معرفی
Shiwei Liu is a Newton International Fellow at the University of Oxford, specializing in sparsity within neural networks. Previously, he earned his Ph.D. (cum laude) from Eindhoven University of Technology in 2022, with his thesis receiving the Informatics Europe Best Dissertation Runner-up Award in 2023. His research impacts efficient training/inference/transfer of large models, robustness, generative AI, and graph learning.
- Education: Ph.D. (cum laude) from Eindhoven University of Technology (2022)
Research Focus: Shiwei's work explores sparsity in neural networks, particularly its applications to large language models (LLMs) and 3D medical image segmentation. His recent publications emphasize dynamic data pruning, neuron revitalization, and outlier-based sparsity optimization.
Key Publications: His 2024 contributions include techniques for sparse feature fusion in medical imaging and layerwise sparsity in LLMs, alongside studies on pruning-induced performance degradation.
Awards:
- Best Dissertation Runner-up, Informatics Europe (2023)
- Newton International Fellowship (2023)
- CPAL Rising Star (2023)
- Best Paper Award of LoG 2022
- PhD Cum Laude/Best Thesis Award (2022)
Collaborations: He collaborates internationally, with recent projects involving institutions in the Netherlands (Eindhoven University of Technology) and Greece (Interspeech 2024). His work spans 2016–2024, with 31 research outputs and 9 peer-reviewed articles in 2024 alone.
