AI.info
Advanced techniques
Ensembles, self-supervision, mixtures of experts — with a decision map for when the simple thing has stopped working.
- Advanced Techniques: A Decision Map
- Ensemble Diversity and Error Correlation
- Bagging, Randomization, and Out-of-Bag Evaluation
- Boosting: From Weak Learners to Gradient Models
- Stacking, Blending, and Weight Averaging
- Transfer Learning and Feature Reuse
- Domain Adaptation and Domain Generalization
- Fine-Tuning and Parameter-Efficient Adaptation
- Few-Shot and In-Context Learning
- Semi-Supervised Learning: Pseudo-Labels, Consistency, and Confirmation Bias
- Self-Supervised Learning: Designing the Objective
- Contrastive and Bootstrap Representation Learning
- Masked Modeling and Predictive Pretraining
- Knowledge Distillation and Teacher–Student Learning
- Meta-Learning and Fast Adaptation
- Continual Learning and Catastrophic Forgetting
- Curriculum Learning, Hard Examples, and Data Scheduling
- Actor–Critic Methods and Proximal Policy Optimization
- Offline Reinforcement Learning and Imitation Learning
- Graph Neural Networks and Message Passing
- Graph Tasks, Sampling, and Failure Modes
- Multi-Task Learning and Mixture-of-Experts
- Evaluating and Debugging Advanced Techniques
- Advanced Techniques Capstone