On Mesa-Optimization in Autoregressively Trained Transformers: Emergence and Capabilityby Chenyu Zheng, Wei Huang, Rongzhen Wang,…
BOLD: Boolean Logic Deep Learningby Van Minh Nguyen, Cristian Ocampo, Aymen Askri, Louis Leconte, Ba-Hien…
BadGD: A unified data-centric framework to identify gradient descent vulnerabilitiesby Chi-Hua Wang, Guang ChengFirst submitted…
Improved Particle Approximation Error for Mean Field Neural Networksby Atsushi NitandaFirst submitted to arxiv on:…
A generalized neural tangent kernel for surrogate gradient learningby Luke Eilers, Raoul-Martin Memmesheimer, Sven GoedekeFirst…
Fundamental computational limits of weak learnability in high-dimensional multi-index modelsby Emanuele Troiani, Yatin Dandi, Leonardo…
Novel Kernel Models and Exact Representor Theory for Neural Networks Beyond the Over-Parameterized Regimeby Alistair…
Minimizing UCB: a Better Local Search Strategy in Local Bayesian Optimizationby Zheyi Fan, Wenyu Wang,…
Parameter-free Clipped Gradient Descent Meets Polyakby Yuki Takezawa, Han Bao, Ryoma Sato, Kenta Niwa, Makoto…
Optimal Rates for Vector-Valued Spectral Regularization Learning Algorithmsby Dimitri Meunier, Zikai Shen, Mattes Mollenhauer, Arthur…