Building high-level features using large scale unsupervised learning

TL;DR

This paper demonstrates that high-level, class-specific features like face detectors can be learned from unlabeled data using large-scale unsupervised learning, significantly improving object recognition performance.

Contribution

It introduces a large-scale unsupervised learning approach with a deep autoencoder trained on internet images, enabling the discovery of high-level features without labels.

Findings

Successfully learned face and object detectors from unlabeled data.

Features are robust to translation, scaling, and rotation.

Achieved 15.8% accuracy on 20,000 ImageNet categories, a 70% improvement.

Abstract

We consider the problem of building high-level, class-specific feature detectors from only unlabeled data. For example, is it possible to learn a face detector using only unlabeled images? To answer this, we train a 9-layered locally connected sparse autoencoder with pooling and local contrast normalization on a large dataset of images (the model has 1 billion connections, the dataset has 10 million 200x200 pixel images downloaded from the Internet). We train this network using model parallelism and asynchronous SGD on a cluster with 1,000 machines (16,000 cores) for three days. Contrary to what appears to be a widely-held intuition, our experimental results reveal that it is possible to train a face detector without having to label images as containing a face or not. Control experiments show that this feature detector is robust not only to translation but also to scaling and out-of-plane rotation. We also find that the same network is sensitive to other high-level concepts such as cat faces and human bodies. Starting with these learned features, we trained our network to obtain 15.8% accuracy in recognizing 20,000 object categories from ImageNet, a leap of 70% relative improvement over the previous state-of-the-art.