CSI: Novelty Detection via Contrastive Learning on Distributionally Shifted Instances

Abstract

Novelty detection, i.e., identifying whether a given sample is drawn from outside the training distribution, is essential for reliable machine learning. To this end, there have been many attempts at learning a representation well-suited for novelty detection and designing a score based on such representation. In this paper, we propose a simple, yet effective method named contrasting shifted instances (CSI), inspired by the recent success on contrastive learning of visual representations. Specifically, in addition to contrasting a given sample with other instances as in conventional contrastive learning methods, our training scheme contrasts the sample with distributionally-shifted augmentations of itself. Based on this, we propose a new detection score that is specific to the proposed training scheme. Our experiments demonstrate the superiority of our method under various novelty detection scenarios, including unlabeled one-class, unlabeled multi-class and labeled multi-class settings, with various image benchmark datasets. Code and pre-trained models are available at https://github.com/alinlab/CSI.

NeurIPS 2020. First two authors contributed equally

References in corpus (28)

• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• Representation Learning with Contrastive Predictive Coding
• Improved Regularization of Convolutional Neural Networks with Cutout
• Accurate, Large Minibatch SGD: Training ImageNet in 1 Hour
• Improved Baselines with Momentum Contrastive Learning
• On Calibration of Modern Neural Networks
• SGDR: Stochastic Gradient Descent with Warm Restarts
• A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks
• A Simple Unified Framework for Detecting Out-of-Distribution Samples and Adversarial Attacks
• Data-Efficient Image Recognition with Contrastive Predictive Coding
• What Makes for Good Views for Contrastive Learning?
• Contrastive Multiview Coding
• A Comprehensive Survey of Data Mining-based Fraud Detection Research
• Large Batch Training of Convolutional Networks
• Using Pre-Training Can Improve Model Robustness and Uncertainty
• Deep Anomaly Detection with Outlier Exposure
• Deep Anomaly Detection Using Geometric Transformations
• WAIC, but Why? Generative Ensembles for Robust Anomaly Detection
• Input complexity and out-of-distribution detection with likelihood-based generative models
• Generative Probabilistic Novelty Detection with Adversarial Autoencoders
• Classification-Based Anomaly Detection for General Data
• A Framework For Contrastive Self-Supervised Learning And Designing A New Approach
• Do Deep Generative Models Know What They Don't Know?
• Self-supervised Label Augmentation via Input Transformations
• Detecting Out-of-Distribution Inputs to Deep Generative Models Using Typicality
• Contrastive Training for Improved Out-of-Distribution Detection
• Hybrid Discriminative-Generative Training via Contrastive Learning
• Confident Multiple Choice Learning

Cited by in corpus (48)

• Generalized Out-of-Distribution Detection: A Survey
• Omni-frequency Channel-selection Representations for Unsupervised Anomaly Detection
• A Survey on Unsupervised Anomaly Detection Algorithms for Industrial Images
• Zero-Shot Out-of-Distribution Detection Based on the Pre-trained Model CLIP
• Learning and Evaluating Representations for Deep One-class Classification
• Self-Supervised Anomaly Detection in Computer Vision and Beyond: A Survey and Outlook
• Unsolved Problems in ML Safety
• Contrastive Out-of-Distribution Detection for Pretrained Transformers
• DASVDD: Deep Autoencoding Support Vector Data Descriptor for Anomaly Detection
• Open-Set Recognition: a Good Closed-Set Classifier is All You Need?
• Detecting Outliers with Foreign Patch Interpolation
• Contrastive Knowledge Graph Error Detection
• Unsupervised Outlier Detection using Memory and Contrastive Learning
• Learning on Graphs with Out-of-Distribution Nodes
• Entropic Out-of-Distribution Detection: Seamless Detection of Unknown Examples
• OpenMatch: Open-set Consistency Regularization for Semi-supervised Learning with Outliers
• Neural Transformation Learning for Deep Anomaly Detection Beyond Images
• Precursor-of-Anomaly Detection for Irregular Time Series
• A Unified Survey on Anomaly, Novelty, Open-Set, and Out-of-Distribution Detection: Solutions and Future Challenges
• A graph neural network-based model with Out-of-Distribution Robustness for enhancing Antiretroviral Therapy Outcome Prediction for HIV-1
• Anomalies, Representations, and Self-Supervision
• Visual Anomaly Detection for Images: A Survey
• OVANet: One-vs-All Network for Universal Domain Adaptation
• Towards Trustworthy Dataset Distillation
• Identifying the Defective: Detecting Damaged Grains for Cereal Appearance Inspection
• A Unified Benchmark for the Unknown Detection Capability of Deep Neural Networks
• Evaluation of Point Pattern Features for Anomaly Detection of Defect within Random Finite Set Framework
• Probing the Purview of Neural Networks via Gradient Analysis
• Diffusion-Scheduled Denoising Autoencoders for Anomaly Detection in Tabular Data
• Focus Your Distribution: Coarse-to-Fine Non-Contrastive Learning for Anomaly Detection and Localization
• Elsa: Energy-based learning for semi-supervised anomaly detection
• Discriminative-Generative Representation Learning for One-Class Anomaly Detection
• SLA${}^2$P: Self-supervised Anomaly Detection with Adversarial Perturbation
• Deep Variational Semi-Supervised Novelty Detection
• Self-supervised Pseudo Multi-class Pre-training for Unsupervised Anomaly Detection and Segmentation in Medical Images
• Exploring Covariate and Concept Shift for Detection and Calibration of Out-of-Distribution Data
• CVAD: A generic medical anomaly detector based on Cascade VAE
• Masked Contrastive Learning for Anomaly Detection
• Deep Out-of-Distribution Uncertainty Quantification via Weight Entropy Maximization
• Test Sample Accuracy Scales with Training Sample Density in Neural Networks
• Contrastive Knowledge Distillation for Anomaly Detection in Multi-Illumination/Focus Display Images
• Deep Fraud Detection on Non-attributed Graph
• Multi-Class Anomaly Detection
• ATAC-Net: Zoomed view works better for Anomaly Detection
• ActiveMatch: End-to-end Semi-supervised Active Representation Learning
• Evaluation of Out-of-Distribution Detection Performance of Self-Supervised Learning in a Controllable Environment
• Data augmentation and pre-trained networks for extremely low data regimes unsupervised visual inspection
• FROB: Few-shot ROBust Model for Classification and Out-of-Distribution Detection