Generative Low-Shot Network Expansion

Conventional deep learning classifiers are static in the sense that they are trained on a predefined set of classes and learning to classify a novel class typically requires re-training. In this work, we address the problem of Low-Shot network-expansion learning. We introduce a learning framework which enables expanding a pre-trained (base) deep network to classify novel classes when the number of examples for the novel classes is particularly small. We present a simple yet powerful hard distillation method where the base network is augmented with additional weights to classify the novel classes, while keeping the weights of the base network unchanged. We show that since only a small number of weights needs to be trained, the hard distillation excels in low-shot training scenarios. Furthermore, hard distillation avoids detriment to classification performance on the base classes. Finally, we show that low-shot network expansion can be done with a very small memory footprint by using a compact generative model of the base classes training data with only a negligible degradation relative to learning with the full training set.

Authors

• Adi Hayat
• Mark Kliger
• Shachar Fleishman
• Daniel Cohen-Or

Keywords

• Training
• Training data
• Data models
• Memory management
• Task analysis
• Robots
• Feature extraction
• Network Expansion
• Training Set
• Deep Network
• Small Footprint
• Base Classes
• Data-generating Model
• Feature Space
• Feature Representation
• Deeper Layers
• Confusion Matrix
• Network Layer
• Classification Network
• Representation Learning
• Fully-connected Layer
• Gaussian Mixture Model
• Training Examples
• Compact Representation
• Feature Extraction Network
• Extreme Learning Machine
• Hard Constraints
• FC Layer
• Single Example
• Distillation Loss
• Entire Training Set
• Class Prototypes
• Training Batch
• Catastrophic Forgetting
• Incremental Learning
• Image Domain