve as a key function in determining the quality of the solution and the efficiency of the search. A trend that has emerged recently is to make the algorithm parameters automatically adapt to different problems during optimization, thereby liberating the user from the tedious and time-consuming task of manual setting. These fine-tuning techniques continue to be the object of ongoing research. Differential evolution (DE) is a simple yet powerful population-based metaheuristic. It has demonstrated good convergence, and its principles are easy to understand. DE is very sensitive to its parameter settings and mutation strategy; thus, this study aims to investigate these settings with the diverse versions of adaptive DE algorithms. This study has two main objectives: (1) to present an extension for the original taxonomy of evolutionary algorithms (EAs) parameter settings that has been overlooked by prior research and therefore minimize any confusion that might arise from the former taxonomy and (2) to investigate the various algorithmic design schemes that have been used in the different variants of adaptive DE and convey them in a new classification style. In other words, this study describes in depth the structural analysis and working principle that underlie the promising and recent work in this field, to analyze their advantages and disadvantages and to gain future insights that can further improve these algorithms. Finally, the interpretation of the literature and the comparative analysis of the algorithmic schemes offer several guidelines for designing and implementing adaptive DE algorithms. The proposed design framework provides readers with the main steps required to integrate any proposed meta-algorithm into parameter and/or strategy adaptation schemes.

An artificial compound eye (ACE) is a bio-inspired vision sensor which mimics a natural compound eye (typical of insects). This artificial eye is able to visualize large fields of the outside world through multi-aperture. Due to its functioning, the ACE is subject to optical flow, that is an apparent motion of the object visualized by the eye. This paper proposes a method to estimate the optical flow based on capturing multiple images (multi-aperture). In this method, based on descriptors-based initial optical flows, a unified global energy function is presented to incorporate the information of multi-aperture and simultaneously recover the optical flows of multi-aperture. The energy function imposes a compound flow fields consistency assumption along with the brightness constancy and piecewise smoothness assumptions. This formula efficiently binds the flow field in time and space, and further enables view-consistent optical flow estimation. Experimental results on real and synthetic data demonstrate that the proposed method recovers view-consistent optical flows crossed multi-aperture and performs better than other optical flow methods on the multi-aperture images.

Epistasis is the correlation between the variables of a function and is a challenge often posed by real-world optimisation problems. Synthetic benchmark problems simulate a highly epistatic problem by performing a so-called problem's rotation. Mutation in Differential Evolution (DE) is inherently rotational invariant since it simultaneously perturbs all the variables. On the other hand, crossover, albeit fundamental for achieving a good performance, retains some of the variables, thus being inadequate to tackle highly epistatic problems. This article proposes an extensive study on rotational invariant crossovers in DE. We propose an analysis of the literature, a taxonomy of the proposed method and an experimental setup where each problem is addressed in both its non-rotated and rotated version. Our experimental study includes 280 problems over five different levels of dimensionality and nine algorithms. Numerical results show that 1) for a fixed quota of transferred design variables, the exponential crossover displays a better performance, on both rotated and non-rotated problems, in high dimensions while the binomial crossover seems to be preferable in low dimensions; 2) the rotational invariant mutation DE/current-to-rand is not competitive with standard DE implementations throughout the entire set of experiments we have presented; 3) DE crossovers that perform a change of coordinates to distribute the moves over the components of the offspring offer high-performance results on some problems. However, on average the standard DE/rand/1/exp appears to achieve the best performance on both rotated and non-rotated testbeds.

Echo state networks (ESNs), belonging to the family of recurrent neural networks (RNNs), are suitable for addressing complex nonlinear tasks due to their rich dynamic characteristics and easy implementation. The reservoir of the ESN is composed of a large number of sparsely connected neurons with randomly generated weight matrices. How to set the structural parameters of the ESN becomes a difficult problem in practical applications. Traditionally, the design of the parameters of the ESN structure is performed manually. The manual adjustment of the ESN parameters is not convenient since it is an extremely challenging and time-consuming task. This paper proposes an ensemble of five particle swarm optimization (PSO) strategies to design the structure of ESN and then reduce the manual intervention in the design process. An adaptive selection mechanism is used for each particle in the evolution to select a strategy from the strategy candidate pool for evolution. In addition, leaky integration neurons are used as reservoir internal neurons, which are added within the adaptive mechanism for optimization. The root mean squared error (RMSE) is adopted as the evaluation criterion. The experimental results on Mackey-Glass time series benchmark dataset show that the proposed method outperforms other traditional evolutionary methods. Furthermore, experimental results on electrocardiogram dataset show that the proposed method on the ensemble of PSO displays an excellent performance on real-world problems.

