Evolutionary Neural Architecture Search and Its Applications in Healthcare

Most of the neural network architectures are based on human experience,which requires a long and tedious trial-and-error process.Neural architecture search(NAS)attempts to detect effective architectures without human intervention.Evolutionary algorithms(EAs)for NAS can find better solutions than human-designed architectures by exploring a large search space for possible architectures.Using multiobjective EAs for NAS,optimal neural architectures that meet various performance criteria can be explored and discovered efficiently.Furthermore,hardware-accelerated NAS methods can improve the efficiency of the NAS.While existing reviews have mainly focused on different strategies to complete NAS,a few studies have explored the use of EAs for NAS.In this paper,we summarize and explore the use of EAs for NAS,as well as large-scale multiobjective optimization strategies and hardware-accelerated NAS methods.NAS performs well in healthcare applications,such as medical image analysis,classification of disease diagnosis,and health monitoring.EAs for NAS can automate the search process and optimize multiple objectives simultaneously in a given healthcare task.Deep neural network has been successfully used in healthcare,but it lacks interpretability.Medical data is highly sensitive,and privacy leaks are frequently reported in the healthcare industry.To solve these problems,in healthcare,we propose an interpretable neuroevolution framework based on federated learning to address search efficiency and privacy protection.Moreover,we also point out future research directions for evolutionary NAS.Overall,for researchers who want to use EAs to optimize NNs in healthcare,we analyze the advantages and disadvantages of doing so to provide detailed guidance,and propose an interpretable privacy-preserving framework for healthcare applications.

Evolutionary Variational YOLOv8 Network for Fault Detection in Wind Turbines

Deep learning has emerged in many practical applications,such as image classification,fault diagnosis,and object detection.More recently,convolutional neural networks(CNNs),representative models of deep learning,have been used to solve fault detection.However,the current design of CNNs for fault detection of wind turbine blades is highly dependent on domain knowledge and requires a large amount of trial and error.For this reason,an evolutionary YOLOv8 network has been developed to automatically find the network architecture for wind turbine blade-based fault detection.YOLOv8 is a CNN-backed object detection model.Specifically,to reduce the parameter count,we first design an improved FasterNet module based on the Partial Convolution(PConv)operator.Then,to enhance convergence performance,we improve the loss function based on the efficient complete intersection over the union.Based on this,a flexible variable-length encoding is proposed,and the corresponding reproduction operators are designed.Related experimental results confirmthat the proposed approach can achieve better fault detection results and improve by 2.6%in mean precision at 50(mAP50)compared to the existing methods.Additionally,compared to training with the YOLOv8n model,the YOLOBFE model reduces the training parameters by 933,937 and decreases the GFLOPS(Giga Floating Point Operations Per Second)by 1.1.

Evolutionary Computation Based Optimization of Image Zernike Moments Shape Feature Vector

The image shape feature can be described by the image Zernike moments. In this paper, we points out the problem that the high dimension image Zernike moments shape feature vector can describe more detail of the original image but has too many elements making trouble for the next image analysis phases. Then the low dimension image Zernike moments shape feature vector should be improved and optimized to describe more detail of the original image. So the optimization algorithm based on evolutionary computation is designed and implemented in this paper to solve this problem. The experimental results demonstrate the feasibility of the optimization algorithm.

Orbit Design for Responsive Space Using Multiple-objective Evolutionary Computation

Responsive orbits have exhibited advantages in emergencies for their excellent responsiveness and coverage to targets.Generally,there are several conflicting metrics to trade in the orbit design for responsive space.A special multiple-objective genetic algorithm,namely the Nondominated Sorting Genetic AlgorithmⅡ(NSGAⅡ),is used to design responsive orbits.This algorithm has considered the conflicting metrics of orbits to achieve the optimal solution,including the orbital elements and launch programs of responsive vehicles.Low-Earth fast access orbits and low-Earth repeat coverage orbits,two subtypes of responsive orbits,can be designed using NSGAI under given metric tradeoffs,number of vehicles,and launch mode.By selecting the optimal solution from the obtained Pareto fronts,a designer can process the metric tradeoffs conveniently in orbit design.Recurring to the flexibility of the algorithm,the NSGAI promotes the responsive orbit design further.

