A novel hybrid BPSO–SCA approach for feature selection

Natural Computing - Nature is a great source of inspiration for solving complex problems in real-world. In this paper, a hybrid nature-inspired algorithm is proposed for feature selection problem....     

A novel hybrid PSO–GWO algorithm for optimization problems

Engineering with Computers - In this study, we propose a new hybrid algorithm fusing the exploitation ability of the particle swarm optimization (PSO) with the exploration ability of the grey wolf...     

A novel minorization–maximization framework for simultaneous feature selection and clustering of high-dimensional count data

Pattern Analysis and Applications - Count data are commonly exploited in machine learning and computer vision applications; however, they often suffer from the well-known curse of dimensionality,...     

A multi-objective cost–benefit optimization algorithm for network hardening

International Journal of Information Security - Network hardening is an optimization problem to find the best combination of countermeasures to protect a network from cyber-attacks. While an...     

An orthogonal opposition-based-learning Yin–Yang-pair optimization algorithm for engineering optimization

Yin–Yang-pair Optimization (YYPO) is a recently developed philosophy-inspired meta-heuristic algorithm, which works with two main points for exploitation and exploration, respectively, and then generates more points via splitting to search the global optimum. However, it suffers from low quality of candidate solutions in its exploration process owing to the lack of elitism. Inspired by this, a new modified algorithm named orthogonal opposition-based-learning Yin–Yang-pair Optimization (OOYO) is proposed to enhance the performance of YYPO. First, the OOYO retains the normalization operation in YYPO and starts with a single point to exploit. A set of opposite points is designed by a method of opposition-based learning with split points generated from the current optimum for exploration. Then, the points, i.e., candidate solutions, are constructed by the randomly selected split point and opposite points through the idea of orthogonal experiment design to make full use of information from the space. The proposed OOYO does not add additional time complexity and eliminates a user-defined parameter in YYPO, which facilitates parameter adjustment. The novel orthogonal opposition-based learning strategy can provide inspirations for the improvement of other optimization algorithms. Extensive test functions containing a classic test suite of 23 standard benchmark functions and 2 test suites of Swarm Intelligence Symposium 2005 and Congress on Evolutionary Computation 2020 from Institute of Electrical and Electronics Engineers are employed to evaluate the proposed algorithm. Non-parametric statistical results demonstrate that OOYO outperforms YYPO and furnishes strong competitiveness compared with other state-of-the-art algorithms. In addition, we apply OOYO to solve four well-known constrained engineering problems and a practical problem of parameters optimization in a rainstorm intensity model.

     

Sequential multi-criteria feature selection algorithm based on agent genetic algorithm

A multi-criteria feature selection method-sequential multi-criteria feature selection algorithm (SMCFS) has been proposed for the applications with high precision and low time cost. By combining the consistency and otherness of different evaluation criteria, the SMCFS adopts more than one evaluation criteria sequentially to improve the efficiency of feature selection. With one novel agent genetic algorithm (chain-like agent GA), the SMCFS can obtain high precision of feature selection and low time cost that is similar as filter method with single evaluation criterion. Several groups of experiments are carried out for comparison to demonstrate the performance of SMCFS. SMCFS is compared with different feature selection methods using three datasets from UCI database. The experimental results show that the SMCFS can get low time cost and high precision of feature selection, and is very suitable for this kind of applications of feature selection.     

A hybrid teaching–learning slime mould algorithm for global optimization and reliability-based design optimization problems

Neural Computing and Applications - Slime mould algorithm (SMA) is a novel metaheuristic algorithm with good performance for optimization problems, but it may encounter premature or low accuracy in...     

BEMD–SIFT feature extraction algorithm for image processing application

Scale-invariant feature transform (SIFT) algorithm has been successfully applied to object recognition and to image feature extraction, which is a major application in the field of image processing. Nonetheless, the SIFT algorithm has not been solved effectively in practical applications that requires real-time performance, much calculation, and high storage capacity given the framework level and the iterative calculation process in the SIFT Gaussian blur operation. The extraction of image feature information is accelerated using the speeded-up robust features algorithm. However, this algorithm remains sensitive to complicated deformation. To address these problems, in this paper, we proposes a novel algorithmic framework based on bidimensional empirical mode decomposition (BEMD) and SIFT to extract self-adaptive features from images. First, the BEMD algorithm is used to decompose the self-adaptive features of the original image and to obtain multiple BIMF components. Second, the SIFT algorithm optimizes the extraction of parameters that reflect characteristic information on BIMF components. Related parameters are obtained through genetic algorithm optimization. Third, the method for extracting the characteristic information of the BIMF components involves synthesizing all of the accumulated characteristic information in the original image. Comparison results show that the method of calculating image feature extraction speed, accuracy, and reliability has a stronger effect than other methods.     

