Yonatan Belinkov
Scholar

Yonatan Belinkov

Google Scholar ID: K-6ujU4AAAAJ
Technion
Natural Language ProcessingModel InterpretabilityArtificial Intelligence
Citations & Impact
All-time
Citations
17,711
 
H-index
51
 
i10-index
103
 
Publications
20
 
Co-authors
28
list available
Resume (English only)
Academic Achievements
  • - ICLR 2025: Arithmetic Without Algorithms: Language Models Solve Math With a Bag of Heuristics
  • - ACL 2024: Diffusion Lens: Interpreting Text Encoders in Text-to-Image Pipelines
  • - ACL 2023: BLIND: Bias Removal With No Demographics
  • - NeurIPS 2022: Locating and Editing Factual Associations in GPT
  • - NeurIPS 2020: Investigating Gender Bias in Language Models Using Causal Mediation Analysis
  • - ACL 2019: Don't Take the Premise for Granted: Mitigating Artifacts in Natural Language Inference
  • - ICLR 2018: Synthetic and Natural Noise Both Break Neural Machine Translation
  • - NIPS 2017: Analyzing Hidden Representations in End-to-End Automatic Speech Recognition Systems
  • - ACL 2017: What do Neural Machine Translation Models Learn about Morphology?
Research Experience
  • - Faculty member at Technion Taub Faculty of Computer Science
  • - Postdoc at Harvard SEAS, working with Stuart Shieber, affiliated with the Mind, Brain, Behavior initiative
  • - Worked with the NLP group and CCNLab at Harvard
  • - Worked with the Spoken Language Systems group at MIT CSAIL
  • - Software engineer at IntuView
Education
  • - PhD: Massachusetts Institute of Technology (MIT) CSAIL, Advisor: James Glass, Thesis: Analyzed internal language representations in deep learning models, with particular applications to machine translation and speech recognition
  • - Master's: Tel Aviv University, Arabic Studies, Thesis: The Arabic dialect of Jisir izZarga
Background
  • - Research interests: Artificial Intelligence and Machine Learning, especially Large Language Models and other Natural Language Processing models
  • - Main research areas: Interpretability, Robustness, safety, and controllability, Emergent and multi-agent communication, Biological language models, NLP for Hebrew and Arabic
Miscellany
  • Information on personal interests and hobbies is not provided