Graceful Degradation and Related Fields

Abstract

When machine learning models encounter data which is out of the distribution on which they were trained they have a tendency to behave poorly, most prominently over-confidence in erroneous predictions. Such behaviours will have disastrous effects on real-world machine learning systems. In this field graceful degradation refers to the optimisation of model performance as it encounters this out-of-distribution data. This work presents a definition and discussion of graceful degradation and where it can be applied in deployed visual systems. Following this a survey of relevant areas is undertaken, novelly splitting the graceful degradation problem into active and passive approaches. In passive approaches, graceful degradation is handled and achieved by the model in a self-contained manner, in active approaches the model is updated upon encountering epistemic uncertainties. This work communicates the importance of the problem and aims to prompt the development of machine learning strategies that are aware of graceful degradation.

References in corpus (39)

• Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
• A Review of Uncertainty Quantification in Deep Learning: Techniques, Applications and Challenges
• Conditional Adversarial Domain Adaptation
• Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm
• A Simple Unified Framework for Detecting Out-of-Distribution Samples and Adversarial Attacks
• Hands-on Bayesian Neural Networks -- a Tutorial for Deep Learning Users
• Zero-Shot Learning by Convex Combination of Semantic Embeddings
• Meta-Learning for Semi-Supervised Few-Shot Classification
• Domain Separation Networks
• VSE++: Improving Visual-Semantic Embeddings with Hard Negatives
• Three scenarios for continual learning
• Evidential Deep Learning to Quantify Classification Uncertainty
• Meta-Learning with Implicit Gradients
• Learning to Learn without Gradient Descent by Gradient Descent
• Unsupervised Domain Adaptation through Self-Supervision
• Beyond temperature scaling: Obtaining well-calibrated multiclass probabilities with Dirichlet calibration
• Learning to Optimize
• Hierarchically Structured Meta-learning
• Sharpness-Aware Minimization for Efficiently Improving Generalization
• Robust Unsupervised Domain Adaptation for Neural Networks via Moment Alignment
• Dual Dynamic Inference: Enabling More Efficient, Adaptive and Controllable Deep Inference
• Implicit Weight Uncertainty in Neural Networks
• Posterior Network: Uncertainty Estimation without OOD Samples via Density-Based Pseudo-Counts
• A Variance Maximization Criterion for Active Learning
• Noise Contrastive Priors for Functional Uncertainty
• Continual Learning of a Mixed Sequence of Similar and Dissimilar Tasks
• Alpha MAML: Adaptive Model-Agnostic Meta-Learning
• Contrastive Training for Improved Out-of-Distribution Detection
• SkipNet: Learning Dynamic Routing in Convolutional Networks
• Why is the Mahalanobis Distance Effective for Anomaly Detection?
• Transferring Knowledge across Learning Processes
• Bias-Reduced Uncertainty Estimation for Deep Neural Classifiers
• Continual Learning with Node-Importance based Adaptive Group Sparse Regularization
• HyperGAN: A Generative Model for Diverse, Performant Neural Networks
• Bayesian Generative Active Deep Learning
• Online Fast Adaptation and Knowledge Accumulation: a New Approach to Continual Learning
• Heuristic Domain Adaptation
• Active Learning with Multiple Kernels
• Novel Uncertainty Framework for Deep Learning Ensembles