Big Decision Models

Thursday, December 15, 2022

Tianjin, China

Introduction

Large sequence models, which feature a great volume of parameters and auto-regressive data processing, have recently provided significant assistance for prediction tasks and (self-)supervised learning in natural language processing (NLP) and computer vision (CV). In the last two years, sequence models, especially the Transformer, have attracted rapidly increasing interest in the RL communities and spawned many approaches. Besides, large sequence models are emerging in decision-making and reinforcement learning (RL) with noticeable effectiveness and generalizability, which implies the potential of building a Big Decision Model (BDM) for general purposes, that is, a large sequence model leveraging a large number of parameters for hundreds or more decision-making tasks like what large sequence models did for NLP and CV. Our workshop is a half-day workshop on BDMs at DAI 2022, with the aim to provide a venue, which can bring together academic researchers and industry practitioners (i) to discuss the potentials, principles, and applications of BDMs and (ii) to foster research on innovative algorithms, novel techniques, and new frameworks for BDMs.

Schedule

Overall Program (Dec. 15, 2022, UTC+8)

14:00-14:10

Openning

Ying Wen

Shanghai Jiao Tong University

14:10-14:40

Large Scale Reinforcement Learning in Industrial Problems: Challenges and Opportunities

Jiangcheng Zhu

Huawei Cloud

14:40-15:10

A General Feedback System Encoder and its Applications

Fanming Luo

Polixir

15:10-15:30

Coffee Break

15:30-16:00

Offline Pre-trained Multi-Agent Reinforcement Learning

Linghui Meng

Institute of Automation, Chinese Academy of Sciences

16:00-16:30

Pre-training in Deep Reinforcement Learning: A Survey

Zhihui Xie

Shanghai Jiao Tong University & Tencent

16:30-17:00

DB1: A Practice of Intelligent Decision-Making Foundation Model

Ziyu Wan

Shanghai Jiao Tong University & Digital Brain Lab

17:00-17:10

Closing

Weinan Zhang

Shanghai Jiao Tong University

Talks

Large Scale Reinforcement Learning in Industrial Problems: Challenges and Opportunities

Time: 14:10-14:40

Speaker: Jiangcheng Zhu, Huawei Cloud

Bio: Jiangcheng Zhu received his Ph.D. and B.Sc. degrees in automation from Zhejiang University, Hangzhou, China, in 2019 and 2012, resp., the M.S. degree in the University of California at San Diego, San Diego, USA, in 2014. He is currently a senior AI scientist in Huawei Cloud. His current research interests include reinforcement learning and large-scale AI system.

A General Feedback System Encoder and its Applications

Time: 14:40-15:10

Speaker: Fanming Luo, Polixir

Bio: Fanming Luo received his B.Sc. degree from Nanjing University, Nanjing, China, in 2019. Currently, he is working towards the Ph.D. degree in the National Key Lab for Novel Software Technology, the School of Artificial Intelligence, Nanjing University, China. His research interests include meta reinforcement learning and imitation learning. His work has been accepted in the conferences of artificial intelligence, including NeurIPS and AAAI. The work shared in DAI2022 is conducted during his internship in Polixir.

Offline Pre-trained Multi-Agent Reinforcement Learning

Time: 15:30-16:00

Speaker: Linghui Meng, Institute of Automation, Chinese Academy of Sciences

Bio: Linghui Meng received his B.Sc. degree from Beijing Jiao Tong University, China in 2019. He is a Ph.D. student at both the Institute of Automation Chinese Academy of Sciences and the University of Chinese Academy of Sciences. His current interests include theoretical research on reinforcement learning, pre-training models, multi-agent system, and speech recognition.

Pre-training in Deep Reinforcement Learning: A Survey

Time: 16:00-16:30

Speaker: Zhihui Xie, Shanghai Jiao Tong University & Tencent

Bio: Zhihui Xie received his bachelor's degree from Shanghai Jiao Tong University. He is now a graduate student in the School of Computer Science and Engineering, Shanghai Jiao Tong University. His research interests include reinforcement learning and its applications. He has published first-authored papers in top AI conferences like SIGIR and EMNLP.

DB1: A Practice of Intelligent Decision-Making Foundation Model

Time: 16:30-17:00

Speaker: Ziyu Wan, Shanghai Jiao Tong University & Digital Brain Lab

Bio: Ziyu Wan is now a second-year Ph.D. candidate at Shanghai Jiao Tong University. His research interest is (multi-agent) reinforcement learning and its applications, foundation models.

Organizers

Weinan Zhang

Shanghai Jiao Tong University

Muning Wen

Shanghai Jiao Tong University

Address

Conrad Tianjin:

46 Tianta Road, Nankai District, Tianjin, 300381, P.R.China.

Sponsor

Page updated

Google Sites

Report abuse