QUBO Decision Tree: Annealing Machine Extends Decision Tree Splitting
This addresses the problem of low accuracy in regression trees for tabular data prediction, though it appears incremental as it builds on existing tree methods with a computational twist.
The paper tackles the insufficient accuracy of regression trees due to simple decision rules by extending them to multi-dimensional boundaries, achieving this by transforming the training process into a QUBO problem solvable with an annealing machine.
This paper proposes an extension of regression trees by quadratic unconstrained binary optimization (QUBO). Regression trees are very popular prediction models that are trainable with tabular datasets, but their accuracy is insufficient because the decision rules are too simple. The proposed method extends the decision rules in decision trees to multi-dimensional boundaries. Such an extension is generally unimplementable because of computational limitations, however, the proposed method transforms the training process to QUBO, which enables an annealing machine to solve this problem.