Towards Robust Pattern Recognition: A Review
arXiv:2006.06976
Abstract
The accuracies for many pattern recognition tasks have increased rapidly year by year, achieving or even outperforming human performance. From the perspective of accuracy, pattern recognition seems to be a nearly-solved problem. However, once launched in real applications, the high-accuracy pattern recognition systems may become unstable and unreliable, due to the lack of robustness in open and changing environments. In this paper, we present a comprehensive review of research towards robust pattern recognition from the perspective of breaking three basic and implicit assumptions: closed-world assumption, independent and identically distributed assumption, and clean and big data assumption, which form the foundation of most pattern recognition models. Actually, our brain is robust at learning concepts continually and incrementally, in complex, open and changing environments, with different contexts, modalities and tasks, by showing only a few examples, under weak or noisy supervision. These are the major differences between human intelligence and machine intelligence, which are closely related to the above three assumptions. After witnessing the significant progress in accuracy improvement nowadays, this review paper will enable us to analyze the shortcomings and limitations of current methods and identify future research directions for robust pattern recognition.
References in corpus (61)
- Deep Learning in Neural Networks: An Overview
- Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
- Show, Attend and Tell: Neural Image Caption Generation with Visual Attention
- Prototypical Networks for Few-shot Learning
- mixup: Beyond Empirical Risk Minimization
- DeCAF: A Deep Convolutional Activation Feature for Generic Visual Recognition
- An Overview of Multi-Task Learning in Deep Neural Networks
- Deep Visual Domain Adaptation: A Survey
- A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks
- Graph Neural Networks: A Review of Methods and Applications
- Improved Techniques for Training GANs
- Momentum Contrast for Unsupervised Visual Representation Learning
- Matching Networks for One Shot Learning
- Domain Adaptation for Statistical Classifiers
- Towards Deep Neural Network Architectures Robust to Adversarial Examples
- Semi-Supervised Learning with Ladder Networks
- Regularization With Stochastic Transformations and Perturbations for Deep Semi-Supervised Learning
- Training Convolutional Networks with Noisy Labels
- A simple neural network module for relational reasoning
- Beyond Sharing Weights for Deep Domain Adaptation
- Weakly Supervised Object Localization with Multi-fold Multiple Instance Learning
- Adversarial Discriminative Domain Adaptation
- Understanding Adversarial Training: Increasing Local Stability of Neural Nets through Robust Optimization
- Aligning Books and Movies: Towards Story-like Visual Explanations by Watching Movies and Reading Books
- Towards Biologically Plausible Deep Learning
- Unsupervised Visual Representation Learning by Context Prediction
- One Model To Learn Them All
- A Kernel Method for the Two-Sample Problem
- Learning feed-forward one-shot learners
- Making Risk Minimization Tolerant to Label Noise
- Stacked What-Where Auto-encoders
- Unsupervised Learning by Predicting Noise
- Few-Shot Adversarial Domain Adaptation
- Realistic Evaluation of Deep Semi-Supervised Learning Algorithms
- Exploiting Unlabeled Data in CNNs by Self-supervised Learning to Rank
- Unsupervised Learning of Visual Representations using Videos
- Masking: A New Perspective of Noisy Supervision
- Robust Loss Functions under Label Noise for Deep Neural Networks
- Joint Unsupervised Learning of Deep Representations and Image Clusters
- Learn to Combine Modalities in Multimodal Deep Learning
- Do CIFAR-10 Classifiers Generalize to CIFAR-10?
- Joint Optimization Framework for Learning with Noisy Labels
- Learning Representations for Automatic Colorization
- Improving the Robustness of Deep Neural Networks via Stability Training
- Learn to Grow: A Continual Structure Learning Framework for Overcoming Catastrophic Forgetting
- PsyPhy: A Psychophysics Driven Evaluation Framework for Visual Recognition
- Revisiting Self-Supervised Visual Representation Learning
- Low-Shot Learning from Imaginary Data
- Few-Shot Learning Through an Information Retrieval Lens
- Generative OpenMax for Multi-Class Open Set Classification
- Learning by Association - A versatile semi-supervised training method for neural networks
- Deep Cocktail Network: Multi-source Unsupervised Domain Adaptation with Category Shift
- CliqueCNN: Deep Unsupervised Exemplar Learning
- Semi-Supervised Learning via Compact Latent Space Clustering
- Look, Listen and Learn
- Memory Matching Networks for One-Shot Image Recognition
- Iterative Learning with Open-set Noisy Labels
- Consistent Multitask Learning with Nonlinear Output Relations
- Recovering the Missing Link: Predicting Class-Attribute Associations for Unsupervised Zero-Shot Learning
- Robust Metric Learning by Smooth Optimization
- MultiNet: Multi-Modal Multi-Task Learning for Autonomous Driving