JevForest: Path Voting for Budgeted Feature Acquisition
Abstract
Choosing which information to observe is central to prediction under limited observation budgets. We study JevForest, a feature acquisition policy that aggregates path-dependent proposals from bootstrapped trees, weights them by global training information gain, and predicts from the acquired values with a shared masked classifier. An online implementation queries Jev for semantic answers selected by this policy. On small balanced held-out samples, four-question forest acquisition achieves accuracy $0.729$ on AG News (n=48), compared with $0.667$ for a static gain ranking and $0.583$ for random ordering. On TREC (n=24), the ordering reverses: forest accuracy is $0.667$, compared with $0.750$ and $0.833$. Asking all eight questions in one batch yields higher accuracy at lower measured cost and latency than four sequential forest queries; direct Jev classification matches the batch accuracy while costing less. Offline MiniBooNE experiments yield accuracy 0.845±0.010 at ten features and 0.885±0.008 at forty features over three jointly varying data and forest seeds (mean ± sample standard deviation). A companion Newton boosting implementation provides preliminary full-feature synthetic results. These exploratory findings establish a working Jev acquisition workflow but do not support a general advantage for path voting: its value depends on the task, predictor, and the distinction between question budgets and actual query costs.
Related
- Non-Myopic Active Feature Acquisition via Pathwise Policy Gradients
- Active Feature Acquisition for Cost-Efficient Temporal Prediction with Reduced Participant Burden
- Constructing Data-Driven Budgeted Uncertainty Sets for Robust Optimization
- Calibration and Risk-Budgeted Decisions with System One Models
- Budgeted Multi-Source Counterfactual Annotation for Off-Policy Evaluation
BibTeX
@misc{yan2026jevforest,
title = {{JevForest: Path Voting for Budgeted Feature Acquisition}},
author = {Yu Yan},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.10615},
eprint = {2610.10615},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.10615},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - JevForest: Path Voting for Budgeted Feature Acquisition
AU - Yu Yan
PY - 2026
DA - 2026/10/07
PB - arXiv
DO - 10.48550/arxiv.2610.10615
UR - https://arxiv.org/abs/2610.10615
AB - Choosing which information to observe is central to prediction under limited observation budgets. We study JevForest, a feature acquisition policy that aggregates path-dependent proposals from bootstrapped trees, weights them by global training information gain, and predicts from the acquired values with a shared masked classifier. An online implementation queries Jev for semantic answers selected by this policy. On small balanced held-out samples, four-question forest acquisition achieves accuracy $0.729$ on AG News (n=48), compared with $0.667$ for a static gain ranking and $0.583$ for random ordering. On TREC (n=24), the ordering reverses: forest accuracy is $0.667$, compared with $0.750$ and $0.833$. Asking all eight questions in one batch yields higher accuracy at lower measured cost and latency than four sequential forest queries; direct Jev classification matches the batch accuracy while costing less. Offline MiniBooNE experiments yield accuracy 0.845±0.010 at ten features and 0.885±0.008 at forty features over three jointly varying data and forest seeds (mean ± sample standard deviation). A companion Newton boosting implementation provides preliminary full-feature synthetic results. These exploratory findings establish a working Jev acquisition workflow but do not support a general advantage for path voting: its value depends on the task, predictor, and the distinction between question budgets and actual query costs.
ER -