Real-world optimisation problems pose domain specific challenges that often require an ad-hoc algorithmic design to be efficiently addressed. The present paper investigates the optimisation of a key stage in data mining, known as instance reduction, which aims to shrink the input data prior to applying a learning algorithm. Performing a smart selection or creation of a reduced number of samples that represent the original data may become a complex large-scale optimisation problem, characterised by a computationally expensive objective function, which has been often tackled by sophisticated population-based metaheuristics that suffer from a high runtime. Instead, by following the Ockham’s Razor in Memetic Computing, we propose a Memetic Computing approach that we refer to as fast Single-Point Memetic Structure with Accelerated Local Search (SPMS-ALS). Using the k-nearest neighbours algorithm as base classifier, we first employ a simple local search for large-scale problems that exploits the search logic of Pattern Search, perturbing an -dimensional vector along the directions identified by its design variables one by one. This point-by-point perturbation mechanism allows us to design a strategy to re-use most of the calculations previously made to compute the objective function of a candidate solution. The proposed Accelerated Local Search is integrated within a single-point memetic framework and coupled with a resampling mechanism and a crossover. A thorough experimental analysis shows that SPMS-ALS, despite its simplicity, displays an excellent performance which is as good as that of the state-of-the-art while reducing up to approximately 85% of the runtime with respect to any other algorithm that performs the same number of function calls.

In classification tasks, feature selection (FS) can reduce the data dimensionality and may also improve classification accuracy, both of which are commonly treated as the two objectives in FS problems. Many meta-heuristic algorithms have been applied to solve the FS problems and they perform satisfactorily when the problem is relatively simple. However, once the dimensionality of the datasets grows, their performance drops dramatically. This paper proposes a self-adaptive multi-objective genetic algorithm (SaMOGA) for FS, which is designed to maintain a high performance even when the dimensionality of the datasets grows. The main concept of SaMOGA lies in the dynamic selection of five different crossover operators in different evolution process by applying a self-adaptive mechanism. Meanwhile, a search stagnation detection mechanism is also proposed to prevent premature convergence. In the experiments, we compare SaMOGA with five multi-objective FS algorithms on sixteen datasets. According to the experimental results, SaMOGA yields a set of well converged and well distributed solutions on most data sets, indicating that SaMOGA can guarantee classification performance while removing many features, and the advantage over its counterparts is more obvious when the dimensionality of datasets grows.

With the development of deep learning, the design of an appropriate network structure becomes fundamental. In recent years, the successful practice of Neural Architecture Search (NAS) has indicated that an automated design of the network structure can efficiently replace the design performed by human experts. Most NAS algorithms make the assumption that the overall structure of the network is linear and focus solely on accuracy to assess the performance of candidate networks. This paper introduces a novel NAS algorithm based on a multi-objective modeling of the network design problem to design accurate Convolutional Neural Networks (CNNs) with a small structure. The proposed algorithm makes use of a graph-based representation of the solutions which enables a high flexibility in the automatic design. Furthermore, the proposed algorithm includes novel ad-hoc crossover and mutation operators. We also propose a mechanism to accelerate the evaluation of the candidate solutions. Experimental results demonstrate that the proposed NAS approach can design accurate neural networks with limited size.

Feature selection (FS) is an important research topic in machine learning. Usually, FS is modelled as a bi-objective optimization problem whose objectives are: 1) classification accuracy; 2) number of features. One of the main issues in real-world applications is missing data. Databases with missing data are likely to be unreliable. Thus, FS performed on a data set missing some data is also unreliable. In order to directly control this issue plaguing the field, we propose in this study a novel modelling of FS: we include reliability as the third objective of the problem. In order to address the modified problem, we propose the application of the non-dominated sorting genetic algorithm-III (NSGA-III). We selected six incomplete data sets from the University of California Irvine (UCI) machine learning repository. We used the mean imputation method to deal with the missing data. In the experiments, k-nearest neighbors (K-NN) is used as the classifier to evaluate the feature subsets. Experimental results show that the proposed three-objective model coupled with NSGA-III efficiently addresses the FS problem for the six data sets included in this study.

