5 papers
Spatial Induction Heads: In-Context Learning of Multidimensional Cellular Automata
Kimia Kazemian, Menghan Xu, John Thickstun +1
Induction heads provide a mechanistic account of in-context learning in sequential data, but existing theory largely assumes that the context relevant to a prediction forms a conti…
Credit-assigned Policy Gradient for Early Stage Retrieval in Two-stage Ranking
Haruka Kiyohara, Mihaela Curmei, Ariel Evnine +7
Large-scale search, recommendation, and retrieval-augmented generation (RAG) systems typically employ a two-stage architecture: an early-stage ranker (ESR) generates a candidate se…
Recommender Systems as Control Systems
Giulia De Pasquale, Sarah Dean, Paolo Frasca
We propose a control-theoretic interpretation of recommender systems and use this perspective to analyze how fairness interventions shape long-term system behavior. Fairness concer…
Datasets for Navigating Sensitive Topics in Recommendation Systems
Amelia Kovacs, Jerry Chee, Kimia Kazemian +1
Personalized AI systems, from recommendation systems to chatbots, are a prevalent method for distributing content to users based on their learned preferences. However, there is gro…
Harm Mitigation in Recommender Systems under User Preference Dynamics
Jerry Chee, Shankar Kalyanaraman, Sindhu Kiranmai Ernala +3
We consider a recommender system that takes into account the interplay between recommendations, the evolution of user interests, and harmful content. We model the impact of recomme…