Scholar
Hubert Baniecki
Google Scholar ID: H72DRC0AAAAJ
PhD student, University of Warsaw
Citations & Impact
Citations
778
H-index
12
i10-index
13
Publications
20
Co-authors
11
list available
Publications
Resume (English only)
Academic Achievements
- Paper 'Explaining similarity in vision-language encoders with weighted Banzhaf interactions' accepted at NeurIPS 2025
- Paper 'Interpreting CLIP with hierarchical sparse autoencoders' accepted at ICML 2025
- Paper 'Efficient and accurate explanation estimation with distribution compression' accepted as a Spotlight at ICLR 2025 (top 5%)
- Paper 'Birds look like cars: Adversarial analysis of intrinsically interpretable deep learning' accepted by Machine Learning journal
- Co-authored paper 'Increasing phosphorus loss despite widespread concentration decline in US rivers' published in PNAS 2024
- Contributions to open-source software and benchmarks published in JMLR 2021 and NeurIPS 2024
- Awarded the START scholarship for young scientists by the Foundation for Polish Science (May 2025)
Background
- 4th (final) year PhD student in Computer Science at the University of Warsaw
- Research focuses on machine learning interpretability and explainable AI (XAI)
- Specific interests include interpreting vision–language models (e.g., CLIP), statistical foundations of explainable ML, open-source software and benchmarks, and applications in medicine
- Active reviewer for conferences such as NeurIPS, ECML, ICLR and journals including JMLR, Machine Learning, and Nature Communications
- On the academic job market for positions starting in 2026/2027
