Unsupervised Learning on Neural Network Outputs: with Application in Zero-shot Learning
This work addresses the challenge of leveraging neural network outputs for unsupervised learning and zero-shot learning, offering a novel application that improves performance in large-scale image classification tasks.
The authors tackled the problem of extracting richer information from neural network outputs beyond classification by applying PCA and ICA to reveal hidden structure, and used these features to achieve state-of-the-art results in zero-shot learning on ImageNet with over 20,000 classes.
The outputs of a trained neural network contain much richer information than just an one-hot classifier. For example, a neural network might give an image of a dog the probability of one in a million of being a cat but it is still much larger than the probability of being a car. To reveal the hidden structure in them, we apply two unsupervised learning algorithms, PCA and ICA, to the outputs of a deep Convolutional Neural Network trained on the ImageNet of 1000 classes. The PCA/ICA embedding of the object classes reveals their visual similarity and the PCA/ICA components can be interpreted as common visual features shared by similar object classes. For an application, we proposed a new zero-shot learning method, in which the visual features learned by PCA/ICA are employed. Our zero-shot learning method achieves the state-of-the-art results on the ImageNet of over 20000 classes.