publications

2026

1. ICLR

   Optimization-Free Image Immunization Against Diffusion-Based Editing

   Tarik Can Ozden, Ozgur Kara, Oguzhan Akcin, and 4 more authors

   ICLR, 2026

   PDF

2025

1. ACL

   INTERACT: Enabling Interactive, Question-Driven Learning in Large Language Models

   Aum Kendapadi*, Kerem Zaman*, Rakesh R Menon*, and 1 more author

   ACL, 2025

   PDF
2. EMNLP

   A Causal Lens for Evaluating Faithfulness Metrics

   Kerem Zaman, and Shashank Srivastava

   EMNLP, 2025

   PDF Code
3. arxiv

   Is Chain-of-Thought Really Not Explainability? Chain-of-Thought Can Be Faithful without Hint Verbalization

   Kerem Zaman, and Shashank Srivastava

   ArXiv, 2025

   PDF

2024

1. EMNLP

   Fuse to Forget: Bias Reduction and Selective Memorization through Model Fusion

   Kerem Zaman, Leshem Choshen, and Shashank Srivastava

   2024

   PDF Code

2023

1. EMNLP

   MaNtLE: Model-agnostic Natural Language Explainer

   Rakesh R Menon, Kerem Zaman, and Shashank Srivastava

   Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, Dec 2023, 2023

   PDF

2022

1. CVPR

   Rank in Style: A Ranking-based Approach to Find Interpretable Directions

   Umut Kocasari*, Kerem Zaman*, Mert Tiftikci*, and 2 more authors

   2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2022

   PDF Code Website
2. EMNLP

   A Multilingual Perspective Towards the Evaluation of Attribution Methods in Natural Language Inference

   Kerem Zaman, and Yonatan Belinkov

   Dec 2022

   Abs PDF Code Website

   Most evaluations of attribution methods focus on the English language. In this work, we present a multilingual approach for evaluating attribution methods for the Natural Language Inference (NLI) task in terms of faithfulness and plausibility. First, we introduce a novel cross-lingual strategy to measure faithfulness based on word alignments, which eliminates the drawbacks of erasure-based evaluations. We then perform a comprehensive evaluation of attribution methods, considering different output mechanisms and aggregation methods. Finally, we augment the XNLI dataset with highlight-based explanations, providing a multilingual NLI dataset with highlights, to support future exNLP studies. Our results show that attribution methods performing best for plausibility and faithfulness are different.

2020

1. SIU

   BURST: Software and Simulation Solutions of an Autonomous Vehicle

   Fatih Kose, Ibrahim Ozcan, Melih Dal, and 5 more authors

   2020 28th Signal Processing and Communications Applications Conference (SIU), Dec 2020