Optimization Spiking Neural P System (OSNPS) is the first membrane computing model to directly derive an approximate solution of combinatorial problems with a specific reference to the 0/1 knapsack problem. OSNPS is composed of a family of parallel Spiking Neural P Systems (SNPS) that generate candidate solutions of the binary combinatorial problem and a Guider algorithm that adjusts the spiking probabilities of the neurons of the P systems. Although OSNPS is a pioneering structure in membrane computing optimization, its performance is competitive with that of modern and sophisticated metaheuristics for the knapsack problem only in low dimensional cases. In order to overcome the limitations of OSNPS, this paper proposes a novel Dynamic Guider algorithm which employs an adaptive learning and a diversity-based adaptation to control its moving operators. The resulting novel membrane computing model for optimization is here named Adaptive Optimization Spiking Neural P System (AOSNPS). Numerical result shows that the proposed approach is effective to solve the 0/1 knapsack problems and outperforms multiple various algorithms proposed in the literature to solve the same class of problems even for a large number of items (high dimensionality). Furthermore, case studies show that a AOSNPS is effective in fault sections estimation of power systems in different types of fault cases: including a single fault, multiple faults and multiple faults with incomplete and uncertain information in the IEEE 39 bus system and IEEE 118 bus system.

Dropout is an effective method of mitigating over-fitting while training deep neural networks (DNNs). This method consists of switching off (dropping) some of the neurons of the DNN and training it by keeping the remaining neurons active. This approach makes the DNN general and resilient to changes in its inputs. However, the probability of a neuron belonging to a layer to be dropped, the 'dropout rate', is a hard-to-tune parameter that affects the performance of the trained model. Moreover, there is no reason, besides being more practical during parameter tuning, why the dropout rate should be the same for all neurons across a layer. This paper proposes a novel method to guide the dropout rate based on an evolutionary algorithm. In contrast to previous studies, we associate a dropout with each individual neuron of the network, thus allowing more flexibility in the training phase. The vector encoding the dropouts for the entire network is interpreted as the candidate solution of a bi-objective optimisation problem, where the first objective is the error reduction due to a set of dropout rates for a given data batch, while the second objective is the distance of the used dropout rates from a prearranged constant. The second objective is used to control the dropout rates and prevent them from becoming too small, hence ineffective; or too large, thereby dropping a too-large portion of the network. Experimental results show that the proposed method, namely GADropout, produces DNNs that consistently outperform DNNs designed by other dropout methods, some of them being modern advanced dropout methods representing the state-of-the-art. GADroput has been tested on multiple datasets and network architectures.

In this article, various metaheuristics for a numerical optimization problem with application to Electric Impedance Tomography are tested and compared. The experimental setup is composed of a real valued Genetic Algorithm, the Differential Evolution, a self adaptive Differential Evolution recently proposed in literature, and two novel Memetic Algorithms designed for the problem understudy. The two proposed algorithms employ different algorithmic philosophies in the field of Memetic Computing. The first algorithm integrates a local search into the operations of the offspring generation, while the second algorithm applies a local search to individuals already generated in the spirit of life-time learning. Numerical results show that the fitness landscape and difficulty of the optimization problem heavily depends on the geometrical configuration, as well the proposed Memetic Algorithms seem to be more promising when the geometrical conditions make the problem harder to solve.

This paper proposes a metaheuristic approach to solve a complex large scale optimization problem that originates from a recently introduced Positron Emission Tomography (PET) data analysis method that provides an estimate of tissue heterogeneity. More specifically three modern metaheuristics have been tested. These metaheustics are based on Differential Evolution, Particle Swarm Optimization, and Memetic Computing. On the basis of a preliminary analysis of the fitness landscape, an intelligent initialization technique has been proposed in this paper. More specifically, since the fitness landscape appears to have a strong basin of attraction containing a multimodal landscape, a local search method is applied to one solution at the beginning of the optimization process and inserted into a randomly generated population. The resulting "doped" population is then processed by the metaheuristics. Numerical results show that the application of the local search at the beginning of the optimization process leads to significant benefits in terms of algorithmic performance. Among the metaheuristics analyzed in this study, the DE based algorithm appears to display the best performance.

