Shih-Yuan Chiu

An improved multi-objective particle swarm optimizer for multi-objective problems

This paper proposes an improved multi-objective particle swarm optimizer with proportional distribution and jump improved operation, named PDJI-MOPSO, for dealing with multi-objective problems. PDJI-MOPSO maintains diversity of new found non-dominated solutions via proportional distribution, and combines advantages of wide-ranged exploration and extensive exploitations of PSO in the external repository with the jump improved operation to enhance the solution searching abilities of particles. Introduction of cluster and disturbance allows the proposed method to sift through representative non-dominated solutions from the external repository and prevent solutions from falling into local optimum. Experiments were conducted on eight common multi-objective benchmark problems. The results showed that the proposed method operates better in five performance metrics when solving these benchmark problems compared to three other related works.

Cross-searching strategy for multi-objective particle swarm optimization

The main difference between an original PSO (single-objective) with a multi-objective PSO (MOPSO) is the local guide (global best solution) distribution must be redefined in order to obtain a set of non-dominated solutions (Pareto front). In MOPSO, the selection of local guide for particles will direct affect the performance of finding Pareto optimum. This paper presents a local guide assignment strategy for MOPSO called cross-searching strategy (CSS) which will distribute suitable local guides for particles to lead them toward to Pareto front and also keeping diversity of solutions. Experiments were conducted on several test functions and metrics from the standard literature on evolutionary multi-objective optimization. The results demonstrate good performance of the CSS for MOPSO in solving multi-objective problems when compare with recent approaches of multi-objective optimizer.

Particle swarm optimizer for multi-objective problems based on proportional distribution and cross-over operation

In multi-objective particle swarm optimization (MOPSO) methods, selecting the best local guide (the global best particle) for each particle of the population provides great benefits on the convergence and diversity of solutions, especially when problems are optimized with a large number of objectives. This paper introduces the proportional distribution based particle swarm optimizer (PSO) with cross-over operation in external repository for dealing with multi-objective problems. It combines advantages of wide-ranged exploration with cross-over operation, to maintain diversity of new found non-dominated solutions via proportional distribution in the external repository, and deep exploitations of PSO to enhance the solution searching abilities of particles. Introduction of cluster and disturbance allows the proposed method to sift through representative non-dominated solutions from the external repository and prevent solutions from falling into the local optimal. Experiments were conducted on four common MO benchmark problems. The results showed that the proposed method accomplishes better in four performance metrics when solving these benchmark problems compared to three other related works.

Differential evolution for strongly noisy optimization: Use 1:01 n resamplings at iteration n and Reach the −1/2 slope

This paper is devoted to noisy optimization in case of a noise with standard deviation as large as variations of the fitness values, specifically when the variance does not decrease to zero around the optimum. We focus on comparing methods for choosing the number of resamplings. Experiments are performed on the differential evolution algorithm. By mathematical analysis, we design a new rule for choosing the number of resamplings for noisy optimization, as a function of the dimension, and validate its efficiency compared to existing heuristics.

Adoptive population differential evolution with local search for solving large scale global optimization

Due to real-world optimization problems become increasingly complex. Algorithms are with higher efficiency and higher solution searching ability for finding global optimal solution in reasonable computing time is always needed. Thus, in this paper, an improved DE is proposed for solving large scale global optimization. The proposed method is incorporated with the population manager to eliminate redundant particles or to hire new ones or to maintain population size according to the solution searching status to make the process more efficient. Besides, a local search strategy is also involved to enhance populations solution search ability. Experiments were conducted on ten CEC 2012 test functions to present performance of the proposed method. The proposed method exhibits better performance than other three related works in solving most test functions.

Real Random Mutation Strategy for Differential Evolution

In this paper, an improved DE is proposed to improve optimization performance by implementing three new schemes: sharing mutation, current-to-better mutation and real-random-mutation. When evolution speed is standstill, sharing mutation can increase the search depth, in addition, real-random mutation can disturb individuals and can help individuals diverge to local optimum. When the evolution progresses well, current-to-better mutation will drive individuals to the correct evolution direction. Experiments were conducted on 15 of CEC 2005 test functions, include unimodal, multimodal and hybrid composition functions, to present performance of the proposed method and to compare with 5 variants of DE includes JADE, jDE, SaDE, DEGL and MDE_pBX. The proposed method exhibits better performance than other five related works in solving all the test functions.

Sharing mutation genetic algorithm for solving multi-objective problems

Multi-objective optimization (MO) has been an active area of research in last two decade. In multi-objective genetic algorithm (MOGA), quality of new generated offspring of population will affect the performance of finding Pareto optimum directly. In this paper, an improved MOGA is proposed named SMGA to solving multi-objective optimization problem. For improving solution searching efficiency, an effective mutation named sharing mutation is adopted for generating potential offspring. Experiments were conducted on CEC-09 MOP test problems. The results showed that the proposed method exhibits better performance when solving these benchmark problems compared to related multi-objective evolutionary algorithm (MOEA).

Parallel Evolutionary Algorithms Performing Pairwise Comparisons

We study mathematically and experimentally the convergence rate of differential evolution and particle swarm optimization for simple unimodal functions. Due to parallelization concerns, the focus is on lower bounds on the runtime, i.e. upper bounds on the speed-up, as a function of the population size. Two cases are particularly relevant: A population size of the same order of magnitude as the dimension and larger population sizes. We use the branching factor as a tool for proving bounds and get, as upper bounds, a linear speed-up for a population size similar to the dimension, and a logarithmic speed-up for larger population sizes. We then propose parametrizations for differential evolution and particle swarm optimization that reach these bounds.

Enhanced Differential Evolution Based on Adaptive Mutation and Wrapper Local Search Strategies for Global Optimization Problems

Differential evolution (DE) is a simple, powerful optimization algorithm, which has been widely used in many areas.However, the choices of the best mutation and search strategies are difficult for the specific issues. To alleviate thesedrawbacks and enhance the performance of DE, in this paper, the hybrid framework based on the adaptive mutationand Wrapper Local Search (WLS) schemes, is proposed to improve searching ability to efficiently guide the evolutionof the population toward the global optimum. Furthermore, the effective particle encoding representation namedParticle Segment Operation-Machine Assignment (PSOMA) that we previously published is applied to always producefeasible candidate solutions for solving the Flexible Job-shop Scheduling Problem (FJSP). Experiments wereconducted on comprehensive set of complex benchmarks including the unimodal, multimodal and hybrid compositionfunction, to validate performance of the proposed method and to compare with other state-of-the art DE variants suchas jDE, JADE, MDE_pBX etc. Meanwhile, the hybrid DE model incorporating PSOMA is used to solve differentrepresentative instances based on practical data for multi-objective FJSP verifications. Simulation results indicate thatthe proposed method performs better for the majority of the single-objective scalable benchmark functions in terms ofthe solution accuracy and convergence rate. In addition, the wide range of Pareto-optimal solutions and more Ganttchart decision-makings can be provided for the multi-objective FJSP combinatorial optimizations.

Elimination search for puzzle games: An application for Hashi solver

This paper proposes an efficient method to solve Hashi, a logical-type puzzle game with N by M grid. By using two methods, intersection method and elimination search, we can solve Hashi quickly and efficiency. The solver is authenticated by solving problems taken from Internet.