A neural anisotropic view of underspecification in deep learning
원문 초록
The underspecification of most machine learning pipelines means that we cannot rely solely on validation performance to assess the robustness of deep learning systems to naturally occurring distribution shifts. Instead, making sure that a neural network can generalize across a large number of different situations requires to understand the specific way in which it solves a task. In this work, we propose to study this problem from a geometric perspective with the aim to understand two key characteristics of neural network solutions in underspecified settings: how is the geometry of the learned function related to the data representation? And, are deep networks always biased towards simpler solutions, as conjectured in recent literature? We show that the way neural networks handle the underspecification of these problems is highly dependent on the data representation, affecting both the geometry and the complexity of the learned predictors. Our results highlight that understanding the architectural inductive bias in deep learning is fundamental to address the fairness, robustness, and generalization of these systems.
함께 살펴볼 논문
공통 태그를 근거로 연결했습니다. 실제 인용 관계나 논문의 우수성을 뜻하지 않습니다.
입력 경로는 소형 트랜스포머의 퓨샷 바인딩을 형성하지만 제로샷은 아니다: 완전 열거 가능한 연구
Input Pathways Shape Few-Shot, Not Zero-Shot, Binding in Tiny Transformers: A Fully-Enumerable Study
Decomposing Yes-No Bias in LLM Moral Judgment: Stance, Position, Word, and Verdict
영향력 있는 다원주의적 정렬 연구를 위한 로드맵
A Roadmap to Impactful Pluralistic Alignment Research
공정한 자기지도 학습을 위한 편향 없는 Open World 정규화
Unbiased Open World Regularization for Fair Self-Supervised Learning
텍스트-이미지 생성에서 대표 다양성을 위한 다축 Max@K 강화학습
Multi-Axis Max@K Reinforcement Learning for Representative Diversity in Text-to-Image Generation
Spectral Margins and Memory Routing in Looped Transformers
토론
이 논문과 연결된 포럼 글이 아직 없습니다.