Emmanuel Abbé
Full Professor
Studies the mathematical foundations of learning and reasoning in neural networks and large language models, including generalization on unseen data and the learnability of compositional tasks.
Emmanuel Abbé's lab at EPFL works on learning theory & optimization, signal processing & information theory and large language models. 36 papers since 2023, 13 at top venues, mostly NeurIPS and ICLR. 5 PhD students. 3 alumni, most often to Apple. Applications close .
At a glance
- papers, 2023–2026
- 9/ year
- at T1 venues
- 36%
- PhD students
- 5
- MS students
- 0
- alumni listed
- 3
Papers by yearfrom the lab homepage, 2023–2026
- 4 papers2026 · T1 2 · collecting
- 6 papers2025 · T1 2
- 13 papers2024 · T1 5
- 13 papers2023 · T1 4
T1 venuesOtherVenues: NeurIPS, ICLR, ICML, ISIT, COLT
Research focuspaper titles that use each topic's words
- Learning theory & optimization14 papers
- Large language models3 papers
- Signal processing & information theory0 papers
- compositional0 papers
- mathematical foundations0 papers
Paper list
From the lab homepage.
20264 papers
- ICLR (T1)AbstRaL: Augmenting LLMs’ Reasoning by Reinforcing Abstract Thinking
- Revue medicale suisseComment l’intelligence artificielle peut-elle améliorer la prévention cardiovasculaire ? (opens in a new tab)
- Medical Image Analysis (T1)Future cardiovascular events prediction from invasive coronary angiography: A graph representation learning perspective (opens in a new tab)
- ISITTensor Reed-Muller Codes: Achieving Capacity with Quasilinear Decoding Time (opens in a new tab)
20256 papers
- The International Journal of Cardiovascular ImagingAI-Driven multi-view learning from CCTA for myocardial infarction diagnosis (opens in a new tab)
- Second Conference on Language Modeling COLMAlgorithm Discovery With LLMs: Evolutionary Search Meets Reinforcement Learning
- EMBCCM-UNet: A Self-Supervised Learning-Based Model for Coronary Artery Segmentation in X-Ray Angiography (opens in a new tab)
- NeurIPS (T1)Inductive Domain Transfer In Misspecified Simulation-Based Inference (opens in a new tab)
- ICLR (T1)Learning High-Degree Parities: The Crucial Role of the Initialization
- ISITReed-Muller Codes for Quantum Pauli and Multiple Access Channels (opens in a new tab)
202413 papers
- EUROINTERVENTIONA study of ChatGPT in facilitating Heart Team decisions on severe aortic stenosis (opens in a new tab)
- IEEE Open Journal of Engineering in Medicine and BiologyAnatomy-Informed Multimodal Learning for Myocardial Infarction Prediction (opens in a new tab)
- International Journal of CardiologyAngioPy segmentation: An open-source, user-guided deep learning tool for coronary artery segmentation (opens in a new tab)
- European Heart JournalangioPy: an open-source, deep learning-driven tool for coronary artery segmentation (opens in a new tab)
- EUROPEAN HEART JOURNALAssessing the need for coronary angiography in high-risk acute coronary syndrome patients using artificial intelligence and computed tomography (opens in a new tab)
- International Journal of Cardiovascular ImagingAssessing the need for coronary angiography in high-risk non-ST-elevation acute coronary syndrome patients using artificial intelligence and computed tomography (opens in a new tab)
- Journal of medical Internet researchGender Bias in AI’s Perception of Cardiovascular Risk (opens in a new tab)
- ICML (T1)Generalization on the Unseen, Logic Reasoning and Degree Curriculum
- ISBIGraph Neural Network based Future Clinical Events Prediction from Invasive Coronary Angiography (opens in a new tab)
- NeurIPS (T1)How Far Can Transformers Reason? the Globality Barrier and Inductive Scratchpad (opens in a new tab)
- ICML (T1)On the Minimal Degree Bias in Generalization on the Unseen for non-Boolean Functions
- NeurIPS (T1)Transformation-invariant Learning and Theoretical Guarantees for Ood Generalization
202313 papers
- FOCS (T1)A proof that Reed-Muller codes achieve Shannon capacity on symmetric channels (opens in a new tab)
- AIMECan Knowledge Transfer Techniques Compensate for the Limited Myocardial Infarction Data by Leveraging Haemodynamics? An in silico Study (opens in a new tab)
- ESCChatGPT takes on the european exam in core cardiology: an AI success story
- European Heart Journal - Digital HealthChatGPT takes on the European Exam in Core Cardiology: an artificial intelligence success story? (opens in a new tab)
- Open HeartDeep learning-based prediction of future myocardial infarction using invasive coronary angiography: a feasibility study (opens in a new tab)
- Annals Of StatisticsLearning Sparse Graphons And The Generalized Kesten-Stigum Threshold (opens in a new tab)
- Revue medicale suisseLes dispositifs intelligents et l’IA en cardiologie peuvent-ils améliorer la pratique clinique ? (opens in a new tab)
- Communications on Pure and Applied MathematicsPolynomial-time universality and limitations of deep learning (opens in a new tab)
- NeurIPS (T1)Provable Advantage of Curriculum Learning on Parity Targets with Mixed Inputs
- Foundations and Trends in Communications and Information TheoryReed-Muller Codes (opens in a new tab)
- COLT (T1)SGD learning on neural networks: leap complexity and saddle-to-saddle dynamics
- EurointerventionTowards AI-assisted cardiology: a reflection on the performance and limitations of using large language models in clinical decision-making (opens in a new tab)
Where alumni went3 listed, 50% to industry
Industry 1Academia 1
Most often to
- Apple
Also to
- Bocconi University
Apply to EPFLFall 2027
EDIC Computer and Communication Sciences PhD
- Round
- Round 1
- Opens
- —
- GRE
- optional
- English
- TOEFL 100 / IELTS 7.0
- Fee
- none
Confirmed on the program page.
Similar labsnearest by research summary
- CAToronto
Nisarg Shah
- Learning theory & optimization
- SGNTU
Atsushi Nitanda
- Learning theory & optimization
- GBCambridge
Simon Godsill
- Probabilistic & causal inference
- HKCUHK
Hanrui Zhang
- Reasoning & neuro-symbolic AI
- USNYU
Emily Black
- Trustworthy & safe ML
- USStanford
Carlos Guestrin
- Trustworthy & safe ML
Your notes
Save this lab to keep notes.
Data
Checked against the lab homepage.
- Collected
- Verified
- Papers
- The lab homepage
- Members
- Lab homepage
- Alumni
- Lab homepage
Tiers are venue classifications, not lab rankings.
EPFLcloses