Scholar
Mingbao Lin
Google Scholar ID: Dp3L1bsAAAAJ
Principal Research Scientist, Rakuten
Model Compression
(Multimodal) LLMs
Diffusion Models
Follow
Homepage
↗
Google Scholar
↗
Citations & Impact
All-time
Citations
4,366
H-index
34
i10-index
54
Publications
20
Co-authors
4
list available
Contact
Email
linmb001@outlook.com
GitHub
Open ↗
LinkedIn
Open ↗
Publications
28 items
Unbiased Dynamic Pruning for Efficient Group-Based Policy Optimization
2026
Cited
0
Reversible Diffusion Decoding for Diffusion Language Models
2026
Cited
0
Autoregressive Image Generation Needs Only a Few Lines of Cached Tokens
2025
Cited
0
Parallel Vision Token Scheduling for Fast and Accurate Multimodal LMMs Inference
2025
Cited
0
Test-Time Temporal Sampling for Efficient MLLM Video Understanding
2025
Cited
0
Progressive Supernet Training for Efficient Visual Autoregressive Modeling
2025
Cited
0
Beyond Benchmarks: Understanding Mixture-of-Experts Models through Internal Mechanisms
2025
Cited
0
ERTACache: Error Rectification and Timesteps Adjustment for Efficient Diffusion
2025
Cited
0
Load more
Resume (English only)
Background
Currently a Principal Research Scientist at Rakuten Singapore.
Research focuses on model compression and acceleration for deep neural networks, with nearly a decade of experience.
Work spans convolutional neural networks, vision transformers, (multimodal) large language models, and diffusion models.
Strong emphasis on bridging academia and industry by transforming research into real-world products.
Actively collaborates with university professors and students to co-author papers and advance technological frontiers.
Co-authors
4 total
Rongrong Ji 纪荣嵘
Professor, Xiamen University
Zhihang Lin
Xiamen University & Shanghai Innovation Institute
Co-author 3
Chia-Wen Lin (林嘉文)
Distinguished Professor of Electrical Engineering, National Tsing Hua University, Taiwan
×
Welcome back
Sign in to Agora
Welcome back! Please sign in to continue.
Email address
Password
Forgot password?
Continue
Do not have an account?
Sign up