Scholar

Enric Boix-Adserà

Google Scholar ID: KOZSrE8AAAAJ

Assistant Professor at Wharton Statistics and Data Science

Citations & Impact

All-time

Citations

1,080

H-index

i10-index

Publications

Co-authors

list available

Contact

Publications

20 items

Browse publications on Google Scholar (top-right) ↗

Resume (English only)

Academic Achievements

Published multiple papers, including:
- FACT: the Features At Convergence Theorem for neural networks
- Toward universal steering and monitoring of AI models
- Let Me Think! A long chain of thought can be worth exponentially many short ones
- The power of fine-grained experts: Granularity boosts expressivity in Mixture of Experts
- On the inductive bias of infinite-depth ResNets and the bottleneck rank
- Prompts have evil twins
- When can transformers reason with abstract symbols?
- Towards a theory of model distillation
- Transformers learn through gradual rank increase
- Tight conditions for when the NTK approximation is valid

Research Experience

Assistant Professor of Statistics and Data Science at Wharton School, University of Pennsylvania.

Background

Research interests include the theory of AI (including distillation, reasoning models, steering, monitoring, training dynamics, inductive bias of architectures, adversarial examples, feature learning, and AI safety).

Co-authors

7 total