When Analytic Calculus Cracks AdaBoost Code

TL;DR

The paper interrogates the claim that AdaBoost achieves a true minimum-risk solution by analyzing a two-class AdaBoost procedure through a truth-table–based logical framework. It constructs a recursive scheme to compute the weights $\beta_k$ for a set of $p$ weak learners and provides explicit closed-form formulas for the $p=3$ case, with a general induction discussed for larger $p$. The authors show that the analytic weights $\boldsymbol{\beta}_\star^{(p)}$ match the weights produced by scikit-learn's AdaBoost to machine precision and that this approach yields substantial speedups, though it does not always minimize the risk and does not implement Freund and Schapire's original minimum-risk rule. Overall, the work clarifies the exact computation behind AdaBoost in practice, exposes limitations of common implementations, and offers a fast analytic route under certain conditions.

Abstract

The principle of boosting in supervised learning involves combining multiple weak classifiers to obtain a stronger classifier. AdaBoost has the reputation to be a perfect example of this approach. This study analyzes the (two classes) AdaBoost procedure implemented in scikit-learn. This paper shows that AdaBoost is an algorithm in name only, as the resulting combination of weak classifiers can be explicitly calculated using a truth table. Indeed, using a logical analysis of the training set with weak classifiers constructing a truth table, we recover, through an analytical formula, the weights of the combination of these weak classifiers obtained by the procedure. We observe that this formula does not give the point of minimum of the risk, we provide a system to compute the exact point of minimum and we check that the AdaBoost procedure in scikit-learn does not implement the algorithm described by Freund and Schapire.