Genetic algorithm with logistic regression feature selection for Alzheimer’s disease classification

Neural Computing and Applications - Cognitive impairment must be diagnosed in Alzheimer’s disease as early as possible. Early diagnosis allows the person to receive effective treatment...     

A directional derivative computer algorithm for non-linear optimization†

This paper describes the computer solution to a non-linear optimization problem by the use of directional derivatives. Newton's method is used to find stationary points and when these points satisfy conditions of sufficiency, they are determined optima.     

A quantum particle swarm optimization for the 0–1 generalized knapsack sharing problem

This study proposes a new hybrid heuristic approach that combines the quantum particle swarm optimization (QPSO) technique with a local search phase to solve the binary generalized knapsack sharing problem (GKSP). The approach also incorporates a heuristic repair operator that uses problem-specific knowledge instead of the penalty function technique commonly used for constrained problems. This study is the first to report on the application of the QPSO method to the GKSP. The efficiency of our proposed approach was tested on a large set of instances, and the results were compared to those produced by the commercial mixed integer programming solver CPLEX 12.5 of IBM-ILOG. The Experimental results demonstrated the good performance of the QPSO in solving the GKSP.     

Solving 0–1 knapsack problem by a novel binary monarch butterfly optimization

Neural Computing and Applications - This paper presents a novel binary monarch butterfly optimization (BMBO) method, intended for addressing the 0–1 knapsack problem (0–1 KP). Two...     

A new conjugate gradient algorithm with cubic Barzilai–Borwein stepsize for unconstrained optimization

In this paper, a new conjugate gradient (CG) algorithm in Dai–Liao (DL) family is presented for solving unconstrained optimization problems. The proposed algorithm tries to adjust positive values for the so-called DL parameter by using quadratic and/or cubic models of the objective function. More precisely, the cubic regularization model of the objective function is properly employed when the non-positive curvature is detected. Besides, the CG parameter is introduced so that the generated CG directions are descent. Under some standard assumptions, we establish the convergence property of the new proposed algorithm. Numerical results on some test problems are reported. The results show that the new algorithm performs well and is competitive with CG_DESCENT method.     

A recursive algorithm for solving “a secret sharing” problem

Abstract

This paper presents a recursive algorithm for solving “a secret sharing” problem. This problem is one of the unsolved problems in the Second International Students Olympiad in Cryptography (NSUCRYPTO2015). Recently, Geut et al. solved the problem in a special case. We show that our algorithm is able to solve it in general.     

A novel optimized GA–Elman neural network algorithm

Neural Computing and Applications - The Elman neural network has good dynamic properties and strong global stability, being most widely used to deal with nonlinear, dynamic, and complex data....     

Feature clustering and feature discretization assisting gene selection for molecular classification using fuzzy c-means and expectation–maximization algorithm

The Journal of Supercomputing - In this paper, a novel gene selection benefiting from feature clustering and feature discretization is developed. In large numbers of genes, unsupervised fuzzy...     

Multiobjective optimization using an immunodominance and clonal selection inspired algorithm

Based on the mechanisms of immunodominance and clonal selection theory, we propose a new multiobjective optimization algorithm, immune dominance clonal multiobjective algorithm （IDCMA）. IDCMA is unique in that its fitness values of current dominated individuals are assigned as the values of a custom distance measure, termed as Ab-Ab affinity, between the dominated individuals and one of the nondominated individuals found so far. According to the values of Ab-Ab affinity, all dominated individuals （antibodies） are divided into two kinds, subdominant antibodies and cryptic antibodies. Moreover, local search only applies to the subdominant antibodies, while the cryptic antibodies are redundant and have no function during local search, but they can become subdominant （active） antibodies during the subsequent evolution. Furthermore, a new immune operation, clonal proliferation is provided to enhance local search. Using the clonal proliferation operation, IDCMA reproduces individuals and selects their improved maturated progenies after local search, so single individuals can exploit their surrounding space effectively and the newcomers yield a broader exploration of the search space. The performance comparison of IDCMA with MISA, NSGA-Ⅱ, SPEA, PAES, NSGA, VEGA, NPGA, and HLGA in solving six well-known multiobjective function optimization problems and nine multiobjective 0/1 knapsack problems shows that IDCMA has a good performance in converging to approximate Pareto-optimal fronts with a good distribution.