Modern Paradigms in Generalization
summary
Program sets out to produce a taxonomy of settings and to adapt old proofs while building new tools for non IID generalization, with concrete activities: four workshops (boot camp, theory–practice gap, distribution shift, future settings) plus weekly reading groups and industry mini workshops. Examples motivated are LLMs broken by adversarial online communities and self driving cars encountering new homes, and a goal to refocus ML theory training so PhD students get tractable questions and actionable results.
AI-written summary · about
chapters (5)
- 0:00Program overview
- 1:36Defining generalization
- 4:31Program goals and focus
- 6:42Activities and workshops
- 8:10Community impact
related talks
01h 10mModern paradigms of generalization, the heliocentric model of Aristarchus,...
53mGeneralization insights from actual cognition
01h 16mGeneralization in natural language processing
36mThe Limits of Post-Selection Generalization
01h 02mOut-of-Distribution Generalization as Reasoning: Are LLMs Competitive?
01h 05mAdaptive Generalization: The Role of Dynamic Evaluation in Low-Resource Settings
42mQuestions for Theory in the New Age of Machine Learning
01h 10mModels and Programs: Better Together