Cover Image

On the Cone Effect in the Learning Dynamics

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Zhou, Zhanpeng, Yang, Yongyi, Ren, Jie, Sugiyama, Mahito, Yan, Junchi
Format:	Preprint
Published:	2025
Subjects:	Machine Learning
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

New Evidence of the Two-Phase Learning Dynamics of Neural Networks
by: Zhou, Zhanpeng, et al.
Published: (2025)

Linear Mode Connectivity in Differentiable Tree Ensembles
by: Kanoh, Ryuichi, et al.
Published: (2024)

A Complete Decomposition of KL Error using Refined Information and Mode Interaction Selection
by: Enouen, James, et al.
Published: (2024)

When Graph Language Models Go Beyond Memorization
by: Yamada, Masatsugu, et al.
Published: (2026)

Pseudo-Nonlinear Data Augmentation: A Constrained Energy Minimization Viewpoint
by: Hu, Pingbang, et al.
Published: (2024)

StiefelGen: A Simple, Model Agnostic Approach for Time Series Data Augmentation over Riemannian Manifolds
by: Cheema, Prasad, et al.
Published: (2024)

Dual Riemannian Newton Method on Statistical Manifolds
by: Zhou, Derun, et al.
Published: (2025)

Bringing Structure to Naturalness: On the Naturalness of ASTs
by: Pârţachi, Profir-Petru, et al.
Published: (2025)

On the Learning Dynamics of Two-layer Linear Networks with Label Noise SGD
by: Zhang, Tongcheng, et al.
Published: (2026)

Sharpness-Aware Minimization Efficiently Selects Flatter Minima Late in Training
by: Zhou, Zhanpeng, et al.
Published: (2024)

An Equivariance Toolbox for Learning Dynamics
by: Yang, Yongyi, et al.
Published: (2025)

On the Emergence of Cross-Task Linearity in the Pretraining-Finetuning Paradigm
by: Zhou, Zhanpeng, et al.
Published: (2024)

Quadratic polarity and polar Fenchel-Young divergences from the canonical Legendre polarity
by: Nielsen, Frank, et al.
Published: (2026)

On the Role of Label Noise in the Feature Learning Process
by: Han, Andi, et al.
Published: (2025)

EasyDGL: Encode, Train and Interpret for Continuous-time Dynamic Graph Learning
by: Chen, Chao, et al.
Published: (2023)

The Sharpness Disparity Principle in Transformers for Accelerating Language Model Pre-Training
by: Wang, Jinbo, et al.
Published: (2025)

Non-stationary Online Learning for Curved Losses: Improved Dynamic Regret via Mixability
by: Zhang, Yu-Jie, et al.
Published: (2025)

How Graph Neural Networks Learn: Lessons from Training Dynamics
by: Yang, Chenxiao, et al.
Published: (2023)

Same Graph, Different Likelihoods: Calibration of Autoregressive Graph Generators via Permutation-Equivalent Encodings
by: Fredsgaard, Laurits, et al.
Published: (2026)

Learning with Incomplete Context: Linear Contextual Bandits with Pretrained Imputation
by: Yan, Hao, et al.
Published: (2025)

Provable Low-Frequency Bias of In-Context Learning of Representations
by: Yang, Yongyi, et al.
Published: (2025)

Swing-by Dynamics in Concept Learning and Compositional Generalization
by: Yang, Yongyi, et al.
Published: (2024)

Learning to Solve Combinatorial Optimization under Positive Linear Constraints via Non-Autoregressive Neural Networks
by: Wang, Runzhong, et al.
Published: (2024)

Learning Divergence Fields for Shift-Robust Graph Representations
by: Wu, Qitian, et al.
Published: (2024)

Topological Invariance and Breakdown in Learning
by: Yang, Yongyi, et al.
Published: (2025)

Euphonium: Steering Video Flow Matching via Process Reward Gradient Guided Stochastic Dynamics
by: Zhong, Ruizhe, et al.
Published: (2026)

Molecule Generation for Drug Design: a Graph Learning Perspective
by: Yang, Nianzu, et al.
Published: (2022)

mHC-lite: You Don't Need 20 Sinkhorn-Knopp Iterations
by: Yang, Yongyi, et al.
Published: (2026)

Learning Interpretable Features in Audio Latent Spaces via Sparse Autoencoders
by: Paek, Nathan, et al.
Published: (2025)

Parallel Recursive LSTM
by: Gaudreault, Tristan, et al.
Published: (2026)

On robust overfitting: adversarial training induced distribution matters
by: Tian, Runzhi, et al.
Published: (2023)

Trustworthy Federated Learning via Blockchain
by: Yang, Zhanpeng, et al.
Published: (2022)

NTKMTL: Mitigating Task Imbalance in Multi-Task Learning from Neural Tangent Kernel Perspective
by: Qin, Xiaohan, et al.
Published: (2025)

On the Cone Effect and Modality Gap in Medical Vision-Language Embeddings
by: Restrepo, David, et al.
Published: (2026)

HERTA: A High-Efficiency and Rigorous Training Algorithm for Unfolded Graph Neural Networks
by: Yang, Yongyi, et al.
Published: (2024)

Learning with Complementary Labels Revisited: The Selected-Completely-at-Random Setting Is More Practical
by: Wang, Wei, et al.
Published: (2023)

Riemannian Langevin Dynamics: Strong Convergence of Geometric Euler-Maruyama Scheme
by: Zhan, Zhiyuan, et al.
Published: (2026)

GSINA: Improving Subgraph Extraction for Graph Invariant Learning via Graph Sinkhorn Attention
by: Yan, Junchi, et al.
Published: (2024)

On the Cost and Benefit of Chain of Thought: A Learning-Theoretic Perspective
by: Zhang, Yue, et al.
Published: (2026)

Fast T2T: Optimization Consistency Speeds Up Diffusion-Based Training-to-Testing Solving for Combinatorial Optimization
by: Li, Yang, et al.
Published: (2025)