Curriculum Learning for Improved Femur Fracture Classification: Scheduling Data with Prior Knowledge and Uncertainty

Keyphrases

• Curriculum Learning 100%
• Scheduling Information 100%
• Proximal Femoral Fracture 100%
• Fracture Classification 100%
• Femoral Fractures 100%
• Training Set 66%
• Training Samples 66%
• Scoring Function 66%
• Publicly Available 33%
• Set-based 33%
• Learning Approaches 33%
• Convolutional Neural Network 33%
• MNIST 33%
• Limited Amount of Data 33%
• Class Imbalance 33%
• X-ray Images 33%
• Automatic Classification 33%
• Trauma Surgeon 33%
• Radiograph 33%
• Treatment Choice 33%
• Fracture Location 33%
• Representative Dataset 33%
• Self-paced 33%
• Novel Formulations 33%
• Knowledge Tree 33%
• Fracture Complexity 33%
• Hierarchical Knowledge 33%
• Proximal Femur 33%
• Class-based 33%
• Label Noise 33%
• Digit Recognition 33%
• AO Systems 33%

Computer Science

• Convolutional Neural Network 100%
• Training Sample 66%
• Scoring Function 66%
• Perform Experiment 33%
• Learning Approach 33%
• Automatic Classification 33%
• Class Imbalance 33%
• Clinical Outcome 33%