paper-with-me

Papers

Are All Linear Regions Created Equal?

2022-02-23 · Matteo Gamba, Adrian Chmielewski-Anders, Josephine Sullivan, Hossein Azizpour, Mårten Björkman

The number of linear regions has been studied as a proxy of complexity for ReLU networks. However, the empirical success of network compression techniques like pruning and knowledge distillation, suggest that in the overparameterized setting, linear regions density might fail to capture the effective nonlinearity. In this work, we propose an efficient algorithm for discovering linear regions and use it to investigate the effectiveness of density in capturing the nonlinearity of trained VGGs and ResNets on CIFAR-10 and CIFAR-100. We contrast the results with a more principled nonlinearity measure based on function variation, highlighting the shortcomings of linear regions density. Furthermore, interestingly, our measure of nonlinearity clearly correlates with model-wise deep double descent, connecting reduced test error with reduced nonlinearity, and increased local similarity of linear regions.

📄 PDF Abstract BibTeX arXiv:2202.11749

Code (1)

magamba/linear-regions 공식 구현 pytorch

Tasks

AllKnowledge Distillation

Methods 이 논문이 사용한 방법론

Pruning 설명 없음

Similar Papers 제목 키워드 기반

Not All Parameters Are Created Equal: Smart Isolation Boosts Fine-Tuning Performance

2025-08-29 · Yao Wang, Di Liang, Minlong Peng arxiv

Supervised fine-tuning (SFT) is a pivotal approach to adapting large language models (LLMs) for downstream tasks; however, performance often suffers from the ``seesaw phenomenon'', where indiscriminate parameter updates …

Not All Character N-grams Are Created Equal: A Study in Authorship Attribution

2015-05-01 · HLT 2015 5 · Thamar Solorio, Upendra Sapkota, Manuel Montes, Steven Bethard
AllAuthorship Attribution

Not All Titles are Created Equal: Financial Document Structure Extraction Shared Task

2021-09-01 · FNP 2021 9 · Anubhav Gupta, Hanna Abi Akl, Hugues de Mazancourt
All

Not All Contexts Are Created Equal: Better Word Representations with Variable Attention

2015-09-01 · EMNLP 2015 9 · Wang Ling, Yulia Tsvetkov, Silvio Amir, Ferm 외
AllDependency ParsingMachine TranslationPart-Of-Speech Tagging

Maximum Likelihood Identification of Linear Models with Integrating Disturbances for Offset-Free Control

2024-06-06 · Steven J. Kuntz, James B. Rawlings

This paper addresses the identification of models for offset-free model predictive control (MPC), where LTI models are augmented with (fictitious) uncontrollable integrating modes, called integrating disturbances. The st…

Model Predictive Control