Modern engineering and scientific problems face a common problem –– how to efficiently find promising candidate solutions from a large space of possible choices of configurations. This complex configuration space could be the hyperparameters of a machine learning model, environment configuration variables of a hardware system, resource allocation of a cluster, or design parameters of a simulated materials.
Searching and tuning these configurations are often resources/time intensive. SigOpt is an optimization platform that automate this process. It streamlines the users process to get to the desirable results, but also encourages the users to experiment with their problem.
SigOpt is a platform designed for sample-efficient search of desirable outcomes in a complex configuration space while considering one or multiple objective metrics.
Multiple objectives - SigOpt supports multiobjective and constrained optimization.
Complex search space - SigOpt supports, continuous, integer, discrete, categorical variables, parameter constraints, and conditional dependencies.
Flexible API - The core SigOpt API follows the RESTful design principle.
Interactive User Experience - Sigopt has a web dashboard that shows an overview of experiments and allows for exploration.
Sign up for a free account and get started with SigOpt today!