Efficient Learning with Partially Observed Attributes

TL;DR

This paper introduces efficient algorithms for learning linear predictors from partially observed data, demonstrating high accuracy even with minimal attribute information, such as only four pixels in image recognition tasks.

Contribution

The paper presents novel algorithms and theoretical analysis for learning with limited attribute observations, reducing the need for full data access.

Findings

High prediction accuracy with only four pixels in digit recognition

Bounded additional examples needed to compensate for partial observations

Algorithms are efficient and effective in practical scenarios

Abstract

We describe and analyze efficient algorithms for learning a linear predictor from examples when the learner can only view a few attributes of each training example. This is the case, for instance, in medical research, where each patient participating in the experiment is only willing to go through a small number of tests. Our analysis bounds the number of additional examples sufficient to compensate for the lack of full information on each training example. We demonstrate the efficiency of our algorithms by showing that when running on digit recognition data, they obtain a high prediction accuracy even when the learner gets to see only four pixels of each image.