Unsupervised Feature Learning via Non-Parametric Instance-level Discrimination

TL;DR

This paper introduces an unsupervised method for learning visual features by discriminating individual instances, outperforming existing approaches on ImageNet and enabling efficient retrieval and transfer to other tasks.

Contribution

It proposes a non-parametric, instance-level discrimination approach using noise-contrastive estimation, advancing unsupervised feature learning beyond class-based methods.

Findings

Surpasses state-of-the-art on ImageNet classification

Improves with more training data and better architectures

Requires minimal storage for large-scale retrieval

Abstract

Neural net classifiers trained on data with annotated class labels can also capture apparent visual similarity among categories without being directed to do so. We study whether this observation can be extended beyond the conventional domain of supervised learning: Can we learn a good feature representation that captures apparent similarity among instances, instead of classes, by merely asking the feature to be discriminative of individual instances? We formulate this intuition as a non-parametric classification problem at the instance-level, and use noise-contrastive estimation to tackle the computational challenges imposed by the large number of instance classes. Our experimental results demonstrate that, under unsupervised learning settings, our method surpasses the state-of-the-art on ImageNet classification by a large margin. Our method is also remarkable for consistently improving test performance with more training data and better network architectures. By fine-tuning the learned feature, we further obtain competitive results for semi-supervised learning and object detection tasks. Our non-parametric model is highly compact: With 128 features per image, our method requires only 600MB storage for a million images, enabling fast nearest neighbour retrieval at the run time.