Scholar
Matthew Kowal
Google Scholar ID: FCg8QxUAAAAJ
FAR AI, York University, Vector Institute
Citations & Impact
Citations
358
H-index
9
i10-index
9
Publications
17
Co-authors
34
list available
Publications
Resume (English only)
Academic Achievements
- Preprint 2025: 'Into the Rabbit Hull: From Task-Relevant Concepts in DINO to Minkowski Geometry'
- Preprint 2025: 'It’s the Thought that Counts: Evaluating the Attempts of Frontier LLMs to Persuade on Harmful Topics' (introduced AttemptPersuadeEval)
- ICML 2025: 'Universal Sparse Autoencoders: Interpretable Cross-Model Concept Alignment'
- ICML 2025: 'Archetypal SAE: Adaptive and Stable Dictionary Learning for Concept Extraction in Large Vision Models'
- CVPR 2024 (Spotlight): 'Visual Concept Connectome (VCC): Open World Concept Discovery and their Interlayer Connections in Deep Models'
- CVPR 2024 (Spotlight): 'Understanding Video Transformers via Universal Concept Discovery'
- CVPR 2022 & XAI4CV Workshop (Spotlight): 'A Deeper Dive Into What Deep Spatiotemporal Networks Encode'
- ICCV 2021: 'Global Pooling, More than Meets the Eye: Position Information is Encoded Channel-Wise in CNNs'
- ICLR 2021: 'Shape or Texture: Understanding Discriminative Features in CNNs'
- BMVC 2020 (Oral): 'Feature Bin'
Research Experience
- Member of Technical Staff at FAR AI, working on AI safety, mechanistic interpretability, and LLM persuasion
- Intern at Ubisoft La Forge, working on generative modeling for character animations
- Intern at Toyota Research Institute, contributing to interpretability research for video transformers
- Faculty affiliate researcher at Vector Institute
- Lead Scientist in Residence at NextAI (2020–2022)
- Worked at Morrison Hershfield post-Bachelor’s, designing buildings, labs, and residential projects
