This paper presents NSGA-PINN, a multi-objective optimization framework for the effective training of physics-informed neural networks (PINNs). The proposed uses non-dominated sorting genetic algorithm (NSGA-II) to enable traditional stochastic gradient algorithms (e.g., ADAM) escape local minima effectively. Additionally, NSGA-II enables satisfying initial and boundary conditions encoded into ...

Although usage of genetic algorithms (GAs) has become widespread, the theoretical work from the genetic and evolutionary computation (GEC) field has been largely ignored by practitioners in realworld applications. This paper provides an overview of a three-step method for utilizing GEC theory to ensure robust search and avoid the common pitfalls in GA applications. Additionally, this study pres...

A kind of unrelated parallel machines scheduling problem is discussed. The memberships of fuzzy due dates denote the grades of satisfaction with respect to completion times with jobs. Objectives of scheduling are to maximize the minimum grade of satisfaction while makespan is minimized in the meantime. Two kind of genetic algorithms are employed to search optimal solution set of the problem. Bo...

In this paper, we propose a new genetic algorithm for multi-objective optimization problems. That is called “Neighborhood Cultivation Genetic Algorithm (NCGA)”. NCGA includes the mechanisms of other methods such as SPEA2 and NSGA-II. Moreover, NCGA has the mechanism of neighborhood crossover. Because of the neighborhood crossover, the effective search can be performed and good results can be de...

We apply the NSGA-II algorithm and its controlled elitist version NSGA-IIc for the intensity modulated beam radiotherapy dose optimization problem. We compare the performance of the algorithms with objectives for which deterministic optimization methods provide global optimal solutions. The number of parameters to be optimized can be up to a few thousands and the number of objectives varies fro...

Most multiobjective evolutionary algorithms are based on Pareto dominance for measuring the quality of solutions during their search, among them NSGA-II is well-known. A very few algorithms are based on decomposition and implicitly or explicitly try to optimize aggregations of the objectives. MOEA/D is a very recent such an algorithm. One of the major advantages of MOEA/D is that it is very eas...

An improved harmony search algorithm for constrained multi-objective optimization problems is proposed in this paper. Inspired by Particle Swarm Optimization, an inductor particle is introduced to speed up the convergence rate of the CMOHS. Two populations are adopted to increase the opportunity of finding the optimal solutions. Numerical experiments are divided into two parts: the first one co...