paper-with-me

홈 › Papers

How You Start Matters for Generalization

2022-06-17 · Sameera Ramasinghe, Lachlan MacDonald, Moshiur Farazi, Hemanth Saratchandran, Simon Lucey

Characterizing the remarkable generalization properties of over-parameterized neural networks remains an open problem. In this paper, we promote a shift of focus towards initialization rather than neural architecture or (stochastic) gradient descent to explain this implicit regularization. Through a Fourier lens, we derive a general result for the spectral bias of neural networks and show that the generalization of neural networks is heavily tied to their initialization. Further, we empirically solidify the developed theoretical insights using practical, deep networks. Finally, we make a case against the controversial flat-minima conjecture and show that Fourier analysis grants a more reliable framework for understanding the generalization of neural networks.

📄 PDF Abstract BibTeX arXiv:2206.08558

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Understanding Decoupled and Early Weight Decay

2020-12-27 · Johan Bjorck, Kilian Weinberger, Carla Gomes

Weight decay (WD) is a traditional regularization technique in deep learning, but despite its ubiquity, its behavior is still an area of active research. Golatkar et al. have recently shown that WD only matters at the st…

How To Start a Grassroots Movement

2022-09-03 · David Ehrlich, Nora Szech

We study the influence of social messages that promote a digital public good, a COVID-19 tracing app. We vary whether subjects receive a digital message from another subject, and, if so, at what cost it came. Observed ma…

Large Loss Matters in Weakly Supervised Multi-Label Classification

2022-06-08 · CVPR 2022 1 · Youngwook Kim, Jae Myung Kim, Zeynep Akata, Jungwoo Lee

Weakly supervised multi-label classification (WSML) task, which is to learn a multi-label classification using partially observed labels per image, is becoming increasingly important due to its huge annotation cost. In t…

ClassificationMemorizationMulti-Label ClassificationMUlTI-LABEL-ClASSIFICATION

Time Matters in Regularizing Deep Networks: Weight Decay and Data Augmentation Affect Early Learning Dynamics, Matter Little Near Convergence

2019-05-30 · NeurIPS 2019 12 · Aditya Golatkar, Alessandro Achille, Stefano Soatto

Regularization is typically understood as improving generalization by altering the landscape of local extrema to which the model eventually converges. Deep neural networks (DNNs), however, challenge this view: We show th…

Data Augmentation

Where You Place the Norm Matters: From Prejudiced to Neutral Initializations

2025-05-16 · Emanuele Francazi, Francesco Pinto, Aurelien Lucchi, Marco Baity-Jesi

Normalization layers, such as Batch Normalization and Layer Normalization, are central components in modern neural networks, widely adopted to improve training stability and generalization. While their practical effectiv…