publications

2026

  1. Steering Pretrained Drafters during Speculative Decoding
    Frédéric Berdoz, Peer Rheinboldt, and Roger Wattenhofer
    In 40th Conference on Artificial Intelligence (AAAI), 2026
  2. Text-to-Scene with Large Reasoning Models
    Frédéric Berdoz, Luca Lanzendörfer, Nick Tuninga, and 1 more author
    In 40th Conference on Artificial Intelligence (AAAI), 2026
  3. Data Attribution in Large Language Models via Bidirectional Gradient Optimization
    Frédéric Berdoz, Luca Lanzendörfer, Kaan Bayraktar, and 1 more author
    In Third International AI Governance Workshop (AIGOV@AAAI), 2026

2025

  1. Recommender Systems for Democracy: Toward Adversarial Robustness in Voting Advice Applications
    Frédéric Berdoz, Dustin Brunner, Yann Vonlanthen, and 1 more author
    In 34th International Joint Conference on Artificial Intelligence (IJCAI), 2025

2024

  1. Can an AI Agent Safely Run a Government? Existence of Probably Approximately Aligned Policies
    Frédéric Berdoz, and Roger Wattenhofer
    In 38th Conference on Neural Information Processing Systems (NeurIPS), 2024

2023

  1. Fundamentals of Task-Agnostic Data Valuation
    Mohammad Mohammadi Amiri, Frédéric Berdoz, and Ramesh Raskar
    In 37th Conference on Artificial Intelligence (AAAI), 2023

2022

  1. Scalable Collaborative Learning via Representation Sharing
    Frédéric Berdoz, Abhishek Singh, Martin Jaggi, and 1 more author
    In Workshop on Decentralization and Trustworthy Machine Learning in Web3 (TSRML@NeurIPS), 2022