This paper proposes the integration of the generalized opposition based learning into compact Differential Evolution frameworks and tests its impact on the algorithmic performance. Opposition-based learning is a technique which has been applied, in several circumstances, to enhance the performance of Differential Evolution. It consists of the generation of additional points by means of a hyper-rectangle. These opposition points are simply generated by making use of a central symmetry within the hyper-rectangle. In the population based Differential Evolution, the inclusion of this search move corrects a limitation of the original algorithm, i.e. the scarcity of search moves, and sometimes leads to benefits in terms of algorithmic performance. The opposition-based learning scheme is further improved in the generalized scheme by integrating some randomness and progressive narrowing of the search. The proposed study shows how the generalized opposition-based learning can be encoded within a compact Differential Evolution framework and displays its effect on a set of diverse problems. Numerical results show that the generalized opposition-based learning is beneficial for compact Differential Evolution employing the binomial crossover while its implementation is not always successful when the exponential crossover is used. In particular, the opposition-based logic appears to be in general promising for non-separable problems whilst it seems detrimental for separable problems.

This paper proposes a novel adaptation scheme for Differential Evolution (DE) frameworks. The proposed algorithm, namely Estimation Distribution Differential Evolution (EDDE), is based on a DE structure and employs randomized scale factor ad crossover rate values. These values are sampled from truncated Gaussian probability distribution functions. These probability functions adaptively vary during the optimization process. At the beginning of the optimization the truncated Gaussian functions are characterized by a large standard deviation values and thus are similar to uniform distributions. During the later stages of the evolution, the probability functions progressively adapt to the most promising values attempting to detect the optimal working conditions of the algorithm. The performance offered by the proposed algorithm has been compared with those given by three modern DE based algorithms which represent the state-of-the-art in DE. Numerical results show that the proposed EDDE, despite its simplicity, is competitive with the other algorithms and in many cases displays a very good performance in terms of both final solution detected and convergence speed.

This paper proposes an artificial player for the Turtle Race game, with the goal of creating an opponent that will provide some amount of challenge to a human player. Turtle Race is a game of imperfect information, where the players know which one of the game pieces is theirs, but do not know which ones belong to the other players and which ones are neutral. Moreover, movement of the pieces is determined by cards randomly drawn from a deck. The artificial player is based oil a non-linear neural network whose training is performed by means of a novel parallel evolutionary algorithm with fitness diversity adaptation. The algorithm handles, in parallel, several populations which cooperate with each other 133, exchanging individuals when a population registers a diversity loss. Four Popular evolutionary algorithms have been tested for the proposed parallel framework. Numerical results show that an evolution strategy can be very efficient for the problem under examination and that the proposed adaptation tends to improve upon the algorithmic performance Without any addition in computational overhead. The resulting artificial player displayed a high performance against other artificial players and a challenging behavior for expert human players.

This paper proposes the Noise Analysis compact Genetic Algorithm (NAcGA). This algorithm integrates a noise analysis component within a compact structure. This fact makes the proposed algorithm appealing for those real-world applications characterized by the necessity of a high performance optimizer despite severe hardware limitations. The noise analysis component adaptively assigns the amount of fitness evaluations to be performed in order to distinguish two candidate solutions. In this way, it is assured that computational resources are not wasted and the selection of the most promising solution is correctly performed. The noise analysis employed in this algorithm spouses very well the pair-wise comparison logic typical of compact evolutionary algorithms. Numerical results show that the proposed algorithm significantly improves upon the performance, in noisy environments, of the standard compact genetic algorithm. Two implementation variants based on the elitist strategy have been tested in this studies. It is shown that the nonpersistent strategy is more robust to the noise than the persistent one and therefore its implementation seems to be advisable in noisy environments.

