Research
Nine lines of research, one lab.
Learning from constraints
Knowledge as constraints that a learner must satisfy.
Learning in visual environments
Agents that learn from a continuous video stream.
Logic and learning
Deep networks that reason with symbolic rules.
Graph neural networks
Learning on structured, relational data.
Language
Language models, dialogue and education.
Bioinformatics
Proteins, molecules and gene expression.
Deep learning foundations
How and why neural networks learn.
Collectionless AI
Learning over time, without stored datasets.
Crosswords
WebCrow and solving puzzles with AI.