This may take a while. Are you sure you want to continue?
COMBO (COMmon Bayesian Optimization Library) uses a Bayesian Optimization approach based on Thompson sampling, fast Cholesky decomposition and automatic hyper-parameter tuning, to guide and optimize experimental research.
Given a small set of data annotated with a property that needs to be optimized, and a large number of candidates, COMBO will suggest the most promising candidates to explore next, for the best chances at quickly maximizing the desired property. In a few iterations of this process, COMBO will converge toward the true optimal value, requiring only a fraction of the experiments that would be needed to achieve the same result through random search.
More information about COMBO’s algorithm, along with examples of applications to Material Science datasets, can be found in our documentation and in the following scientific publication:Ueno, T., Rhone, T.D., Hou, Z., Mizoguchi, T. and Tsuda, K., 2016. COMBO: An efficient Bayesian optimization library for materials science. Materials Discovery.