Abstract:
In the real-world of Logistics network, Supply Chain Management (SCM) & Automatic Guided Vehicle (AGV), there are many combinatorial optimization problems (COPs) imposing on more complex structure, nonlinear constraints, multiple objectives and uncertainty. The COPs make the problem intractable to the traditional approaches because of NP-hard ones. As one of the most typical scheduling problems, Flexible Jobshop Scheduling Problem (FJSP) is a generalization of the jobshop and parallel machine environment, which provides closer real manufacturing and logistics systems.
In order to develop an efficient algorithm whose reasonable computational time for NP-hard COPs, we have to consider 1) quality of solution, 2) computational time and 3) effectiveness of the nondominated solutions for multiobjective COP. As the most typical metaheuristics, Genetic Algorithm (GA) is a generic population-based metaheuristic such as Memetic Algorithm (MA), Particle Swarm Optimization (PSO), and Differential Evolution (DE). Metaheuristics is a very powerful and broadly applicable stochastic search and optimization technique which is effective for solving various NP hard problems. However, in order to expand traditional GA in the quality, computational time and effectiveness, we have to combine it with Machine Learning algorithm for creating hybrid metaheuristics.
The advanced seminar will concentrate as follows:
13:30-15:00, Jan. 9, 2023: Introducing Traditional Metaheuristics:
Hybrid GA, PSO, EDA and DE will introduce with applications. Hybrid evolutionary optimization with learning such as Teaching-Leaning based Optimization algorithms will summarize for scheduling & logistics.
13:30-15:00, Jan. 10, 2023: Multiobjective Hybrid GA and MoEA-HSS algorithms:
Basic MoHGA, MoHGA/TOPSIS (Technique for Order of Preference by Similarity to Ideal Solution), and MoEA-HSS (Hybrid Sampling Strategy) will introduce with applications of Reverse Logistics, FJSP-SDST (Sequence Dependent & Setup Time), and Logistics/SCM network design problems.
13:30-15:00, Jan. 12, 2023: Hybrid GA & PSO for Uncertain FJSP and MoCLN Design models:
Hybrid GA combined PSO will introduce for solving Fuzzy-FJSP (Flexible Jobshop Scheduling Problem). Hybrid co-evolution algorithm will also introduce by Fuzzy-FJSP with uncertain processing time. Multi- objective closed-loop network (MoCLN) design problem will introduce by NSGA-II.
13:30-15:00, Jan. 13, 2023: Hybrid Metaheuristics for MoSCM and Machine Learning for AGV:
Expanding GA and PSO with TLBO to Hybrid Metaheuristics will introduce for SCM network model. Reinforcement Learning for AGV (Automatic Guided Vehicle), MoGA for Multi-objective AGV, and MoGA+PSO+Q-Learning for MoFSP (Multi-objective Flexible Shop Scheduling) will introduce respectively.
Online Meeting ID: 388-4254-2507 (Tencent Voov Meeting)
Bio-Sketch: Google Scholar Citation: Mitsuo Gen.
Mitsuo Gen received his PhD in Engineering from Kogakuin University, PhD degree in Informatics from Kyoto University and is now Senior Research Scientist at FLSI and Visiting Prof. at TUS. He was faculties at Ashikaga Institute of Tech. for 1974-2003, at Waseda Univ. for 2003-2010. He was visiting faculties at Univ. of California at Berkeley for 1999.8-2000.3, Texas A&M Univ. for 2000.1-3 & 2000.8-9, Hanyang Univ. in S. Korea for 2010-2012 and National Tsing Hua Univ. in Taiwan for 2012-2014. His research field is Evolutionary Computation, Manufacturing Scheduling and Logistics Systems. He is a coauthor of the following books: G enetic Algorithms and Engineering D esign, 1997 and G enetic Algorithms and Engineering Optimization, 2000, John Wiley & Sons, New York; N etwork Models and Optimization: Multiobjective Genetic Algorithm; pproach, 2008 and I ntroduction to Evolutionary Algorithms, 2010, Springer, London.
