Naoki Nishikawa

5

Papers

8

Total Citations

Papers (5)

State Space Models are Provably Comparable to Transformers in Dynamic Token Selection

From Shortcut to Induction Head: How Data Diversity Shapes Algorithm Selection in Transformers

NeurIPS 2025arXiv

Degrees of Freedom for Linear Attention: Distilling Softmax Attention with Optimal Feature Efficiency

NeurIPS 2025arXiv

Two-layer neural network on infinite dimensional data: global optimization guarantee in the mean-field regime

Adaptive Topological Feature via Persistent Homology: Filtration Learning for Point Clouds

NeurIPS 2023arXiv