Frequency Bias in Neural Networks for Input of Non-Uniform Density
Jul 12, 2020
Speakers
Ronen Basri
Speaker · 0 followers
Meirav Galun
Speaker · 0 followers
Amnon Geifman
Speaker · 0 followers
David Jacobs
Speaker · 0 followers
Yoni Kasten
Speaker · 0 followers
Shira Kritchman
Speaker · 0 followers
About
Recent works have partly attributed the generalization ability of over-parameterized neural networks to frequency bias – networks trained with gradient descent on data drawn from a uniform distribution find a low frequency fit before high frequency ones. As realistic training sets are not drawn from a uniform distribution, we here use the Neural Tangent Kernel (NTK) model to explore the effect of variable density on training dynamics. Our results, which combine analytic and empirical observations, show that when learning a pure harmonic function of frequency κ, convergence at a point ∈^d-1 occurs in time O(κ^d/p()) where p() denotes the local density at . Specifically, for data in ^1 we analytically derive the eigenfunctions of the kernel associated with the NTK for two-layer networks. We further prove convergence results for deep, fully connected networks with respect to the spectral decomposition of the NTK. Our empirical study highlights similarities and differences between deep and shallow networks in this model.Recent works have partly attributed the generalization ability of over-parameterized neural networks to frequency bias – networks trained with gradient descent on data drawn from a uniform distribution find a low frequency fit before high frequency ones. As realistic training sets are not drawn from a uniform distribution, we here use the Neural Tangent Kernel (NTK) model to explore the effect of variable density on training dynamics. Our results, which combine analytic and empirical observation…
Categories
AI & Data Science
Category · 10.8k presentations
About ICML 2020
The International Conference on Machine Learning (ICML) is the premier gathering of professionals dedicated to the advancement of the branch of artificial intelligence known as machine learning. ICML is globally renowned for presenting and publishing cutting-edge research on all aspects of machine learning used in closely related areas like artificial intelligence, statistics and data science, as well as important application areas such as machine vision, computational biology, speech recognition, and robotics. ICML is one of the fastest growing artificial intelligence conferences in the world. Participants at ICML span a wide range of backgrounds, from academic and industrial researchers, to entrepreneurs and engineers, to graduate students and postdocs.
Like the format? Trust SlidesLive to capture your next event!
Professional recording and live streaming, delivered globally.
Sharing
Recommended Videos
Presentations on similar topic, category or speaker
A Meta-Learning Approach for Image Classification Architecture Recommendation
Watch later
Accelerating Large-Scale Inference with Anisotropic Vector Quantization
Watch later
Ruiqi Guo, …
Source Separation with Deep Generative Priors
Watch later
Counterfactual Data Augmentationi using Locally Factored Dynamics
Watch later
Silviu Pitis, …
Proper Network Interpretability Helps Adversarial Robustness in Classification
Watch later
