Efficient Learning by Directed Acyclic Graph For Resource Constrained Prediction

TL;DR

This paper introduces a dynamic programming-based method for learning decision rules in a DAG structure to adaptively select sensors, reducing test-time costs in classification systems with proven convergence and superior empirical performance.

Contribution

It proposes an efficient algorithm for training sensor selection policies in a DAG framework, with convergence guarantees and applicability to various budgeted learning problems.

Findings

Outperforms state-of-the-art algorithms on sensor-rich datasets.

Efficient training with convergence guarantees.

Applicable to diverse budgeted learning scenarios.

Abstract

We study the problem of reducing test-time acquisition costs in classification systems. Our goal is to learn decision rules that adaptively select sensors for each example as necessary to make a confident prediction. We model our system as a directed acyclic graph (DAG) where internal nodes correspond to sensor subsets and decision functions at each node choose whether to acquire a new sensor or classify using the available measurements. This problem can be naturally posed as an empirical risk minimization over training data. Rather than jointly optimizing such a highly coupled and non-convex problem over all decision nodes, we propose an efficient algorithm motivated by dynamic programming. We learn node policies in the DAG by reducing the global objective to a series of cost sensitive learning problems. Our approach is computationally efficient and has proven guarantees of convergence to the optimal system for a fixed architecture. In addition, we present an extension to map other budgeted learning problems with large number of sensors to our DAG architecture and demonstrate empirical performance exceeding state-of-the-art algorithms for data composed of both few and many sensors.