High Dimensional Restrictive Federated Model Selection with multi-objective Bayesian Optimization over shifted distributions
This addresses the challenge of model selection in privacy-sensitive domains like healthcare where data sharing is restricted, offering an incremental improvement over existing federated learning methods by focusing on hyper-parameter optimization rather than direct training.
The paper tackles the problem of selecting machine learning models under restrictive federated settings where data cannot leave local sites, proposing Restrictive Federated Model Selection (RFMS) to optimize hyper-parameters across sites using multi-objective Bayesian optimization for high-dimensional data with shifted distributions. Empirical results show that using local and remote performances improves generalization across sites, with multi-objective Bayesian optimization achieving increased performance in terms of dominated hypervolumes.
A novel machine learning optimization process coined Restrictive Federated Model Selection (RFMS) is proposed under the scenario, for example, when data from healthcare units can not leave the site it is situated on and it is forbidden to carry out training algorithms on remote data sites due to either technical or privacy and trust concerns. To carry out a clinical research under this scenario, an analyst could train a machine learning model only on local data site, but it is still possible to execute a statistical query at a certain cost in the form of sending a machine learning model to some of the remote data sites and get the performance measures as feedback, maybe due to prediction being usually much cheaper. Compared to federated learning, which is optimizing the model parameters directly by carrying out training across all data sites, RFMS trains model parameters only on one local data site but optimizes hyper-parameters across other data sites jointly since hyper-parameters play an important role in machine learning performance. The aim is to get a Pareto optimal model with respective to both local and remote unseen prediction losses, which could generalize well across data sites. In this work, we specifically consider high dimensional data with shifted distributions over data sites. As an initial investigation, Bayesian Optimization especially multi-objective Bayesian Optimization is used to guide an adaptive hyper-parameter optimization process to select models under the RFMS scenario. Empirical results show that solely using the local data site to tune hyper-parameters generalizes poorly across data sites, compared to methods that utilize the local and remote performances. Furthermore, in terms of dominated hypervolumes, multi-objective Bayesian Optimization algorithms show increased performance across multiple data sites among other candidates.