Recent Advances in Open Set Recognition: A Survey

Abstract

In real-world recognition/classification tasks, limited by various objective factors, it is usually difficult to collect training samples to exhaust all classes when training a recognizer or classifier. A more realistic scenario is open set recognition (OSR), where incomplete knowledge of the world exists at training time, and unknown classes can be submitted to an algorithm during testing, requiring the classifiers to not only accurately classify the seen classes, but also effectively deal with the unseen ones. This paper provides a comprehensive survey of existing open set recognition techniques covering various aspects ranging from related definitions, representations of models, datasets, evaluation criteria, and algorithm comparisons. Furthermore, we briefly analyze the relationships between OSR and its related tasks including zero-shot, one-shot (few-shot) recognition/learning techniques, classification with reject option, and so forth. Additionally, we also overview the open world recognition which can be seen as a natural extension of OSR. Importantly, we highlight the limitations of existing approaches and point out some promising subsequent research directions in this field.

arXiv commentAccepted by IEEE TPAMI

References in corpus (30)

• Explaining and Harnessing Adversarial Examples
• Intriguing properties of neural networks
• Prototypical Networks for Few-shot Learning
• Deep Visual Domain Adaptation: A Survey
• NIPS 2016 Tutorial: Generative Adversarial Networks
• Zero-Shot Learning Through Cross-Modal Transfer
• Towards Open World Recognition
• The Extreme Value Machine
• Learning feed-forward one-shot learners
• Selective Classification for Deep Neural Networks
• A PAC-Bayesian bound for Lifelong Learning
• Reducing Network Agnostophobia
• Large-Scale Long-Tailed Recognition in an Open World
• Metric Learning for Novelty and Anomaly Detection
• Unseen Class Discovery in Open-world Classification
• Generative OpenMax for Multi-Class Open Set Classification
• Prediction of Search Targets From Fixations in Open-World Settings
• Open Category Detection with PAC Guarantees
• Recent Advances in Zero-shot Recognition
• Multi-target Unsupervised Domain Adaptation without Exactly Shared Categories
• Lasso type classifiers with a reject option
• DOC: Deep Open Classification of Text Documents
• Open Set Domain Adaptation by Backpropagation
• Online Open World Recognition
• Extreme Value Theory for Open Set Classification -- GPD and GEV Classifiers
• Open-World Visual Recognition Using Knowledge Graphs
• Open-Category Classification by Adversarial Sample Generation
• Query Attack via Opposite-Direction Feature:Towards Robust Image Retrieval
• Vocabulary-informed Extreme Value Learning
• Out-of-Distribution Detection Using an Ensemble of Self Supervised Leave-out Classifiers

Cited by in corpus (81)

