Raghuram Shankar
← Bayesian Optimization

Part 2 of 8

Intuition

The core idea fits in one sentence: maintain a probabilistic belief about the whole function, and let the next evaluation be chosen by a decision rule that weighs predicted value against remaining ignorance. Everything else is machinery for making that sentence precise.

  1. The loop: model, decide, evaluate, repeat
  2. Uncertainty is the resource
  3. Drive the loop yourself
  4. Does the model really help? Run the experiment
  5. The catch: what can go wrong
  6. Quiz: 5 questions