A sample complexity separation between non-convex and convex meta-learning

Mathematics

• Meta-learning 100%
• Reptile 30%
• Learning 23%
• Initialization 20%
• Training 17%
• Black-box Optimization 13%
• Trajectory Optimization 12%
• Subspace 11%
• Formulation 10%
• Learning Algorithm 9%
• Trends 8%
• Linear regression 8%
• Gradient 6%
• Model 5%

Engineering & Materials Science

• Linear networks 32%
• Linear regression 25%
• Learning algorithms 21%
• Trajectories 17%