rFITb: Random Forest in the Browser

This paper introduces rFITb, a distributed computing platform that enables the execution of computationally intensive random forest jobs on personal devices such as smartphones and personal computers. The platform leverages the increased computational capacity of personal devices to distribute and execute jobs globally, providing an efficient alternative to cloud-based services. The paper describes rFITb's architecture and design optimizations, along with a comparative evaluation of its performance against Python's sklearn ensemble random forest classifier on various datasets. The results show that rFITb outperforms the sklearn classifier in terms of model time, while also providing a mechanism for managing failure-prone volunteers.