Flexible Modeling and Multitask Learning using Differentiable Tree Ensembles
This work addresses the problem of restricted functionality in tree ensemble toolkits for machine learning practitioners, offering incremental improvements in flexibility and efficiency.
The authors tackled the limited modeling capabilities of existing tree ensemble toolkits by proposing a flexible framework that supports arbitrary loss functions, missing responses, and multi-task learning, resulting in tree ensembles that are 100x more compact and 23% more expressive than those from popular toolkits.
Decision tree ensembles are widely used and competitive learning models. Despite their success, popular toolkits for learning tree ensembles have limited modeling capabilities. For instance, these toolkits support a limited number of loss functions and are restricted to single task learning. We propose a flexible framework for learning tree ensembles, which goes beyond existing toolkits to support arbitrary loss functions, missing responses, and multi-task learning. Our framework builds on differentiable (a.k.a. soft) tree ensembles, which can be trained using first-order methods. However, unlike classical trees, differentiable trees are difficult to scale. We therefore propose a novel tensor-based formulation of differentiable trees that allows for efficient vectorization on GPUs. We perform experiments on a collection of 28 real open-source and proprietary datasets, which demonstrate that our framework can lead to 100x more compact and 23% more expressive tree ensembles than those by popular toolkits.