Active Multi-Task Representation Learning

TL;DR

This paper introduces an active learning-based method for selecting relevant source tasks in multi-task representation learning, significantly reducing sample complexity and improving efficiency in leveraging source data for target tasks.

Contribution

It presents the first formal study on resource task sampling using active learning techniques, with a novel algorithm that estimates task relevance and improves sample efficiency.

Findings

Theoretically reduces source task sample complexity by up to a factor of the number of source tasks.

Demonstrates effectiveness on real-world computer vision datasets with linear and CNN representations.

Provides a new perspective on task selection in multi-task learning using active learning principles.

Abstract

To leverage the power of big data from source tasks and overcome the scarcity of the target task samples, representation learning based on multi-task pretraining has become a standard approach in many applications. However, up until now, choosing which source tasks to include in the multi-task learning has been more art than science. In this paper, we give the first formal study on resource task sampling by leveraging the techniques from active learning. We propose an algorithm that iteratively estimates the relevance of each source task to the target task and samples from each source task based on the estimated relevance. Theoretically, we show that for the linear representation class, to achieve the same error rate, our algorithm can save up to a \textit{number of source tasks} factor in the source task sample complexity, compared with the naive uniform sampling from all source tasks. We also provide experiments on real-world computer vision datasets to illustrate the effectiveness of our proposed method on both linear and convolutional neural network representation classes. We believe our paper serves as an important initial step to bring techniques from active learning to representation learning.