LDAS&ET-AD:Learnable Distillation Attack Strategies and Evolvable Teachers Adversarial Distillation

Adversarial distillation(AD)has emerged as a potential solution to tackle the challenging optimization problem of loss with hard labels in adversarial training.However,fixed sample-agnostic and student-egocentric attack strategies are unsuitable for distillation.Additionally,the reliability of guidance from static teachers diminishes as target models become more robust.This paper proposes an AD method called Learnable Distillation Attack Strategies and Evolvable Teachers Adversarial Distillation(LDAS&ET-AD).Firstly,a learnable distillation attack strategies generating mechanism is developed to automatically generate sample-dependent attack strategies tailored for distillation.A strategy model is introduced to produce attack strategies that enable adversarial examples(AEs)to be created in areas where the target model significantly diverges from the teachers by competing with the target model in minimizing or maximizing the AD loss.Secondly,a teacher evolution strategy is introduced to enhance the reliability and effectiveness of knowledge in improving the generalization performance of the target model.By calculating the experimentally updated target model’s validation performance on both clean samples and AEs,the impact of distillation from each training sample and AE on the target model’s generalization and robustness abilities is assessed to serve as feedback to fine-tune standard and robust teachers accordingly.Experiments evaluate the performance of LDAS&ET-AD against different adversarial attacks on the CIFAR-10 and CIFAR-100 datasets.The experimental results demonstrate that the proposed method achieves a robust precision of 45.39%and 42.63%against AutoAttack(AA)on the CIFAR-10 dataset for ResNet-18 and MobileNet-V2,respectively,marking an improvement of 2.31%and 3.49%over the baseline method.In comparison to state-of-the-art adversarial defense techniques,our method surpasses Introspective Adversarial Distillation,the top-performing method in terms of robustness under AA attack for the CIFAR-10 dataset,with enhancements of 1.40%and 1.43%for ResNet-18 and MobileNet-V2,respectively.These findings demonstrate the effectiveness of our proposed method in enhancing the robustness of deep learning networks(DNNs)against prevalent adversarial attacks when compared to other competing methods.In conclusion,LDAS&ET-AD provides reliable and informative soft labels to one of the most promising defense methods,AT,alleviating the limitations of untrusted teachers and unsuitable AEs in existing AD techniques.We hope this paper promotes the development of DNNs in real-world trust-sensitive fields and helps ensure a more secure and dependable future for artificial intelligence systems.

Evolution Performance of Symbolic Radial Basis Function Neural Network by Using Evolutionary Algorithms

Radial Basis Function Neural Network(RBFNN)ensembles have long suffered from non-efficient training,where incorrect parameter settings can be computationally disastrous.This paper examines different evolutionary algorithms for training the Symbolic Radial Basis Function Neural Network(SRBFNN)through the behavior’s integration of satisfiability programming.Inspired by evolutionary algorithms,which can iteratively find the nearoptimal solution,different Evolutionary Algorithms(EAs)were designed to optimize the producer output weight of the SRBFNN that corresponds to the embedded logic programming 2Satisfiability representation(SRBFNN-2SAT).The SRBFNN’s objective function that corresponds to Satisfiability logic programming can be minimized by different algorithms,including Genetic Algorithm(GA),Evolution Strategy Algorithm(ES),Differential Evolution Algorithm(DE),and Evolutionary Programming Algorithm(EP).Each of these methods is presented in the steps in the flowchart form which can be used for its straightforward implementation in any programming language.With the use of SRBFNN-2SAT,a training method based on these algorithms has been presented,then training has been compared among algorithms,which were applied in Microsoft Visual C++software using multiple metrics of performance,including Mean Absolute Relative Error(MARE),Root Mean Square Error(RMSE),Mean Absolute Percentage Error(MAPE),Mean Bias Error(MBE),Systematic Error(SD),Schwarz Bayesian Criterion(SBC),and Central Process Unit time(CPU time).Based on the results,the EP algorithm achieved a higher training rate and simple structure compared with the rest of the algorithms.It has been confirmed that the EP algorithm is quite effective in training and obtaining the best output weight,accompanied by the slightest iteration error,which minimizes the objective function of SRBFNN-2SAT.

An Evolution Strategy for the Induction of Fuzzy Finite-State Automata

This paper presents an evolution strategy to induce fuzzy finite-state automata from examples of fuzzy languages. The coding, fitness function of a generated automaton and corresponding mutation operators are given respectively. The application example given at last shows the effectiveness of the proposed evolution strategy for automata induction.

