Boosting Federated Multitask Learning: Transfer Effects in Cross-Domain Drug-Target Interaction Prediction

Using federated learning to collaborate with other parties is becoming common when conducting machine learning on high-value data. In our work, we try to expand the possibilities of existing federated models to apply them to multitask problems. Previously we presented FedMTBoost, which used boosting to enhance predictive performance in a small drug-target interaction problem. This paper investigates the algorithm’s performance on a larger scale using a cross-domain benchmark data set. The original motivation for boosting was to weigh the data adaptively; thus, the multitask transfer can happen on different tasks in different iterations. However, our results suggest that improvement is mostly present in federated scenarios, leading us to believe that the data and model weights can improve the federated transfer by adapting the models to the clients’ data. Furthermore, the boosting algorithms generally outperform traditional baselines when fewer data are available, either in tasks or samples. In this paper, we examine these findings in multiple experiments and try to explain the improvements achieved.