BRiG-AFA: Bellman Risk-to-Go Learning for Non-Myopic Active Feature Acquisition
2026-08-03 • Machine Learning
Machine LearningArtificial Intelligence
AI summaryⓘ
The authors present a new method called GABA for deciding which missing features to measure next when making predictions under a limited budget. Instead of complex methods, their supervised approach learns risk estimates backward in time to better handle multi-step decisions. Testing on datasets like Fashion-MNIST showed improved accuracy compared to simpler one-step methods, especially when several features can be acquired. Results on another dataset were mixed, indicating the method works better under certain conditions. Overall, their work supports using direct backward risk learning for this problem and outlines what further comparison studies are needed.
Active feature acquisitionGreedy algorithmsReinforcement learningBellman equationRisk-to-go functionSupervised learningTerminal riskFashion-MNISTMiniBooNEBackward fitting
Authors
Jiaorong Feng, Qian Li, Ying Li
Abstract
Active feature acquisition (AFA) asks which unobserved feature to measure next for each test instance under a budget. Greedy rules are easy to train but can overlook context features whose value is realized only through later acquisitions, while reinforcement-learning and generative approaches introduce difficult optimization or conditional-density estimation. We introduce \method, a deployable, supervised alternative that learns a separate candidate-conditioned risk-to-go function for every remaining budget. Starting from the one-step terminal classification risk, the functions are fitted backward with Bellman targets; inference greedily minimizes the learned terminal risk using only observed values, the mask, candidate identity, and remaining budget. A controlled non-myopic benchmark shows the expected mechanism: at budgets two and three, \method improves accuracy over its one-step ablation by $4.84\pm2.17$ and $4.39\pm1.10$ percentage points (mean $\pm$ standard error over five seeds). On Fashion-MNIST with 20 candidate pixels, it improves accuracy at every nontrivial reported budget on average, including $10.20\pm0.74$ points at four acquisitions; its mean paired gain across budgets $\{2,4,8,12,16\}$ is $3.50\pm0.37$ points. A three-seed MiniBooNE study is mixed at small budgets but positive at 8 and 16 acquisitions, identifying a current boundary rather than supporting a universal claim. These results establish a reproducible mechanism-level case for direct Bellman risk regression and delimit the experiments still needed for state-of-the-art comparison.