A New Model for Network Security Situation Assessment of the Industrial Internet

To address the problem of network security situation assessment in the Industrial Internet,this paper adopts the evidential reasoning(ER)algorithm and belief rule base(BRB)method to establish an assessment model.First,this paper analyzes the influencing factors of the Industrial Internet and selects evaluation indicators that contain not only quantitative data but also qualitative knowledge.Second,the evaluation indicators are fused with expert knowledge and the ER algorithm.According to the fusion results,a network security situation assessment model of the Industrial Internet based on the ER and BRB method is established,and the projection covariance matrix adaptive evolution strategy(P-CMA-ES)is used to optimize the model parameters.This method can not only utilize semiquantitative information effectively but also use more uncertain information and prevent the problem of combinatorial explosion.Moreover,it solves the problem of the uncertainty of expert knowledge and overcomes the problem of low modeling accuracy caused by insufficient data.Finally,a network security situation assessment case of the Industrial Internet is analyzed to verify the effectiveness and superiority of the method.The research results showthat this method has strong applicability to the network security situation assessment of complex Industrial Internet systems.It can accurately reflect the actual network security situation of Industrial Internet systems and provide safe and reliable suggestions for network administrators to take timely countermeasures,thereby improving the risk monitoring and emergency response capabilities of the Industrial Internet.

Secured Cloud Data Migration Technique by Competent Probabilistic Public Key Encryption

Cloud computing, a recently emerged paradigm faces major challenges in achieving the privacy of migrated data, network security, etc. Too many cryptographic technologies are raised to solve these issues based on identity, attributes and prediction algorithms yet;these techniques are highly prone to attackers. This would raise a need of an effective encryption technique, which would ensure secure data migration. With this scenario, our proposed methodology Efficient Probabilistic Public Key Encryption(EPPKE) is optimized with Covariance Matrix Adaptation Evolution Strategies(CMA-ES). It ensures data integrity through the Luhn algorithm with BLAKE 2b encapsulation. This enables an optimized security to the data which is migrated through cloud. The proposed methodology is implemented in Open Stack with Java Language. It achieves better results by providing security compared to other existing techniques like RSA, IBA, ABE, PBE, etc.

Quantum learning control using differential evolution with equally-mixed strategies

Learning control has been recognized as a powerful approach in quantum information technology. In this paper, we extend the application of differential evolution （DE） to design optimal control for various quantum systems. Various DE methods are introduced and analyzed, and EMSDE featuring in equally mixed strategies is employed for quantum control. Two classes of quantum control problems, including control of four-level open quantum ensembles and quantum superconducting systems, are investigated to demonstrate the performance of EMSDE for learning control of quantum systems. Numerical results verify the effectiveness of the FMSDE method for various quantum systems and show the potential for complex quantum control problems.

AERODYNAMIC OPTIMIZATION DESIGN OF LOW ASPECT RATIO TRANSONIC TURBINE STAGE

The advanced optimization method named as adaptive range differential evolution （ARDE） is developed. The optimization performance of ARDE is demonstrated using a typical mathematical test and compared with the standard genetic algorithm and differential evolution. Combined with parallel ARDE, surface modeling method and Navier-Stokes solution, a new automatic aerodynamic optimization method is presented. A low aspect ratio transonic turbine stage is optimized for the maximization of the isentropic efficiency with forty-one design variables in total. The coarse-grained parallel strategy is applied to accelerate the design process using 15 CPUs. The isentropic efficiency of the optimum design is 1.6% higher than that of the reference design. The aerodynamic performance of the optimal design is much better than that of the reference design.

Aberration correction for multiphoton microscopy using covariance matrix adaptation evolution strategy

Multiphoton microscopy is the enabling tool for biomedical research,but the aberrations of biological tissues have limited its imaging performance.Adaptive optics(AO)has been developed to partially overcome aberration to restore imaging performance.For indirect AO,algorithm is the key to its successful implementation.Here,based on the fact that indirect AO has an analogy to the black-box optimization problem,we successfully apply the covariance matrix adaptation evolution strategy(CMA-ES)used in the latter,to indirect AO in multiphoton microscopy(MPM).Compared with the traditional genetic algorithm(GA),our algorithm has a greater improvement in convergence speed and convergence accuracy,which provides the possibility of realizing real-time dynamic aberration correction for deep in vivo biological tissues.