• Domain Generalization: A Survey
• Few-Shot Hyperspectral Image Classification With Unknown Classes Using Multitask Deep Learning
• Visual Identification of Individual Holstein-Friesian Cattle via Deep Metric Learning
• SR2CNN: Zero-Shot Learning for Signal Recognition
• Valid prediction intervals for regression problems
• Uncertainty for Identifying Open-Set Errors in Visual Object Detection
• Collective Decision of One-vs-Rest Networks for Open Set Recognition
• Network Embedding with Completely-imbalanced Labels
• Embrace Limited and Imperfect Training Datasets: Opportunities and Challenges in Plant Disease Recognition Using Deep Learning
• On-Device Machine Learning: An Algorithms and Learning Theory Perspective
• Open-World Semi-Supervised Learning
• Machine Learning Techniques for Sensor-based Human Activity Recognition with Data Heterogeneity -- A Review
• Learning without Forgetting for Vision-Language Models
• Open-Set Recognition Using Intra-Class Splitting
• Deep Fidelity in DNN Watermarking: A Study of Backdoor Watermarking for Classification Models
• Open-Set Support Vector Machines
• Learning Discriminative Representations and Decision Boundaries for Open Intent Detection
• Open World Object Detection: A Survey
• Attention2Minority: A salient instance inference-based multiple instance learning for classifying small lesions in whole slide images
• Anomaly Detection Based on Isolation Mechanisms: A Survey
• OpenMatch: Open-set Consistency Regularization for Semi-supervised Learning with Outliers
• A Unified Survey on Anomaly, Novelty, Open-Set, and Out-of-Distribution Detection: Solutions and Future Challenges
• ProtoGCD: Unified and Unbiased Prototype Learning for Generalized Category Discovery
• Automatic Open-World Reliability Assessment
• A fast learning algorithm for One-Class Slab Support Vector Machines
• Understanding Open-Set Recognition by Jacobian Norm and Inter-Class Separation
• OVANet: One-vs-All Network for Universal Domain Adaptation
• Robust Semi-Supervised Learning in Open Environments
• Stylometry recognizes human and LLM-generated texts in short samples
• Extreme Value Theory for Open Set Classification -- GPD and GEV Classifiers
• Unified Classification and Rejection: A One-versus-All Framework
• Hybrid Open-set Segmentation with Synthetic Negative Data
• Adversarial Motorial Prototype Framework for Open Set Recognition
• Evidential Deep Learning for Open Set Action Recognition
• Robust Semi-Supervised Learning for Self-learning Open-World Classes
• Learning Adaptive Embedding Considering Incremental Class
• An Unbiased Risk Estimator for Learning with Augmented Classes
• A Critical Evaluation of Open-World Machine Learning
• Out-of-domain Detection for Natural Language Understanding in Dialog Systems
• Conditional Extreme Value Theory for Open Set Video Domain Adaptation
• Hybrid Models for Open Set Recognition
• AutoScale: Learning to Scale for Crowd Counting and Localization
• Simple Domain Generalization Methods are Strong Baselines for Open Domain Generalization
• Few-shot Open-set Recognition by Transformation Consistency
• Class Distribution Monitoring for Concept Drift Detection
• Open-Set Biometrics: Beyond Good Closed-Set Models
• On-Demand Myoelectric Control Using Wake Gestures to Eliminate False Activations During Activities of Daily Living
• Learning Placeholders for Open-Set Recognition
• Evaluation of Various Open-Set Medical Imaging Tasks with Deep Neural Networks
• End-to-End Learning of Speech 2D Feature-Trajectory for Prosthetic Hands
• Continual Learning in the Presence of Repetition
• SDGMNet: Statistic-based Dynamic Gradient Modulation for Local Descriptor Learning
• ProtoInfoMax: Prototypical Networks with Mutual Information Maximization for Out-of-Domain Detection
• Unsupervised detection and open-set classification of fast-ramped flexibility activation events
• On the link between generative semi-supervised learning and generative open-set recognition
• torchosr -- a PyTorch extension package for Open Set Recognition models evaluation in Python
• Learning without Seeing nor Knowing: Towards Open Zero-Shot Learning
• CheXseen: Unseen Disease Detection for Deep Learning Interpretation of Chest X-rays
• A Generalizable Model-and-Data Driven Approach for Open-Set RFF Authentication
• Multi-task Learning by Leveraging the Semantic Information
• Teacher-Explorer-Student Learning: A Novel Learning Method for Open Set Recognition
• More Information Supervised Probabilistic Deep Face Embedding Learning
• S2OSC: A Holistic Semi-Supervised Approach for Open Set Classification
• Resilience to the Flowing Unknown: an Open Set Recognition Framework for Data Streams
• Explaining Adversarial Vulnerability with a Data Sparsity Hypothesis
• DeepStreamCE: A Streaming Approach to Concept Evolution Detection in Deep Neural Networks
• Generator evaluator-selector net for panoptic image segmentation and splitting unfamiliar objects into parts
• A Glimpse of Physical Layer Decision Mechanisms: Facts, Challenges, and Remedies
• Learning the Compositional Spaces for Generalized Zero-shot Learning
• Non-Exhaustive Learning Using Gaussian Mixture Generative Adversarial Networks
• Open-Set Representation Learning through Combinatorial Embedding
• Deep Learning and Open Set Malware Classification: A Survey
• Visual and Semantic Prototypes-Jointly Guided CNN for Generalized Zero-shot Learning
• Open-Set Vein Biometric Recognition with Deep Metric Learning
• A Symmetric Loss Perspective of Reliable Machine Learning
• Recognition Awareness: An Application of Latent Cognizance to Open-Set Recognition
• Do Visual-Language Grid Maps Capture Latent Semantics?
• Why Out-of-distribution Detection in CNNs Does Not Like Mahalanobis -- and What to Use Instead
• Taking Class Imbalance Into Account in Open Set Recognition Evaluation
• Model Rectification via Unknown Unknowns Extraction from Deployment Samples
• A Review of Computer Vision Methods in Network Security