This paper proposes the introduction of a generator of random individuals within the ring topology of a Parallel Differential Evolution algorithm. The generated random individuals are then injected within a sub-population. A crucial point in the proposed study is that a proper balance between migration and random injection can determine the success of a distributed Differential Evolution scheme. An experimental study of this balance is carried out in this paper. Numerical results show that the proposed Parallel Random Injection Differential Evolution seems to be a simple, robust, and efficient algorithm which can be used for various applications. An important finding of this paper is that premature convergence problems due to an excessively frequent migration can be overcome by the injection of random individuals. In this way, the resulting algorithm maintains the high convergence speed properties of a parallel algorithm with high migration but differs in that it is able to continue improving upon the available genotypes and detect high quality solutions.

This paper proposes a Differential Evolution based algorithm for numerical optimization in the presence of noise. The proposed algorithm, namely Noise Analysis Differential Evolution (NADE), employs a randomized scale factor in order to overcome the structural difficulties of a Differential Evolution in a noisy environment as well as a noise analysis component which determines the amount of samples required for characterizing the stochastic process and thus efficiently performing pairwise comparisons between parent and offspring solutions. The NADE has been compared, for a benchmark set composed of various fitness landscapes tinder several levels of noise bandwidth, with a classical evolutionary algorithm for noisy optimization and two recently proposed metaheuristics. Numerical results show that the proposed NADE has a very good performance in detecting high quality solutions despite the presence of noise. The NADE seems, in most cases, very fast and reliable in detecting promising search directions and continuing evolution towards the optimum.

Automatic target detection and tracking systems are used extensively in complex scenes. In long-term tracking, some visual attributes of objects are changing, such as illumination, size, profile, and so on. To address the issue, it is particularly important to describe the essential properties of the objects in tracking. An enhanced kernelized correlation filter tracking strategy fused multiple features with location prediction is proposed. To make the object appearance models more accuracy and robustness, based on the original histogram of oriented gradient features, we integrate the hue, saturation, value, and grayscale information to construct a new descriptor to represent the target appearance. Moreover, location prediction and bi-linear interpolation are employed to obtain the more accurate target position. Experiments show that the proposed strategy can obtain superior or competitive performance in challenging benchmark data sets. In practice, the algorithm is applied to track shuttle bus targets in the airport apron.

Back propagation (BP) is widely used for parameter search of fully-connected layers in many neural networks. Although BP has the potential of quickly converging to a solution, due to its gradient-based nature, it tends to fall into a local optimum. Metaheuristics such as evolutionary computation (EC) techniques, as gradient-free methods, may have excellent global search capability due to their stochastic nature. However, these techniques tend to perform worse than BP in terms of convergence speed. In this paper, a hybrid gradient descent search algorithm (HGDSA) is proposed for training the parameters in fully-connected neural networks. HGDSA initially searches the space extensively by means of an ensemble of gradient descent strategies in the early stage and then uses BP as an exploitative local search operator. Moreover, a self-adaptive method which selects strategies and updates the learning rates of strategies has been designed and embedded in the global search operators to prevent stagnation in local optima. To verify the effectiveness of HGDSA, experiments were performed on eleven classification datasets. Experimental results demonstrate that the proposed HGDSA possesses both powerful global and local search abilities. Furthermore, the proposed approach appears to be promising also on high-dimensional datasets.

A Generative Adversarial Network (GAN) can learn the relationship between two image domains and achieve unpaired image-to-image translation. One of the breakthroughs was Cycle-consistent Generative Adversarial Networks (CycleGAN), which is a popular method to transfer the content representations from the source domain to the target domain. Existing studies have gradually improved the performance of CycleGAN models by modifying the network structure or loss function of CycleGAN. However, these methods tend to suffer from training instability and the generators lack the ability to acquire the most discriminating features between the source and target domains, thus making the generated images of low fidelity and few texture details. To overcome these issues, the present paper proposes a new method that combines Evolutionary Algorithms (EAs) and Attention Mechanisms to train GANs. Specifically, from an initial CycleGAN, binary vectors indicating the activation of the weights of the generators are progressively improved upon by means of an EA. At the end of this process, the best-performing configurations of generators can be retained for image generation. In addition, to address the issues of low fidelity and lack of texture details on generated images, we make use of the channel attention mechanism. The latter component allows the candidate generators to learn important features of real images and thus generate images with higher quality. The experiments demonstrate qualitatively and quantitatively that the proposed method, namely Attention evolutionary GAN (AevoGAN) alleviates the training instability problems of CycleGAN training. In the test results, the proposed method can generate higher-quality images and obtain better results than the CycleGAN training methods present in the literature, in terms of Inception Score (IS), Fréchet Inception Distance (FID) and Kernel Inception Distance (KID).