Modeling and Simulation for High Energy Sub-Nuclear Interactions Using Evolutionary Computation Technique

High energy sub-nuclear interactions are a good tool to dive deeply in the core of the particles to recognize their structures and the forces governed. The current article focuses on using one of the evolutionary computation techniques, the so-called genetic programming (GP), to model the hadron nucleus (h-A) interactions through discovering functions. In this article, GP is used to simulate the rapidity distribution of total charged, positive and negative pions for p--Ar and p--Xe interactions at 200 GeV/c and charged particles for p-pb collision at 5.02 TeV. We have done so many runs to select the best runs of the GP program and finally obtained the rapidity distribution as a function of the lab momentum , mass number (A) and the number of particles per unit solid angle (Y). In all cases studied, we compared our seven discovered functions produced by GP technique with the corresponding experimental data and the excellent matching was so clear.

Computational Simulations of Bone Remodeling under Natural Mechanical Loading or Muscle Malfunction Using Evolutionary Structural Optimization Method

Live bone inherently responds to applied mechanical stimulus by altering its internal tissue composition and ultimately biomechanical properties, structure and function. The final formation may structurally appear inferior by design but complete by function. To understand the loading response, this paper numerically investigated structural remodeling of mature sheep femur using evolutionary structural optimization method (ESO). Femur images from Computed Tomography scanner were used to determine the elastic modulus variation and subsequently construct finite element model of the femur with stiffest elasticity measured. Major muscle forces on dominant phases of healthy sheep gait were imposed on the femur under static mode. ESO was applied to progressively alter the remodeling of numerically simulated femur from its initial to final design by iteratively removing elements with low strain energy density (SED). The computations were repeated with two different mesh sizes to test the convergence. The elements within the medullary canal had low SEDs and therefore were removed during the optimization. The SEDs in the remaining elements varied with angle around the circumference of the shaft. Those elements with low SED were inefficient in supporting the load and thus fundamentally explained how bone remodels itself with less stiff inferior tissue to meet load demand. This was in line with the Wolff’s law of transformation of bone. Tissue growth and remodeling process was found to shape the sheep femur to a mechanically optimized structure and this was initiated by SED in macro-scale according to traditional principle of Wolff’s law.

Coordinated Route Planning via Nash Equilibrium and Evolutionary Computation

The coordinated route planning problem for multiple unmanned air vehicles （UAVs）, a cooperative optimization problem, also a non-cooperative game, is addressed in the framework of game theory, A Nash equilibrium based route planner is proposed. The rational is that the structure of UAV subteam usually provides some inherent and implicit preference information, which help to find the optimum coordinated routes and the optimum combination of the various objective functions. The route planner combines the concepts of evolutionary computation with problem-specific chromosome structures and evolutionary operators and handles different kinds of mission constraints in hierarchical style. Cooperation and competition among UAVs are reflected by the definition of fitness function. Simulations validate the feasibility and superiority of the game-theoretic coordinated routes planner.

Time Complexity of Evolutionary Algorithms for Combinatorial Optimization:A Decade of Results

Computational time complexity analyzes of evolutionary algorithms （EAs） have been performed since the mid-nineties. The first results were related to very simple algorithms, such as the （1＋1）-EA, on toy problems. These efforts produced a deeper understanding of how EAs perform on different kinds of fitness landscapes and general mathematical tools that may be extended to the analysis of more complicated EAs on more realistic problems. In fact, in recent years, it has been possible to analyze the （1＋1）-EA on combinatorial optimization problems with practical applications and more realistic population-based EAs on structured toy problems. This paper presents a survey of the results obtained in the last decade along these two research lines. The most common mathematical techniques are introduced, the basic ideas behind them are discussed and their elective applications are highlighted. Solved problems that were still open are enumerated as are those still awaiting for a solution. New questions and problems arisen in the meantime are also considered.

