Learning to Support: Exploiting Structure Information in Support Sets for One-Shot Learning

Keyphrases

• Support Set 100%
• One-shot Learning 100%
• Query Point 100%
• CNN Embeddings 100%
• Attention Mechanism 100%
• Meta-learner 100%
• Structure Information 100%
• Trajectory Optimization 50%
• Training Procedure 50%
• Classification Problem 50%
• Deep Learning 50%
• Overfitting 50%
• Deep Convolutional Neural Network (deep CNN) 50%
• One Sample 50%
• Final Model 50%
• Art Performance 50%
• Class Representative 50%
• Competitive Learning 50%
• Class Prototype 50%
• Learning Approaches 50%
• Class Representation 50%
• Model Features 50%
• Unseen Class 50%
• Deep Nets 50%
• Labeled Data 50%
• Large Capacity 50%
• Few Samples 50%
• Few-shot Classification 50%
• Meta-learning 50%

Computer Science

• Convolutional Neural Network 100%
• Attention (Machine Learning) 100%
• Deep Learning 50%
• Learning Approach 50%
• Approximation (Algorithm) 50%
• Classification Problem 50%
• Art Performance 50%
• Good Performance 50%
• Labeled Data Set 50%