Good Papers

Showing papers from Centrale Supélec Show all papers

80%Must read
?Must readVote to see the score
NeurIPS 2026Paris Dauphine - PSLCentrale SupélecINRIA - Ecole Normale SuperieureEPFLUniversité Paris DauphineAdversarial robustness

Frame the adversary: a structure-aware attack methodology

A structure-aware optimization framework crafts frequency adversarial attacks via weighted projections onto structured non-orthogonal transform constraints, yielding effective cross-architecture vulnerabilities.

Vicky Kouni, Stelios Perrakis, Francis Bach, Pascal Frossard and 1 more

Paris Poster Session 4, Thu, Dec 10, 5:30 PM–7:30 PM, Paris Poster Hall · Published 2026

– ReadersNo votes yet
12/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

AI panel: 12 of 20 reviewers recommend it
lenient 5/5
medium 7/10
strict 0/5
72%Highly rated
?Highly ratedVote to see the score
NeurIPS 2026U FribourgJohns HopkinsTechnical University of MunichU FerraraCentrale SupélecGraph neural networks

On the Rademacher Complexity of Graph Neural Networks: Unifying Expressivity and Geometry

Rademacher complexity bounds unify GNN expressivity and input geometry through equivalence classes, covering numbers, and Wasserstein robustness.

Martin Carrasco, Caio Deberaldini Netto, Ehimare Okoyomon, Aneeqa Mehrab and 2 more

Paris Poster Session 2, Wed, Dec 9, 5:00 PM–7:00 PM, Paris Poster Hall · Published 2026

– ReadersNo votes yet
8/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

AI panel: 8 of 20 reviewers recommend it
lenient 2/5
medium 5/10
strict 1/5
83%Must read
?Must readVote to see the score
NeurIPS 2026PleiasPleias & KU LeuvenEPFL - EPF LausanneTechnical University of Applied École Normale Supérieure PSLLLM pretraining & scaling laws

It's All Training: A Fully Synthetic Single-Stage Recipe for LLMs

SYNTH is an open-source synthetic dataset derived from Wikipedia that collapses pre-, mid-, and post-training into one stage, training competitive small models with 10-140x fewer tokens and higher factual precision than web-crawled data.

Pierre-Carl Langlais, Pieter Delobelle, Yannick Detrois, Pavel Chizhov and 6 more

Paris Poster Session 5, Fri, Dec 11, 11:30 AM–1:30 PM, Paris Poster Hall · Published 2026

– ReadersNo votes yet
13/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

AI panel: 13 of 20 reviewers recommend it
lenient 5/5
medium 7/10
strict 1/5
80%Must read
?Must readVote to see the score
NeurIPS 2026Université Paris-Saclay/CentraleCentrale SupélecOnera - The french aerospace labCentraleSupélecDistribution shift & OOD detection

Backbone-Equated Diffusion OOD via Sparse Internal Snapshots

A mutualized backbone-equated protocol reveals that sparse internal diffusion snapshots at low noise detect OOD better than full denoising with minimal heads, via encoder-decoder complementarity and diagonal-score separation.

Yadang Alexis Rouzoumka, Jean Pinsolle, Eugénie TERREAUX, Christèle Morisseau and 2 more

Paris Poster Session 6, Fri, Dec 11, 2:30 PM–4:30 PM, Paris Poster Hall · Published 2026

– ReadersNo votes yet
12/20 AI panelreviewers recommend it

Readers and the AI panel: vote on this paper to see what they said.

Only vote on papers you've read. Sign in with GitHub to vote.

AI panel: 12 of 20 reviewers recommend it
lenient 3/5
medium 8/10
strict 1/5