6 papers
Concept-Guided Fine-Tuning: Steering ViTs away from Spurious Correlations to Improve Robustness
Yehonatan Elisha, Oren Barkan, Noam Koenigstein
Vision Transformers (ViTs) often degrade under distribution shifts because they rely on spurious correlations, such as background cues, rather than semantically meaningful features…
Fidelity-Aware Recommendation Explanations via Stochastic Path Integration
Oren Barkan, Yahlly Schein, Yehonatan Elisha +3
Explanation fidelity, which measures how accurately an explanation reflects a model's true reasoning, remains critically underexplored in recommender systems. We introduce SPINRec…
Extracting Interaction-Aware Monosemantic Concepts in Recommender Systems
Dor Arviv, Yehonatan Elisha, Oren Barkan +1
We present a method for extracting \emph{monosemantic} neurons, defined as latent dimensions that align with coherent and interpretable concepts, from user and item embeddings in r…
Rethinking Saliency Maps: A Cognitive Human Aligned Taxonomy and Evaluation Framework for Explanations
Yehonatan Elisha, Seffi Cohen, Oren Barkan +1
Saliency maps are widely used for visual explanations in deep learning, but a fundamental lack of consensus persists regarding their intended purpose and alignment with diverse use…
Forget What You Know about LLMs Evaluations -- LLMs are Like a Chameleon
Nurit Cohen-Inger, Yehonatan Elisha, Bracha Shapira +2
Large language models (LLMs) often appear to excel on public benchmarks, but these high scores may mask an overreliance on dataset-specific surface cues rather than true language u…
BEE: Metric-Adapted Explanations via Baseline Exploration-Exploitation
Oren Barkan, Yehonatan Elisha, Jonathan Weill +1
Two prominent challenges in explainability research involve 1) the nuanced evaluation of explanations and 2) the modeling of missing information through baseline representations. T…