In this paper, a hybrid feature selection algorithm based on a multi-objective algorithm with ReliefF (MOFS-RFGA) is proposed. Combining the advantages of filter and wrapper methods, the two types of algorithms are hybridized to improve the capability when solving feature selection problems. First, the ReliefF algorithm is used to score the features according to their importance to the instance class. Then, the feature scoring information is used to initialize the population. Also, a new crossover and mutation operator are designed in this paper to guide the crossover and mutation process based on feature scoring information to improve the search direction of MOFS-RFGA in search space and enhance the convergence performance. In the experiments, MOFS-RFGA is compared with seven advanced multi-objective feature selection algorithms on 20 datasets, and the results show that MOFS-RFGA can fully utilize the advantages of filter and wrapper methods, beating the excellent performance of the comparison algorithms on a large number of datasets, and ensuring good classification performance while cutting a large number of features.

Biological brains have a natural capacity for resolving certain classification tasks. Studies on biologically plausible spiking neurons, architectures and mechanisms of artificial neural systems that closely match biological observations while giving high classification performance are gaining momentum. Spiking neural P systems (SN P systems) are a class of membrane computing models and third-generation neural networks that are based on the behavior of biological neural cells and have been used in various engineering applications. Furthermore, SN P systems are characterized by a highly flexible structure that enables the design of a machine learning algorithm by mimicking the structure and behavior of biological cells without the over-simplification present in neural networks. Based on this aspect, this paper proposes a novel type of SN P system, namely, layered SN P system (LSN P system), to solve classification problems by supervised learning. The proposed LSN P system consists of a multi-layer network containing multiple weighted fuzzy SN P systems with adaptive weight adjustment rules. The proposed system employs specific ascending dimension techniques and a selection method of output neurons for classification problems. The experimental results obtained using benchmark datasets from the UCI machine learning repository and MNIST dataset demonstrated the feasibility and effectiveness of the proposed LSN P system. More importantly, the proposed LSN P system presents the first SN P system that demonstrates sufficient performance for use in addressing real-world classification problems.

Spiking neural P systems (SN P systems), inspired by biological neurons, are introduced as symbolical neural-like computing models that encode information with multisets of symbolized spikes in neurons and process information by using spike-based rewriting rules. Inspired by neuronal activities affected by enzymes, numerical variants of SN P systems called enzymatic numerical spiking neural P systems (ENSNP systems) are proposed wherein each neuron has a set of variables with real values and a set of enzymatic activation-production spiking rules, and each synapse has an assigned weight. By using spiking rules, ENSNP systems can directly implement mathematical methods based on real numbers and continuous functions. Furthermore, ENSNP systems are used to model ENSNP membrane controllers for robots implementing wall following. The trajectories, distances from the wall, and wheel speeds of robots with ENSNP membrane controllers for wall following are compared with those of a robot with a membrane controller for wall following. The average error values of the designed ENSNP membrane controllers are compared with three recently fuzzy logical controllers with optimization algorithms for wall following. The experimental results showed that the designed ENSNP membrane controllers can be candidates as efficient controllers to control robots implementing the task of wall following.

Spiking neural P systems (abbreviated as SNP systems) are models of computation that mimic the behavior of biological neurons. The spiking neural P systems with communication on request (abbreviated as SNQP systems) are a recently developed class of SNP system, where a neuron actively requests spikes from the neighboring neurons instead of passively receiving spikes. It is already known that small SNQP systems, with four unbounded neurons, can achieve Turing universality. In this context, ‘unbounded’ means that the number of spikes in a neuron is not capped. This work investigates the dependency of the number of unbounded neurons on the computation capability of SNQP systems. Specifically, we prove that (1) SNQP systems composed entirely of bounded neurons can characterize the family of finite sets of numbers; (2) SNQP systems containing two unbounded neurons are capable of generating the family of semilinear sets of numbers; (3) SNQP systems containing three unbounded neurons are capable of generating nonsemilinear sets of numbers. Moreover, it is obtained in a constructive way that SNQP systems with two unbounded neurons compute the operations of Boolean logic gates, i.e., OR, AND, NOT, and XOR gates. These theoretical findings demonstrate that the number of unbounded neurons is a key parameter that influences the computation capability of SNQP systems.

