mq.nguyen.2023(at)phdcs.smu(dot)edu.sg
I am Nguyen Minh Quang (Nguyen is family name), Ph.D. Student at School of Computing and Information Systems, Singapore Management University, under the supervision of Prof. Hady W. Lauw. I am also a member of Perferred.AI. My research focuses on Reinforcement Learning problem.
I received bachelor's degree (first class, honors program) in Information Technology from University of Engineering and Technology, Vietnam National University Hanoi (UET, VNU) in 2023. During my undergraduate, I was a member of DS&KT Laboratory, Faculty of Information Technology, and my initial research interest is Natural Language Processing (NLP) problems.
Reinforcement Learning
Self-critic Agent
Multi-Agent Reinforcement Learning
Nguyen Minh Quang and Hady W. Lauw
The Forty-first International Conference on Machine Learning (ICML 2024)
TL;DR: We find out that Bellman-based Learning Scheme can be "softly constrained" if we prompt learners with hypothesis - a weak environment representation, created by M augementor in the right figure.
Value-based reinforcement learning is the current State-Of-The-Art due to high sampling efficiency. However, our theoretical and empirical studies show evidence that it suffers from low exploitation in early training period and bias sensitiveness. To address these issues, we propose to augment the decision-making process with hypothesis, a weak form of environment description. Our approach relies on prompting the learning agent with accurate hypotheses, and designing a ready-to-adapt policy through incremental learning. We propose the ALH algorithm and monitor it on detailed analyses on a typical learning scheme and a diverse set of Mujoco benchmarks. Our algorithm produces a significant improvement over value-based learning algorithms and other strong baselines.
Minh-Quang Nguyen*, Duy-Cat Can* and Hoang-Quynh Le
European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML-PKDD 2024) (Research Track)
TL;DR: As its name, we propose to directly utilize pre-optimized BART (on Single-Doc Summarization) for Multi-Doc Summarization task. We show this is possible, and also "very simple", by the MCA arechitecture in the left figure.
We propose a novel vertical scaling approach. In this approach, we conditionally factorize the multi-document output probability by lower-complexity components. Specifically, these components are estimated by estimators optimized for single-doc data. Unlike the full-attention approach, vertical scaling has complexity that scales linearly with the number of single documents, making it more efficient for long documents or large numbers of documents. To further enhance the efficiency and effectiveness of our approach, we introduce the Multi-Channel Attention architecture. This architecture enables us to fully utilize BART's single-doc pre-optimized parameters, while does not require re-optimization, leading to a zero-cost transition. Our approach maintains promising accuracy and computing efficiency.
*co-first authors
Duy-Cat Can, Quoc-An Nguyen, Binh-Nguyen Nguyen, Minh-Quang Nguyen, Khanh-Vinh Nguyen, Trung-Hieu Do and Hoang-Quynh Le
Duy-Cat Can, Quoc-An Nguyen, Quoc-Hung Duong, Minh-Quang Nguyen, Huy-Son Nguyen, Linh Nguyen Tran Ngoc, Quang-Thuy Ha, Mai-Vu Tran
Presented at BioNLP@NAACL 2021
Hoang-Quynh Le, Quoc-An Nguyen, Quoc-Hung Duong, Minh-Quang Nguyen, Huy-Son Nguyen, Tam Doan Thanh, Hai-Yen Thi Vuong and Trang M. Nguyen
Presented at BioNLP@NAACL 2021