Cover Image

Saddle-To-Saddle Dynamics in Deep ReLU Networks: Low-Rank Bias in the First Saddle Escape

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Bantzis, Ioannis, Simon, James B., Jacot, Arthur
Format:	Preprint
Published:	2025
Subjects:	Machine Learning Artificial Intelligence
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Loss Landscape of Shallow ReLU-like Neural Networks: Stationary Points, Saddle Escape, and Network Embedding
by: Wu, Frank Zhengqing, et al.
Published: (2024)

Dimer-Enhanced Optimization: A First-Order Approach to Escaping Saddle Points in Neural Network Training
by: Hu, Yue, et al.
Published: (2025)

Saddle-to-Saddle Dynamics Explains A Simplicity Bias Across Neural Network Architectures
by: Zhang, Yedi, et al.
Published: (2025)

Stochastic Gradient Descent in the Saddle-to-Saddle Regime of Deep Linear Networks
by: Corlouer, Guillaume, et al.
Published: (2026)

A Theory of Saddle Escape in Deep Nonlinear Networks
by: Rawal, Divit, et al.
Published: (2026)

N-ReLU: Zero-Mean Stochastic Extension of ReLU
by: Manik, Md Motaleb Hossen, et al.
Published: (2025)

The Geometry of ReLU Networks through the ReLU Transition Graph
by: Dhayalkar, Sahil Rajesh
Published: (2025)

The Resurrection of the ReLU
by: Horuz, Coşku Can, et al.
Published: (2025)

Dimension-Free Saddle-Point Escape in Muon
by: Long, Yanlin, et al.
Published: (2026)

Only Strict Saddles in the Energy Landscape of Predictive Coding Networks?
by: Innocenti, Francesco, et al.
Published: (2024)

Pathwise Explanation of ReLU Neural Networks
by: Lim, Seongwoo, et al.
Published: (2025)

Beyond ReLU: Chebyshev-DQN for Enhanced Deep Q-Networks
by: Yazdannik, Saman, et al.
Published: (2025)

Bilevel Optimization over Saddle Points of Zero-Sum Markov Games
by: Zheng, Zihao, et al.
Published: (2026)

Sufficient Conditions for Stability of Minimum-Norm Interpolating Deep ReLU Networks
by: Harzli, Ouns El, et al.
Published: (2026)

Efficiently Escaping Saddle Points for Policy Optimization
by: Khorasani, Sadegh, et al.
Published: (2023)

Neural Network-based High-index Saddle Dynamics Method for Searching Saddle Points and Solution Landscape
by: Liu, Yuankai, et al.
Published: (2024)

Unveiling the Training Dynamics of ReLU Networks through a Linear Lens
by: Ye, Longqing
Published: (2025)

Three Quantization Regimes for ReLU Networks
by: Ou, Weigutian, et al.
Published: (2024)

$λ$-GELU: Learning Gating Hardness for Controlled ReLU-ization in Deep Networks
by: Pérez-Corral, Cristian, et al.
Published: (2026)

Activation-Descent Regularization for Input Optimization of ReLU Networks
by: Yu, Hongzhan, et al.
Published: (2024)

Bottleneck Structure in Learned Features: Low-Dimension vs Regularity Tradeoff
by: Jacot, Arthur
Published: (2023)

Relating Piecewise Linear Kolmogorov Arnold Networks to ReLU Networks
by: Schoots, Nandi, et al.
Published: (2025)

ReLU Networks for Exact Generation of Similar Graphs
by: Ghafoor, Mamoona, et al.
Published: (2026)

Benign Overfitting for Regression with Trained Two-Layer ReLU Networks
by: Park, Junhyung, et al.
Published: (2024)

Deep ReLU Networks Have Surprisingly Simple Polytopes
by: Fan, Feng-Lei, et al.
Published: (2023)

Convergence of Shallow ReLU Networks on Weakly Interacting Data
by: Dana, Léo, et al.
Published: (2025)

Expressive Power of ReLU and Step Networks under Floating-Point Operations
by: Park, Yeachan, et al.
Published: (2024)

Is ReLU Adversarially Robust?
by: Sooksatra, Korn, et al.
Published: (2024)

Geometry-induced Regularization in Deep ReLU Neural Networks
by: Bona-Pellissier, Joachim, et al.
Published: (2024)

RePO: Understanding Preference Learning Through ReLU-Based Optimization
by: Wu, Junkang, et al.
Published: (2025)

Hessian-guided Perturbed Wasserstein Gradient Flows for Escaping Saddle Points
by: Yamamoto, Naoya, et al.
Published: (2025)

Uncertainty Quantification with Bayesian Higher Order ReLU KANs
by: Giroux, James, et al.
Published: (2024)

Precise Verification of Transformers through ReLU-Catalyzed Abstraction Refinement
by: Liu, Hengjie, et al.
Published: (2026)

Uncovering Layer-Dependent Activation Sparsity Patterns in ReLU Transformers
by: Wild, Cody, et al.
Published: (2024)

Detecting Invariant Manifolds in ReLU-Based RNNs
by: Eisenmann, Lukas, et al.
Published: (2025)

Mixed Dynamics In Linear Networks: Unifying the Lazy and Active Regimes
by: Tu, Zhenfeng, et al.
Published: (2024)

Topological Signatures of ReLU Neural Network Activation Patterns
by: Bosca, Vicente, et al.
Published: (2025)

A Lower Bound for the Number of Linear Regions of Ternary ReLU Regression Neural Networks
by: Nakahara, Yuta, et al.
Published: (2025)

Does Flatness imply Generalization for Logistic Loss in Univariate Two-Layer ReLU Network?
by: Qiao, Dan, et al.
Published: (2025)

Stable Minima Cannot Overfit in Univariate ReLU Networks: Generalization by Large Step Sizes
by: Qiao, Dan, et al.
Published: (2024)