Differential Evolution-Boosted Sine Cosine Golden Eagle Optimizer with Lévy Flight

Golden eagle optimizer(GEO)is a recently introduced nature-inspired metaheuristic algorithm,which simulates the spiral hunting behavior of golden eagles in nature.Regrettably,the GEO suffers from the challenges of low diversity,slow iteration speed,and stagnation in local optimization when dealing with complicated optimization problems.To ameliorate these deficiencies,an improved hybrid GEO called IGEO,combined with Lévy flight,sine cosine algorithm and differential evolution(DE)strategy,is developed in this paper.The Lévy flight strategy is introduced into the initial stage to increase the diversity of the golden eagle population and make the initial population more abundant;meanwhile,the sine-cosine function can enhance the exploration ability of GEO and decrease the possibility of GEO falling into the local optima.Furthermore,the DE strategy is used in the exploration and exploitation stage to improve accuracy and convergence speed of GEO.Finally,the superiority of the presented IGEO are comprehensively verified by comparing GEO and several state-of-the-art algorithms using(1)the CEC 2017 and CEC 2019 benchmark functions and(2)5 real-world engineering problems respectively.The comparison results demonstrate that the proposed IGEO is a powerful and attractive alternative for solving engineering optimization problems.

Automatic Approach to Ontology Evolution Based on Change Impact Comparisons

Ontology evolution is the timely adaptation of ontologies to changing requirements, which is becoming more and more important as ontologies become widely used in different fields. This paper shows how to address the problem of evolving ontologies with less manual case-based reasoning using an automatic selection mechanism. An automatic ontology evolution strategy selection framework is presented that automates the evolution. A minimal change impact algorithm is also developed for the framework. The method is shown to be effective in a case study.

Efficient multi-objective CMA-ES algorithm assisted by knowledge-extraction-based variable-fidelity surrogate model

To accelerate the multi-objective optimization for expensive engineering cases, a Knowledge-Extraction-based Variable-Fidelity Surrogate-assisted Covariance Matrix Adaptation Evolution Strategy(KE-VFS-CMA-ES) is presented. In the first part, the KE-VFS model is established. Firstly, the optimization is performed using the low-fidelity surrogate model to obtain the Low-Fidelity Non-Dominated Solutions(LF-NDS). Secondly, aiming to obtain the High-Fidelity(HF) sample points located in promising areas, the K-means clustering algorithm and the space-filling strategy are used to extract knowledge from the LF-NDS to the HF space. Finally,the KE-VFS model is established by means of the obtained HF and LF sample points. In the second part, a novel model management based on the Modified Hypervolume Improvement(MHVI) criterion and pre-screening strategy is proposed. In each generation of KE-VFS-CMA-ES, excessive candidate points are firstly generated and then calculated by the MHVI criterion to find out a few potential points, which will be evaluated by the HF model. Through the above two parts,the promising areas can be detected and the potential points can be screened out, which contributes to speeding up the optimization process twofold. Three classic benchmark functions and a time-consuming engineering case of the aerospace integrally stiffened shell are studied, and results illustrate the excellent efficiency, robustness and applicability of KE-VFS-CMA-ES compared with other four known multi-objective optimization algorithms.

A hybrid multi-objective optimization method considering optimization problems in power distribution systems

Various kinds of new engineering technologies have been studied to realize the low-carbon and sustainable power supply systems all over the world.In actual implementation of these technologies,mostly,there are multiple objectives with trade off relationships among each other,and also various constraints in the achievement of these objectives.Therefore,it should be essential to solve multiobjective optimization problems effectively in the applications of these new technologies in power systems.This paper proposes an improved method to realize multiobjective optimization for critical challenges in advanced power systems.To realize that,in an optimal dispersed generation installation problem,that is,one of effective measures for low-carbon power systems,various optimization methods and their combination methods are evaluated and a hybrid method for evolutionary algorithms was developed.The method can provide improved results compared with other state-of-the-art multi-objective optimization methods.

Network security situation automatic prediction model based on accumulative CMA-ES optimization

To improve the accuracy of the network security situation, a security situation automatic prediction model based on accumulative data preprocess and support vector machine （SVM） optimized by covariance matrix adaptive evolutionary strategy （CMA-ES） is proposed. The proposed model adopts SVM which has strong nonlinear ability. Also, the hyper parameters for SVM are optimized through the CMA-ES which owns good performance in finding optimization automatically. Considering the irregularity of network security situation values, we accumulate the original sequence, so that the internal rules of discrete data can be revealed and it is easy to model. Simulation experiments show that the proposed model has faster convergence-speed and higher prediction accuracy than other extant prediction models.