Open-set recognition of unseen macromolecules in cellular electron cryo-tomograms by soft large margin centralized cosine loss

Keyphrases

• 3D Image 20%
• 3D Imaging 20%
• Cellular Electron Cryo-Tomography 60%
• Cosine Loss 100%
• Cryo 100%
• Deep Learning 20%
• Deep Neural Network 20%
• Detection Task 20%
• Different Datasets 20%
• Discriminative Features 20%
• Feature Descriptor 20%
• Feature Guided 20%
• High-throughput 20%
• Hypersphere 20%
• Imaging Tools 20%
• Intra-class Variation 40%
• Label Space 20%
• Large Margin 100%
• Learned Features 20%
• Loss Function 20%
• Macromolecular Structure 20%
• Macromolecules 100%
• Native State 20%
• Near-native 20%
• Novel Loss Function 20%
• Open Set Recognition 100%
• Recognition Task 20%
• Simulation Data 20%
• Softmax Loss 20%
• Spatial Organization 20%
• Structural Class 20%
• Structure Recognition 20%
• Submolecular Resolution 20%
• Subtomogram 20%
• Subtomogram Classification 20%
• Testing Data 40%
• Traditional Features 20%
• Training Data 40%
• Unseen 100%
• Verification Accuracy 20%

Computer Science

• 3D Imaging 50%
• Class Variance 100%
• Data Simulation 50%
• Deep Learning Method 50%
• Deep Neural Network 50%
• Discriminative Feature 50%
• feature descriptor 50%
• High Throughput 50%
• Learned Feature 50%
• Open Set Recognition 100%
• Spatial Organization 50%
• Systematic Analysis 50%
• Training Data 100%