Evolutionary Multi-objective Portfolio Optimization in Practical Context

This paper addresses evolutionary multi-objective portfolio optimization in the practical context by incorporating realistic constraints into the problem model and preference criterion into the optimization search process. The former is essential to enhance the realism of the classical mean-variance model proposed by Harry Markowitz, since portfolio managers often face a number of realistic constraints arising from business and industry regulations, while the latter reflects the fact that portfolio managers are ultimately interested in specific regions or points along the efficient frontier during the actual execution of their investment orders. For the former, this paper proposes an order-based representation that can be easily extended to handle various realistic constraints like floor and ceiling constraints and cardinality constraint. An experimental study, based on benchmark problems obtained from the OR-library, demonstrates its capability to attain a better approximation of the efficient frontier in terms of proximity and diversity with respect to other conventional representations. The experimental results also illustrated its viability and practicality in handling the various realistic constraints. A simple strategy to incorporate preferences into the multi-objective optimization process is highlighted and the experimental study demonstrates its capability in driving the evolutionary search towards specific regions of the efficient frontier.

A New Algorithm for On-Line Handwriting Signature Verification Based on Evolutionary Computation

The paper proposes an on-line signature verification algorithm, through which test sample and template signatures can be optimizedly matched, based on evolutionary computation （EC）. Firstly, the similarity of signature curve segment is defined, and shift and scale transforms are also introduced due to the randoness of on-line signature. Secondly, this paper puts forward signature verification matching algorithm after establishment of the mathematical model. Thirdly, the concrete realization of the algorithm based on EC is discussed as well. In addition, the influence of shift and scale on the matching result is fully considered in the algorithm. Finally, a computation example is given, and the matching results between the test sample curve and the template signature curve are analyzed in detail, The preliminary experiments reveal that the type of signature verification problem can be solved by EC.

A Fast Clustering Based Evolutionary Algorithm for Super-Large-Scale Sparse Multi-Objective Optimization

During the last three decades,evolutionary algorithms(EAs)have shown superiority in solving complex optimization problems,especially those with multiple objectives and non-differentiable landscapes.However,due to the stochastic search strategies,the performance of most EAs deteriorates drastically when handling a large number of decision variables.To tackle the curse of dimensionality,this work proposes an efficient EA for solving super-large-scale multi-objective optimization problems with sparse optimal solutions.The proposed algorithm estimates the sparse distribution of optimal solutions by optimizing a binary vector for each solution,and provides a fast clustering method to highly reduce the dimensionality of the search space.More importantly,all the operations related to the decision variables only contain several matrix calculations,which can be directly accelerated by GPUs.While existing EAs are capable of handling fewer than 10000 real variables,the proposed algorithm is verified to be effective in handling 1000000 real variables.Furthermore,since the proposed algorithm handles the large number of variables via accelerated matrix calculations,its runtime can be reduced to less than 10%of the runtime of existing EAs.

Integrating Conjugate Gradients Into Evolutionary Algorithms for Large-Scale Continuous Multi-Objective Optimization

Large-scale multi-objective optimization problems(LSMOPs)pose challenges to existing optimizers since a set of well-converged and diverse solutions should be found in huge search spaces.While evolutionary algorithms are good at solving small-scale multi-objective optimization problems,they are criticized for low efficiency in converging to the optimums of LSMOPs.By contrast,mathematical programming methods offer fast convergence speed on large-scale single-objective optimization problems,but they have difficulties in finding diverse solutions for LSMOPs.Currently,how to integrate evolutionary algorithms with mathematical programming methods to solve LSMOPs remains unexplored.In this paper,a hybrid algorithm is tailored for LSMOPs by coupling differential evolution and a conjugate gradient method.On the one hand,conjugate gradients and differential evolution are used to update different decision variables of a set of solutions,where the former drives the solutions to quickly converge towards the Pareto front and the latter promotes the diversity of the solutions to cover the whole Pareto front.On the other hand,objective decomposition strategy of evolutionary multi-objective optimization is used to differentiate the conjugate gradients of solutions,and the line search strategy of mathematical programming is used to ensure the higher quality of each offspring than its parent.In comparison with state-of-the-art evolutionary algorithms,mathematical programming methods,and hybrid algorithms,the proposed algorithm exhibits better convergence and diversity performance on a variety of benchmark and real-world LSMOPs.

