LARF: Two-level Attention-based Random Forests with a Mixture of Contamination Models

TL;DR

LARF introduces a novel two-level attention mechanism in random forests, utilizing leaf and tree attention with a mixture of contamination models, optimized via quadratic programming, to improve model performance.

Contribution

The paper proposes LARF, a new attention-based random forest model with a two-level attention mechanism and a mixture of contamination models, enhancing interpretability and tuning simplicity.

Findings

Numerical experiments demonstrate LARF's effectiveness on real datasets.

The model's attention parameters are efficiently trained through quadratic optimization.

LARF outperforms traditional random forests in various tasks.

Abstract

New models of the attention-based random forests called LARF (Leaf Attention-based Random Forest) are proposed. The first idea behind the models is to introduce a two-level attention, where one of the levels is the "leaf" attention and the attention mechanism is applied to every leaf of trees. The second level is the tree attention depending on the "leaf" attention. The second idea is to replace the softmax operation in the attention with the weighted sum of the softmax operations with different parameters. It is implemented by applying a mixture of the Huber's contamination models and can be regarded as an analog of the multi-head attention with "heads" defined by selecting a value of the softmax parameter. Attention parameters are simply trained by solving the quadratic optimization problem. To simplify the tuning process of the models, it is proposed to make the tuning contamination parameters to be training and to compute them by solving the quadratic optimization problem. Many numerical experiments with real datasets are performed for studying LARFs. The code of proposed algorithms can be found in https://github.com/andruekonst/leaf-attention-forest.