Planning and scheduling with reinforcement learning
Watch the third Sustainable Transformation Seminar of this spring season 2021: Planning and scheduling with reinforcement learning.
Time: Fri 2021-05-07 12.00 - 13.00
Participating: Jong Hun Woo
About this seminar
The industrial paradigm is changing with artificial intelligence methodologies. In particular, The deep neural network theory, proven through Dr.Hinton's 'Deep belief network', overcomes the limitations of past artificial intelligence research, and numerous studies triggered by it, such as AlphaGo and AlphaStar, are yielding practical results. Reinforcement learning is also an artificial intelligence algorithm that is re-emerging with the advent of deep neural networks although it has been eliminated for a while due to the curse of dimensionality. Recently, reinforcement learning is being actively studied by replacing tabular calculation with multi-layer perceptron. The study on planning and scheduling to be introduced in this seminar is also an artificial intelligence study using multi-layer perceptron-based reinforcement learning. Past planning and scheduling studies have solved problems using optimization techniques such as MIP (mixed integer programming), metaheuristic, and CST (constraint satisfaction technique). These optimization techniques have been well applied to problems that have a standardized and definite problem environment but have limited application to varying environments. Also, as the size of the planning and scheduling problem increases, the computation time increases exponentially, so there are many problems that cannot be solved realistically. In this seminar, I would like to introduce examples of applying reinforcement learning to planning and scheduling problems in a varying environment and to present opinions on the future development potential of reinforcement learning.
About the speaker
Jong Hun Woo is an associate professor in the Department of Naval Architecture and Ocean Engineering at Seoul National University in South Korea. He holds a PhD in Naval Architecture and Ocean Engineering from Seoul National University. His research interest is in DES simulation, APS (Advanced Planning and Scheduling), machine learning and queuing theory. He has relevant research experiences in the application areas of shipbuilding. His email address is firstname.lastname@example.org .