Scholar
Alexandre Allauzen
Google Scholar ID: B2-gXkkAAAAJ
Université Paris-Dauphine and ESPCI, PSL
Citations & Impact
Citations
1,646
H-index
18
i10-index
28
Publications
20
Co-authors
76
list available
Contact
Publications
Resume (English only)
Academic Achievements
- Paper 'Bridging the Theoretical Gap in Randomized Smoothing' accepted at AISTATS 2025
- Paper 'Highway Back-Propagation' accepted at ICLR 2025
- Paper 'SCOPE: A Self-supervised Framework for Improving Faithfulness in Conditional Text Generation' accepted at ICLR 2025
- Paper 'Chain and Causal Attention for Efficient Entity Tracking' (ChaCal) accepted at EMNLP 2024
- Paper 'Exploring Precision and Recall to assess the quality and diversity of LLMs' accepted at ACL 2024
- Paper 'LOCOST: State-Space Models for Long Document Abstractive Summarization' received Best Paper Award at EACL 2024
- Paper 'Lipschitz-Variance-Margin Tradeoff for Enhanced Randomized Smoothing' accepted at ICLR 2024
- Released LeBenchmark 2.0 with more data and additional tasks
- Paper 'Efficient Bound of Lipschitz Constant for Convolutional Layers by Gram Iteration' accepted at ICML 2023
- Paper 'Curriculum learning for data-driven modeling of dynamical systems' published in European Physical Journal
- Paper 'A Unified Algebraic Perspective on Lipschitz Neural Networks' accepted at ICLR 2023 (top 25%)
- Paper 'A Dynamical System Perspective for Lipschitz Neural Networks' accepted at ICML 2022
- LeBenchmark project paper accepted at NeurIPS 2021 (Datasets and Benchmarks Track)
Background
- Professor at ESPCI (École Supérieure de Physique et de Chimie Industrielles de la Ville de Paris)
- Affiliated with the MILES team at LAMSADE, Paris-Dauphine University and PSL (Paris Sciences & Lettres)
- Co-head of the MILES team (with Benjamin Negrevergne), focusing on Trustworthy and Frugal Machine Learning
- Head of the DATA program of PSL
- Main research interests include Natural Language Processing (NLP) and Deep Learning
