Understanding classifier errors by examining influential neighbors
The DOI matches its Crossref record; 9 of the 9 references it cites were found in a registry or an index.
Abstract
Modern supervised learning algorithms can learn very accurate and complex discriminating functions. But when these classifiers fail, this complexity can also be a drawback because there is no easy, intuitive way to diagnose why they are failing and remedy the problem. This important question has received little attention. To address this problem, we propose a novel method to analyze and understand a classifier's errors. Our method centers around a measure of how much influence a training example has on the classifier's prediction for a test example. To understand why a classifier is mispredicting the label of a given test example, the user can find and review the most influential training examples that caused this misprediction, allowing them to focus their attention on relevant areas of the data space. This will aid the user in determining if and how the training data is inconsistently labeled or lacking in diversity, or if the feature representation is insufficient. As computing the influence of each training example is computationally impractical, we propose a novel distance metric to approximate influence for boosting classifiers that is fast enough to be used interactively. We also show several novel use paradigms of our distance metric. Through experiments, we show that it can be used to find incorrectly or inconsistently labeled training examples, to find specific areas of the data space that need more training data, and to gain insight into which features are missing from the current representation.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
ImageNet: A large-scale hierarchical image database
Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., & Fei-Fei, L. (2009) · 2009 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2009.5206848 · Cited by 65,333 · cited 7×
“We restricted our analysis to the 28 categories below the " insect " category in the ImageNet Large Scale Visual Recognition Challenge (ILSVRC) 2010 data set.”
BackgroundMethodInfluentialCortes, C., & Vapnik, V. (1995) · Machine Learning · doi:10.1007/bf00994018 · Cited by 59,421
Torralba, A., & Efros, A. A. (2011) · CVPR 2011 · doi:10.1109/cvpr.2011.5995347 · Cited by 2,440 · cited 1×
“This is often not the case when a classifier is used in practice [28, 22].”
BackgroundDescribing objects by their attributes
Farhadi, A., Endres, I., Hoiem, D., & Forsyth, D. (2009) · 2009 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2009.5206772 · Cited by 1,917
The 2005 PASCAL Visual Object Classes Challenge
Everingham, M., Zisserman, A., Williams, C. K. I., Van Gool, L., Allan, M., Bishop, C. M., Chapelle, O., Dalal, N., Deselaers, T., Dorkó, G., et al. (2006) · Lecture Notes in Computer Science · doi:10.1007/11736790_8 · Cited by 765
JAABA: Interactive machine learning for automatic annotation of animal behavior
Kabra, M., Robie, A. A., Rivera-Alba, M., Branson, S., & Branson, K. (2013) · Nature Methods · doi:10.1038/nmeth.2281 · Cited by 644
Self-citationIdentifying Mislabeled Training Data
Brodley, C. E., & Friedl, M. A. (1999) · Journal of Artificial Intelligence Research · doi:10.1613/jair.606 · Cited by 612
Diagnosing Error in Object Detectors
Hoiem, D., Chodpathumwan, Y., & Dai, Q. (2012) · Lecture Notes in Computer Science · doi:10.1007/978-3-642-33712-3_25 · Cited by 520
HOGgles: Visualizing Object Detection Features
Vondrick, C., Khosla, A., Malisiewicz, T., & Torralba, A. (2013) · 2013 IEEE International Conference on Computer Vision · doi:10.1109/iccv.2013.8 · Cited by 282 · cited 1×
“A few recent approaches [29, 32] have been developed for visualizing feature representations.”
Background
Cited by
A Survey on the Explainability of Supervised Machine Learning
Burkart, N., & Huber, M. F. (2021) · Journal of Artificial Intelligence Research · doi:10.1613/jair.1.12228 · Cited by 1,008
Auditing black-box models for indirect influence
Adler, P., Falk, C., Friedler, S. A., Nix, T., Rybeck, G., Scheidegger, C., Smith, B., & Venkatasubramanian, S. (2018) · Knowledge and Information Systems · doi:10.1007/s10115-017-1116-3 · Cited by 280
“visualizing the behavior of neurons and inputs have also been studied [28], [14], [16].”
Auditing Black-Box Models for Indirect Influence
Adler, P., Falk, C., Friedler, S. A., Rybeck, G., Scheidegger, C., Smith, B., & Venkatasubramanian, S. (2016) · 2016 IEEE 16th International Conference on Data Mining (ICDM) · doi:10.1109/icdm.2016.0011 · Cited by 280
The Dangers of Post-hoc Interpretability: Unjustified Counterfactual Explanations
Laugel, T., Lesot, M.-J., Marsala, C., Renard, X., & Detyniecki, M. (2019) · Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2019/388 · Cited by 194
FairTest: Discovering Unwarranted Associations in Data-Driven Applications
Tramer, F., Atlidakis, V., Geambasu, R., Hsu, D., Hubaux, J.-P., Humbert, M., Juels, A., & Lin, H. (2017) · 2017 IEEE European Symposium on Security and Privacy (EuroS&P) · doi:10.1109/eurosp.2017.29 · Cited by 171
“A number of methods exist in ML to understand data features and modeling errors [33]–[36].”
MethodFairTest: Discovering Unwarranted Associations in Data-Driven Applications
Tramèr, F., Atlidakis, V., Geambasu, R., Hsu, D., Hubaux, J.-P., Humbert, M., Juels, A., & Lin, H. (2015) · arXiv · doi:10.48550/arxiv.1510.02377 · Cited by 171
“Fourth, debugging the ML component in an application is often conducted in an ad hoc fashion, with the ML practitioner manually employing exploratory data analysis to understand data features and modeling errors [24, 29, 40, 41].”
MethodUnjustified Classification Regions and Counterfactual Explanations in Machine Learning
Laugel, T., Lesot, M.-J., Marsala, C., Renard, X., & Detyniecki, M. (2020) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-46147-8_3 · Cited by 28
Metadata
- Authors
- Kabra, M., Robie, A., & Branson, K.
- Venue
- 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)
- Publisher
- IEEE
- Published
- 2015-06
- Pages
- 3917-3925
- ISSN
- 1063-6919 (ISSN-L)
- OpenAlex
- W1918466020
- Cited by
- 22
- References
- 9
- FWCI
- 1.33
- Type
- conference paper
- Language
- en
- Open access
- closed
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.