Current browse context:
stat.ML
Change to browse by:
References & Citations
Computer Science > Machine Learning
Title: Estimating informativeness of samples with Smooth Unique Information
(Submitted on 17 Jan 2021 (v1), last revised 28 Mar 2021 (this version, v2))
Abstract: We define a notion of information that an individual sample provides to the training of a neural network, and we specialize it to measure both how much a sample informs the final weights and how much it informs the function computed by the weights. Though related, we show that these quantities have a qualitatively different behavior. We give efficient approximations of these quantities using a linearized network and demonstrate empirically that the approximation is accurate for real-world architectures, such as pre-trained ResNets. We apply these measures to several problems, such as dataset summarization, analysis of under-sampled classes, comparison of informativeness of different data sources, and detection of adversarial and corrupted examples. Our work generalizes existing frameworks but enjoys better computational properties for heavily over-parametrized models, which makes it possible to apply it to real-world networks.
Submission history
From: Hrayr Harutyunyan [view email][v1] Sun, 17 Jan 2021 10:29:29 GMT (12262kb,D)
[v2] Sun, 28 Mar 2021 08:24:40 GMT (12256kb,D)
Link back to: arXiv, form interface, contact.