• shareshare
  • link
  • cite
  • add
auto_awesome_motion View all 4 versions
Publication . Part of book or chapter of book . 2017

Improve Deep Learning with Unsupervised Objective

Shufei Zhang; Kaizhu Huang; Rui Zhang; Amir Hussain;
Open Access
Published: 01 Jan 2017
Publisher: Springer International Publishing

We propose a novel approach capable of embedding the unsupervised objective into hidden layers of the deep neural network (DNN) for preserving important unsupervised information. To this end, we exploit a very simple yet effective unsupervised method, i.e. principal component analysis (PCA), to generate the unsupervised “label" for the latent layers of DNN. Each latent layer of DNN can then be supervised not just by the class label, but also by the unsupervised “label" so that the intrinsic structure information of data can be learned and embedded. Compared with traditional methods which combine supervised and unsupervised learning, our proposed model avoids the needs for layer-wise pre-training and complicated model learning e.g. in deep autoencoder. We show that the resulting model achieves state-of-the-art performance in both face and handwriting data simply with learning of unsupervised “labels".

Subjects by Vocabulary

Microsoft Academic Graph classification: Embedding Face (geometry) Unsupervised learning Artificial intelligence business.industry business Autoencoder Principal component analysis Artificial neural network Multilayer perceptron Pattern recognition Class (biology) Computer science Deep learning

ACM Computing Classification System: ComputingMethodologies_PATTERNRECOGNITION


Deep learning, Multi-layer perceptron, Unsupervised learning, Recognition

Related Organizations
Download fromView all 2 sources
CORE (RIOXX-UK Aggregator)
Part of book or chapter of book . Conference object . 2017
License: cc0