Keito Kozaki

Keito Kozaki

First-year master's student working on recommender systems, with an interest in interpreting model internals and datasets.

Residual Dominance as a Structural Account of Last-Item Reliance in Causal Self-Attention Recommenders

Published in ACM Conference on Recommender Systems (RecSys), 2026

Keito Kozaki, Keigo Sakurai, Ren Togo, Takahiro Ogawa, Miki Haseyama

pdf

Recommended citation: Keito Kozaki, Keigo Sakurai, Ren Togo, Takahiro Ogawa, Miki Haseyama. (2026). "Residual Dominance as a Structural Account of Last-Item Reliance in Causal Self-Attention Recommenders." Proceedings of the ACM Conference on Recommender Systems (RecSys).