Publications by Clément Hongler

Smart Proofs via Recursive Information Gathering: Decentralized Refereeing by Smart Contracts

We introduce the SPRIG (Smart Proofs via Recursive Information Gathering) protocol. SPRIG allows agents to propose, question, and defend mathematical proofs in a decentralized fashion. A structure of stakes and bounties aims at producing debates in good fa ...

Association for Computing Machinery (ACM)2024

Lattice Models and Conformal Field Theory

Clément Hongler, Francesco Spadaro, Franck Raymond Gabriel

This book introduces the mathematical ideas connecting Statistical Mechanics and Conformal Field Theory (CFT). Building advanced structures on top of more elementary ones, the authors map out a well-posed road from simple lattice models to CFTs. Structure ...

Courant Institute of Mathematical Sciences2024

Arrows of Time for Large Language Models

Clément Hongler, Vassilis Joseph Papadopoulos

We study the probabilistic modeling performed by Autoregressive Large Language Models (LLMs) through the angle of time directionality, addressing a question first raised in (Shannon, 1951). For large enough models, we empirically find a time asymmetry in t ...

ACM2024

MAGNETIZATION IN THE ZIG-ZAG LAYERED ISING MODEL AND ORTHOGONAL POLYNOMIALS

Clément Hongler

— We discuss the magnetization Mm in the m-th column of the zig-zag layered 2D Ising model on a half-plane using Kadanoff–Ceva fermions and orthogonal polynomials techniques. Our main result gives an explicit representation of Mm via m × m Hankel determina ...

2024

Looking for Complexity at Phase Boundaries in Continuous Cellular Automata

Clément Hongler, Vassilis Joseph Papadopoulos

One key challenge in Artificial Life is designing systems that display an emergence of complex behaviors. Many such systems depend on a high-dimensional parameter space, only a small subset of which displays interesting dynamics. Focusing on the case of co ...

Association for Computing Machinery, Inc2024

Conformal Field Theory at the Lattice Level: Discrete Complex Analysis and Virasoro Structure

Clément Hongler

Critical statistical mechanics and Conformal Field Theory (CFT) are conjecturally connected since the seminal work of Beliavin et al. (Nucl Phys B 241(2):333-380, 1984). Both exhibit exactly solvable structures in two dimensions. A long-standing question ( ...

SPRINGER2022

Geometry of the Loss Landscape in Overparameterized Neural Networks: Symmetries and Invariances

Wulfram Gerstner, Clément Hongler, Johanni Michael Brea, Francesco Spadaro, Berfin Simsek, Arthur Jacot

We study how permutation symmetries in overparameterized multi-layer neural networks generate `symmetry-induced' critical points. Assuming a network with

L

layers of minimal widths

r_1^*, \ldots, r_{L-1}^*

reaches a zero-loss minimum at $ r_1^*! \c ...

2021

Neural Tangent Kernel: Convergence and Generalization in Neural Networks (Invited Paper)

Clément Hongler, Arthur Ulysse Jacot-Guillarmod, Franck Raymond Gabriel, Arthur Jacot

The Neural Tangent Kernel is a new way to understand the gradient descent in deep neural networks, connecting them with kernel methods. In this talk, I'll introduce this formalism and give a number of results on the Neural Tangent Kernel and explain how th ...

ASSOC COMPUTING MACHINERY2021

Implicit Regularization of Random Feature Models

Clément Hongler, Francesco Spadaro, Franck Raymond Gabriel, Berfin Simsek, Arthur Ulysse Jacot-Guillarmod

Random Feature (RF) models are used as efficient parametric approximations of kernel methods. We investigate, by means of random matrix theory, the connection between Gaussian RF models and Kernel Ridge Regression (KRR). For a Gaussian RF model with P feat ...

JMLR-JOURNAL MACHINE LEARNING RESEARCH2020

Scaling description of generalization with number of parameters in deep learning

Matthieu Wyart, Clément Hongler, Mario Geiger, Stefano Spigler, Levent Dogus Sagun, Franck Raymond Gabriel, Arthur Ulysse Jacot-Guillarmod, Arthur Jacot

Supervised deep learning involves the training of neural networks with a large number N of parameters. For large enough N, in the so-called over-parametrized regime, one can essentially fit the training data points. Sparsitybased arguments would suggest th ...

IOP PUBLISHING LTD2020