Scholar
Yong Liu
Google Scholar ID: 2ejuK8UAAAAJ
National University of Singapore
Citations & Impact
Citations
1,087
H-index
11
i10-index
11
Publications
20
Co-authors
0
Publications
Resume
Academic Achievements
- NeurIPS 2025: 'Sparse MeZO: Less Parameters for Better Performance in Zeroth-Order LLM Fine-Tuning'
- ICML 2025: Two papers including 'SeedLoRA: A Fusion Approach to Efficient LLM Fine-Tuning' and 'MERIT: Maximum-normalized Element-wise Ratio for Language Model Large-batch Training'
- WWW 2024: One paper accepted
- NeurIPS 2022: 'Random Sharpness-Aware Minimization'
- CVPR 2022: 'Towards Efficient and Scalable Sharpness-Aware Minimization'
- ICLR 2022: 'Concurrent Adversarial Learning for Large-Batch Training'
- AAAI 2022: Contributed to 'Go Wider Instead of Deeper'
- ICASSP 2021: Published work on a quantitative metric for privacy leakage in federated learning
- IJCAI 2019: Proposed value function transfer for deep multi-agent reinforcement learning based on N-step returns
- AAAI 2020: Introduced a novel game abstraction method using graph attention neural networks
Background
- Research interests include Large-Batch Training, Multi-Agent Systems, Reinforcement Learning, and Transfer Learning
- Focuses on Large-Batch Training on large-scale distributed systems to accelerate deep neural network training
- Works on simplifying the learning process in multi-agent systems, e.g., through game abstraction
- Studies algorithmic frameworks of reinforcement learning and their applications in multi-agent settings
- Explores transfer learning in multi-agent systems, especially across environments with different numbers of agents