Liya Gurevitch

Tel Aviv University
Learning Counterfactual Explanations for Recommender Systems
Liya Gurevitch

Bio

Liya is a second-year Master’s student in the Industrial Engineering Department at Tel Aviv University, after graduating from her BA studies Summa Cum Laude and first in her class. She is researching the fields of Explainable Artificial Intelligence (XAI) and Deep Learning. Her thesis work focuses on the critical area of enhancing transparency and understandability in recommender systems through XAI principles. An article she authored based on this research is set to be published at the prestigious World Wide Web Conference in Singapore later this year. In tandem with her research endeavors, she serves as a teaching assistant for the “Machine Learning” and “Theory of Learning” courses at the university. She also practices Yoga and Meditation and is a long-distance swimmer.

Abstract

This study introduces the Learning to eXplain Recommendations (LXR) framework, aimed at generating counterfactual explanations in recommender systems. The central problem addressed is the lack of an effective post-hoc, model-agnostic framework tailored for generating such explanations. Prior research has not adequately addressed this need. To bridge this gap, we developed LXR, incorporating a novel self-supervised counterfactual loss term to prioritize the significance of user data in item recommendations. Our approach involved proposing several counterfactual evaluation metrics and implementing the LXR framework across various recommendation algorithms and datasets. The research demonstrates LXR’s capability in providing transparent and insightful counterfactual explanations, thereby enhancing user understanding and trust in recommendation outcomes.

Agenda

8:45 Reception
9:30 Opening remarks by WiDS TLV ambassadors Noah Eyal Altman, Or Basson, and Nitzan Gado
9:45 Dr. Aya Soffer, IBM: "Putting Generative AI to Work: What Have We Learned So Far?"
10:15 Prof. Reut Tsarfaty, Bar-llan University: "Will Hebrew Speakers Be Able to Use Generative AI in Their Native Tongue?"
10:45 Poster Pitches
10:55 Break
11:10 Lightning talks
12:30 Lunch & poster session
13:30 Roundtable session & poster session
14:15 Roundtable closing
14:30 Break
14:40 Naomi Ken Korem, Lightricks: "Mastering the Art of Generative Models: Training and Controlling Text-to-Video Models"
15:00 Dr. Yael Mathov, Intuit: "Surviving the AI-pocalypse: Your Guide to LLM Security"
15:20 Closing remarks
15:30 The end