Using Evolutionary Computation to Solve Problems in Nonparametric Regression

This paper studies evolutionary mechanism of parameter selection in the construction of weight function for Nearest Neighbour Estimate in nonparametric regression. Construct an algorithm which adaptively evolves fine weight and makes good prediction about unknown points. The numerical experiments indicate that this method is effective. It is a meaningful discussion about practicability of nonparametric regression and methodology of adaptive model-building.

A New Evolutionary Algorithm for Solving Multi-Objective Optimization Problems

Multi-objective optimization is a new focus of evolutionary computation research. This paper puts forward a new algorithm, which can not only converge quickly, but also keep diversity among population efficiently, in order to find the Pareto-optimal set. This new algorithm replaces the worst individual with a newly-created one by 'multi-parent crossover' , so that the population could converge near the true Pareto-optimal solutions in the end. At the same time, this new algorithm adopts niching and fitness-sharing techniques to keep the population in a good distribution. Numerical experiments show that the algorithm is rather effective in solving some Benchmarks. No matter whether the Pareto front of problems is convex or non-convex, continuous or discontinuous, and the problems are with constraints or not, the program turns out to do well.

Bio-inspired computational techniques based on advanced condition monitoring

The application of bio-inspired computational techniques to the field of condition monitoring is addressed. First, the bio-inspired computational techniques are briefly addressed; the advantages and disadvantages of these computational methods are made clear. Then, the roles of condition monitoring in the predictive maintenance and failures prediction and the development trends of condition monitoring are discussed. Finally, a case study on the condition monitoring of grinding machine is described, which shows the application of bio-inspired computational technique to a practical condition monitoring system.

A GLOBAL OPTIMALITY CRITERION FOR EVOLUTIONARY COMPUTATION

Focussing on unification of concrete portions into a generic form of computational evolution,a generalized theoretical framework is necessary and imperative to be built to construct a universal computational theory of evolution machine.The NP problem solving capacity can be traced to the nature of metaevolution mechanism with emergence features that determine corresponding homeostasis and diversity ranging in the domain of nonlinnear mapping from genotype to phenotype.In this paper a criterion that guarantees the global optimality of evolutionary computation process is proposed and proven rigorously.The global optimization criterion obtained is based on the nonparametric measarement for the whole evolution system and has great flexibility and evolvability.It leaves room for evolutionary system designing and developement.The formulization of the global description in statistical manifold space of information object family expresses evoluable evolutionary operator architecture and operation procedure in terms of evolution by evolution.The theoretical results are helpful to applications such as machine learning for automatic knowledge acquisition,pattern classification and recognition of complex images(e.q.OCR) and unsupervised system identification of nonlinear dynamical systems as well as chaos phenomena.The kernal of the formal system guided by global evolutionary optimization is proper to the implementation with objectoriented programming paradigm and abstract machine modelling.

Evolutionary Computation for Image Feature Extraction

Modifications to an image feature extraction approach involving evolutionary computation and autonomous agents are proposed. The described algorithm allows extraction of features with certain specified characteristics, while omitting other undesirable details in the image. Experimental results are presented with remarks.

Evolutionary computation in China： A literature survey

Evolutionary computation （EC） has received significant attention in China during the last two decades. In this paper, we present an overview of the current state of this rapidly growing field in China. Chinese research in theoretical foundations of EC, EC-based optimization, EC-based data mining, and EC-based real-world applications are summarized.

The Applications of Evolutionary Computation in Software Reliability

Software reliability models(SRMs) are the theoretic foundation of software reliability. However, the existence of intrinsic limitation of the preposition in traditional model building confines the applications of SRMs. In this paper, a new method,evolutionary computation,is used to estimate parameters of SRMs .At the same time, new algorithms are also proposed and employed to build SRMs. As the experiment results demonstrate, evolutionary computation method is po'verful and effective.

Recommender Systems Based on Evolutionary Computing: A Survey

Data mining techniques and information personalization have made significant growth in the past decade. Enormous volume of data is generated every day. Recommender systems can help users to find their specific information in the extensive volume of information. Several techniques have been presented for development of Recommender System (RS). One of these techniques is the Evolutionary Computing (EC), which can optimize and improve RS in the various applications. This study investigates the number of publications, focusing on some aspects such as the recommendation techniques, the evaluation methods and the datasets which are used.