0
User
Recs.
Recs.
4
Updates
Last
Updated
Updated
Activity
6 Options
Chocolate is a completely asynchronous optimisation framework. Information between workers is shared relying only on a database.
SpecsUpdate
Pros
Pro Designed for distributed parallelization
Chocolate uses no master process for distributing tasks. Every task is completely independent and only gets its information from a database. Chocolate is thus ideal in controlled computing environments where it is hard to maintain a master process for the duration of the optimisation.
Know any positive aspects of this option?
Cons
Con Bayesian optimization requires making a smoothness assumption
The Bayesian optimization implementation needs the user to specify a kernel width that determines how smooth the system assumes functions will be with no data. Other implementations do not require this assumption and integrate out the smoothness assumption using Markov Chain Monte Carlo (MCMC) sampling.
Know any negative aspects of this option?