Adaptive Histogram-based Gradient Boosted Trees For Federated Learning

Abstract

Federated Learning is a novel approach to machine learning without the need to collect data in a central place. Issues of privacy concerns and regulatory restrictions make it impossible or expensive to bring all data to a centralized machine learning cluster. Federated learning helps to overcome these issues by collaboratively training a machine learning model without transmitting any raw data. In this talk, we present a novel implementation of a Histogram-Based Gradient Boosting Tree algorithm for Federated Learning and its advantages over various other Federated Learning approaches.