Sequential Model-based Optimization Algorithm

A Sequential Model-based Optimization Algorithm is an Optimization Algorithm that sequentially selects the next point to evaluate based on a model of the objective function.

• Context:
  • It aims to balance exploration and exploitation by using the model to guide search.
  • It builds a surrogate model of the objective function and uses it to select the next point to evaluate.
  • The model is updated as new points are evaluated, allowing it to improve over time.
  • It can use models such as: Gaussian Processes, Random Forests, and Bayesian Neural Networks.
  • It evaluates points one at a time (or in small batches), sequentially updating the model.
  • It aims to minimize simple regret over the sequence of evaluations.
  • It Uses a model of the objective function to guide search.
  • It Balances exploration and exploitation.
  • It Evaluates points sequentially and updates model.
  • It Optimizes over the sequence of evaluations.
  • ...
• Examples:
  • Bayesian Optimization Algorithm.
  • Upper Confidence Bound Algorithm.
  • Thompson Sampling.
  • Efficient Global Optimization
• See: Derivative-Free Optimization, Model-Based Optimization, Bayesian Optimization.

---

---