AI Roadmap
by Taeksoo Kim
1. Autonomous
1.1. AutoML
1.2. Meta learning
2. Resource Management
2.1. Student-teacher
2.2. Model compression
2.3. Asynchronous algorithm
2.4. Data augmentation
3. Better Perception & Understanding
3.1. Residual concept
3.2. Attention-based model
3.3. Reasoning
3.3.1. Graph networks
4. Generation & Creativity
4.1. Implicit generative models
4.2. Image-to-Image
4.3. Program induction
4.4. AutoEncoder
4.5. Factor disentanglement
5. Model Interpretation
5.1. Data/Feature Visualization
5.2. XAI (Explainable AI)
6. Control in RL
6.1. Imitation learning
6.2. Inverse RL
7. Safety & Privacy
7.1. Adversarial examples
7.2. Training from private samples
8. Human related
8.1. Active learning
8.2. Human in the loop
8.3. Crowd sourcing
9. General Topic
9.1. Overfitting
9.2. Class imbalance
10. Etc.
10.1. various sensor
10.1.1. RGB, Depth, IR, LiDAR