Scholar
Boris Knyazev
Google Scholar ID: Dp9VFB0AAAAJ
Research Scientist, Samsung - SAIT AI Lab
Citations & Impact
Citations
1,457
H-index
16
i10-index
16
Publications
20
Co-authors
26
list available
Publications
Resume (English only)
Academic Achievements
- Multiple papers accepted at top-tier venues including ICLR, NeurIPS, ICML, EMNLP, and TMLR
- 2025: "(Almost) Free Modality Stitching of Foundation Models" accepted to EMNLP 2025 (Main Conference)
- 2025: "Celo" and "Meta-learning Optimizers for Communication-Efficient Learning" accepted to TMLR
- 2025: Invited talk at ICLR 2025 Workshop on Weight Space Learning
- 2024: "Neural Graphs" accepted as oral at ICLR 2024
- 2023: Two workshop papers accepted at ICML 2023; "GHN-3" accepted at ICML 2023
- 2022: PhD thesis approved and published; "Model Zoos" accepted to NeurIPS 2022 Datasets and Benchmarks Track; "Hyper-Representations" accepted to NeurIPS 2022
- 2021: Two papers accepted at ICCV 2021; two papers accepted at NeurIPS 2021; named "Outstanding Reviewer" at ICCV 2021 (top 5% student reviewers)
- Regular reviewer for ICML, NeurIPS, ICLR, CVPR, and other top conferences
Background
- Research Scientist at Samsung SAIT AI Lab
- Adjunct Professor at the University of Montreal
- Research interests include graph neural networks (GNNs), large (language) models (LLMs), optimization, and meta-learning
- Applications span computer vision, language modeling, and molecule discovery
- Open to supervising graduate students on topics such as neural network weight representation, learning to optimize, LLMs and GNNs for scientific discovery, and compressing/merging large models
