Estimating Training Data Influence by Tracking Gradient Descent

We introduce a method called TrackIn that computes the influence of a training example on a prediction made by the model, by tracking how the loss on the test point changes during the training process whenever the training example of interest was utilized. We provide a scalable implementation of TrackIn via a combination of a few key ideas: (a) a first-order approximation to the exact computation, (b) using random projections to speed up the computation of the first-order approximation for large models, (c) using saved checkpoints of standard training procedures, and (d) cherry-picking layers of a deep neural network. An experimental evaluation shows that TrackIn is more effective in identifying mislabelled training examples than other related methods such as influence functions and representer points. We also discuss insights from applying the method on vision, regression and natural language tasks.

[1]  S. Weisberg,et al.  Residuals and Influence in Regression , 1982 .

[2]  Barak A. Pearlmutter.  Fast Exact Multiplication by the Hessian , 1994, Neural Computation.

[3]  R. Pace,et al.  Sparse spatial autoregressions , 1997 .

[4]  Bernhard Schölkopf,et al.  A Generalized Representer Theorem , 2001, COLT/EuroCOLT.

[5]  Li Fei-Fei,et al.  ImageNet: A large-scale hierarchical image database , 2009, CVPR.

[6]  Alex Krizhevsky,et al.  Learning Multiple Layers of Features from Tiny Images , 2009 .

[7]  R. Tibshirani,et al.  Prototype selection for interpretable classification , 2011, 1202.5933.

[8]  Thomas B. Moeslund,et al.  Long-Term Occupancy Analysis Using Graph-Based Optimisation in Thermal Imagery , 2013, 2013 IEEE Conference on Computer Vision and Pattern Recognition.

[9]  Art B. Owen,et al.  Sobol' Indices and Shapley Value , 2014, SIAM/ASA J. Uncertain. Quantification.

[10]  Cynthia Rudin,et al.  The Bayesian Case Model: A Generative Approach for Case-Based Reasoning and Prototype Classification , 2014, NIPS.

[11]  David P. Woodruff.  Sketching as a Tool for Numerical Linear Algebra , 2014, Found. Trends Theor. Comput. Sci..

[12]  Xiang Zhang,et al.  Character-level Convolutional Networks for Text Classification , 2015, NIPS.

[13]  Rico Sennrich,et al.  Edinburgh Neural Machine Translation Systems for WMT 16 , 2016, WMT.

[14]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[15]  Oluwasanmi Koyejo,et al.  Examples are not enough, learn to criticize! Criticism for Interpretability , 2016, NIPS.

[16]  Carlos Guestrin,et al.  "Why Should I Trust You?": Explaining the Predictions of Any Classifier , 2016, ArXiv.

[17]  Subhashini Venugopalan,et al.  Development and Validation of a Deep Learning Algorithm for Detection of Diabetic Retinopathy in Retinal Fundus Photographs. , 2016, JAMA.

[18]  Percy Liang,et al.  Understanding Black-box Predictions via Influence Functions , 2017, ICML.

[19]  Scott Lundberg,et al.  A Unified Approach to Interpreting Model Predictions , 2017, NIPS.

[20]  Timo Aila,et al.  Temporal Ensembling for Semi-Supervised Learning , 2016, ICLR.

[21]  Art B. Owen,et al.  On Shapley Value for Measuring Importance of Dependent Inputs , 2016, SIAM/ASA J. Uncertain. Quantification.

[22]  Ankur Taly,et al.  Axiomatic Attribution for Deep Networks , 2017, ICML.

[23]  Guoyin Wang,et al.  Baseline Needs More Love: On Simple Word-Embedding-Based Models and Associated Pooling Mechanisms , 2018, ACL.

[24]  Pradeep Ravikumar,et al.  Representer Point Selection for Explaining Deep Neural Networks , 2018, NeurIPS.

[25]  Taku Kudo,et al.  SentencePiece: A simple and language independent subword tokenizer and detokenizer for Neural Text Processing , 2018, EMNLP.

[26]  Takanori Maehara,et al.  Data Cleansing for Models Trained with SGD , 2019, NeurIPS.

[27]  Shujian Huang,et al.  Online Distilling from Checkpoints for Neural Machine Translation , 2019, NAACL.

[28]  Yanyao Shen,et al.  Learning with Bad Training Data via Iterative Trimmed Loss Minimization , 2018, ICML.

[29]  Costas J. Spanos,et al.  Towards Efficient Data Valuation Based on the Shapley Value , 2019, AISTATS.

[30]  James Y. Zou,et al.  Data Shapley: Equitable Valuation of Data for Machine Learning , 2019, ICML.

[31]  Yulia Tsvetkov,et al.  Fortifying Toxic Speech Detectors Against Veiled Toxicity , 2020, EMNLP.