Generative adversarial networks have made remarkable achievements in generative tasks. However, instability and mode collapse are still frequent problems. We improve the framework of evolutionary generative adversarial networks (E-GANs), calling it phased evolutionary generative adversarial networks (PEGANs), and adopt a self-attention module to improve upon the disadvantages of convolutional operations. During the training process, the discriminator will play against multiple generators simultaneously, where each generator adopts a different objective function as a mutation operation. Every time after the specified number of training iterations, the generator individuals will be evaluated and the best performing generator offspring will be retained for the next round of evolution. Based on this, the generator can continuously adjust the training strategy during training, and the self-attention module also enables the model to obtain the modeling ability of long-range dependencies. Experiments on two datasets showed that PEGANs improve the training stability and are competitive in generating high-quality samples.

Adam is an adaptive gradient descent approach that is commonly used in back-propagation (BP) algorithms for training feed-forward neural networks (FFNNs). However, it has the defect that it may easily fall into local optima. To solve this problem, some metaheuristic approaches have been proposed to train FFNNs. While these approaches have stronger global search capabilities enabling them to more readily escape from local optima, their convergence performance is not as good as that of Adam. The proposed algorithm is an ensemble of differential evolution and Adam (EDEAdam), which integrates a modern version of the differential evolution algorithm with Adam, using two different sub-algorithms to evolve two sub-populations in parallel and thereby achieving good results in both global and local search. Compared with traditional algorithms, the integration of the two algorithms endows EDEAdam with powerful capabilities to handle different classification problems. Experimental results prove that EDEAdam not only exhibits improved global and local search capabilities, but also achieves a fast convergence speed.

To guarantee their locomotion, biped robots need to walk stably. The latter is achieved by a high performance in joint control. This article addresses this issue by proposing a novel human-simulated fuzzy (HF) membrane control system of the joint angles. The proposed control system, human-simulated fuzzy membrane controller (HFMC), contains several key elements. The first is an HF algorithm based on human-simulated intelligent control (HSIC). This HF algorithm incorporates elements of both multi-mode proportional-derivative (PD) and fuzzy control, aiming at solving the chattering problem of multi-mode switching while improving control accuracy. The second is a membrane architecture that makes use of the natural parallelisation potential of membrane computing to improve the real-time performance of the controller. The proposed HFMC is utilised as the joint controller for a biped robot. Numerical tests in a simulation are carried out with the planar and slope walking of a five-link biped robot, and the effectiveness of the HFMC is verified by comparing and evaluating the results of the designed HFMC, HSIC and PD. Experimental results demonstrate that the proposed HFMC not only retains the advantages of traditional PD control but also improves control accuracy, real-time performance and stability.

Feature selection (FS) is an important data pre-processing technique in classification. In most cases, FS can improve classification accuracy and reduce feature dimension, so it can be regarded as a multi-objective optimization problem. Many evolutionary computation techniques have been applied to FS problems and achieved good results. However, an increase in data dimension means that search difficulty also greatly increases, and EC algorithms with insufficient search ability maybe only find sub-optimal solutions in high probability. Moreover, an improper initial population may negatively affect the convergence speed of algorithms. To solve the problems highlighted above, this paper proposes MOEA-ISa: a multi-objective evolutionary algorithm with interval based initialization and self-adaptive crossover operator for large-scale FS. The proposed interval based initialization can limit the number of selected features for solution to improve the distribution of the initial population in the target space and reduce the similarity of the initial population in the decision space. The proposed self-adaptive crossover operator can determine the number of nonzero genes in offspring according to the similarity of parents, and it combines with the feature weights obtained by ReliefF method to improve the quality of offspring. In the experiments, the proposed algorithm was compared with six other algorithms on 13 benchmark UCI datasets and two benchmark LIBSVM datasets, and an ablation experiment was performed on MOEA-ISa. The results show that MOEA-ISa's performance is better than the six other algorithms for solving large-scale FS problems, and the proposed interval based initialization and self-adaptive crossover operator can effectively improve the performance of MOEA-ISa. The source code of MOEA-ISa is available on GitHub at https://github.com/xueyunuist/MOEA-ISa.