Stochastic Hill-Climbing Algorithm

A Stochastic Hill-Climbing Algorithm is a Hill-Climbing Algorithm that Randomly chooses the starting move and/or samples the slope.

• AKA: Stochastic Hill-Climbing,
• Example:
  • a Stochastic Gradient Descent / Iterative Gradient Descent Algorithm.
• See: Monte Carlo Algorithm.

---

---

References

2009

• http://en.wikipedia.org/wiki/Stochastic_hill_climbing
  • Stochastic hill climbing is a variant of the basic hill climbing method. While basic hill climbing always chooses the steepest uphill move, stochastic hill climbing chooses at random from among the uphill moves. The probability of selection may vary with the steepness of the uphill move.

2007

• (Culotta et al., 2007b) ⇒ Aron Culotta, Michael Wick, Robert Hall, Matthew Marzilli, and Andrew McCallum. (2007). “Canonicalization of Database Records using Adaptive Similarity Measures.” In: Proceedings of the 13th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD-2007).
  • We describe how to learn these costs from a small amount of manually annotated data using stochastic hill-climbing.

---