Neural Active Learning with Performance Guarantees

TL;DR

This paper introduces a neural active learning algorithm with performance guarantees in non-parametric streaming settings, leveraging Neural Tangent Kernel approximations and adaptive thresholding to efficiently select labels without prior knowledge.

Contribution

It develops a novel neural active learning method that is agnostic to prior knowledge and provides theoretical guarantees on regret and label complexity.

Findings

Guarantees on cumulative regret and label requests depend on function complexity.

In the linear case, recovers known minimax generalization bounds.

Algorithm is computationally efficient and adaptive to unknown function complexity.

Abstract

We investigate the problem of active learning in the streaming setting in non-parametric regimes, where the labels are stochastically generated from a class of functions on which we make no assumptions whatsoever. We rely on recently proposed Neural Tangent Kernel (NTK) approximation tools to construct a suitable neural embedding that determines the feature space the algorithm operates on and the learned model computed atop. Since the shape of the label requesting threshold is tightly related to the complexity of the function to be learned, which is a-priori unknown, we also derive a version of the algorithm which is agnostic to any prior knowledge. This algorithm relies on a regret balancing scheme to solve the resulting online model selection problem, and is computationally efficient. We prove joint guarantees on the cumulative regret and number of requested labels which depend on the complexity of the labeling function at hand. In the linear case, these guarantees recover known minimax results of the generalization error as a function of the label complexity in a standard statistical learning setting.