Selective Classification for Deep Neural Networks

By Yonatan Geifman, Ran El-Yaniv. 2017

953 citations recorded by the source corpus.

Selective classification techniques (also known as reject option) have not yet been considered in the context of deep neural networks (DNNs). These techniques can potentially significantly improve DNNs prediction performance by trading-off coverage. In this paper we propose a method to construct a selective classifier given a trained neural network. Our method allows a user to set a desired risk level. At test time, the classifier rejects instances as needed, to grant the desired risk (with high probability). Empirical results over CIFAR and ImageNet convincingly demonstrate the viability of our method, which opens up possibilities to operate DNNs in mission-critical applications. For example, using our method an unprecedented 2% error in top-5 ImageNet classification can be guaranteed with probability 99.9%, and almost 60% test coverage.

Subfields: Artificial Intelligence, Safety, Robustness & Interpretability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Safety, Robustness & Interpretability

Citations in the corpus

Improving alignment of dialogue agents via targeted human judgements (2022, background): “This ability can be viewed as a selective prediction of using evidence (or prediction with a reject option) (El-Yaniv and Wiener, 2010; Geifman and El-Yaniv, 2017, 2019; Kamath et al., 2020).”
Teaching language models to support answers with verified quotes (2022, background): “This setup could be described as “selective prediction” (also known as prediction with a reject option) (El-Yaniv et al., 2010; Geifman and El-Yaniv, 2017, 2019; Kamath et al., 2020).”
Consistent Accelerated Inference via Confident Adaptive Transformers (2021, background): “More similar to our work, selective classification (Geifman and El-Yaniv, 2017) allows the model to abstain from answering when not confident, in order to maintain a target error rate only over answered inputs.”
Deep Active Learning with a Neural Architecture Search (2018, background): “The poor performance of MC-dropout over CIFAR-100 may be caused by the large number of classes, as pointed out by [8] in the context of selective classification.”
Can You Trust Your Model's Uncertainty? Evaluating Predictive Uncertainty Under Dataset Shift (2019, methodology): “, 2018), and related work on selective classification (Geifman and El-Yaniv, 2017).”
WILDS: A Benchmark of in-the-Wild Distribution Shifts (2020, neutral): “Many methods for selective prediction have been developed, from simply using softmax probabilities as a proxy for confidence (Cordella et al., 1995; Geifman and El-Yaniv, 2017), to methods involving ensembles of models (Gal and Ghahramani, 2016; Lakshminarayanan et al., 2017; Geifman et al., 2018)…”
Safer Classification by Synthesis (2017, background): “This problem, typically explored in the context of discriminative classification, is known as selective classification [Geifman and El-Yaniv, 2017].”
Recent Advances in Open Set Recognition: A Survey (2018, background): “It should be noted that there have been already a variety of works in the literature regarding classification with reject option [34]–[44].”
Does Your Model Know the Digit 6 Is Not a Cat? A Less Biased Evaluation of "Outlier" Detectors (2018, background): “Geifman and El-Yaniv [29] present a framework for selective classification with deep neural networks that follows the abstention view.”
Selective Question Answering under Domain Shift (2020, background): “This general goal motivates the setting of selective prediction, in which a model outputs both a prediction and a scalar conﬁdence, and abstains on inputs where its conﬁdence is low (El-Yaniv and Wiener, 2010; Geifman and El-Yaniv, 2017).”
Solving Long-tailed Recognition with Deep Realistic Taxonomic Classifier (2020, methodology): “Since the introduction of deep learning has made the estimation of the posterior distribution central to classification, most recent rejection functions consist of thresholding posteriors or derived quantities, such as the posterior entropy [24,25,57].”
Deep Anomaly Detection Using Geometric Transformations (2018, background): “Finally, it would be interesting to attempt using our techniques to leverage deep uncertainty estimation [12, 11], and deep active learning [10].”
Selective clustering for representative paintings selection (2019, background): “[11] constructed a selective classifier by guaranteed risk control during testing phase of deep convolutional networks opening up possibilities to operate DNNs in mission-critical applications.”
A Unifying Review of Deep and Shallow Anomaly Detection (2020, background): “This is known as the problem of classification with a reject option, and it has been studied extensively [480]–[486].”
Generalized ODIN: Detecting Out-of-Distribution Image Without Learning From Out-of-Distribution Data (2020, background): “This problem has been studied for many years [12] and has been discussed in several views such as rejection [8, 5], anomaly detection [1], open set recognition [2], and uncertainty estimation [22, 23, 24].”
On Mixup Training: Improved Calibration and Predictive Uncertainty for Deep Neural Networks (2019, background): “Further mixup based models can also be combined with rejection classifiers, both during training [31] for dealing with label noise, as well as inference [7] to improve the training and classification pipeline in modern deep learning.”