Infinite Variational Autoencoder for Semi-Supervised Learning
Authors:
- M. Ehsan AbbasnejadMAM. Ehsan Abbasnejad
- Anthony DickADAnthony Dick
- Anton van Den HengelAHAnton van Den Hengel
- 1 July 2017
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 10636919,p. 781-790
Abstract
This paper presents an infinite variational autoencoder (VAE) whose capacity adapts to suit the input data. This is achieved using a mixture model where the mixing coefficients are modeled by a Dirichlet process, allowing us to integrate over the coefficients when performing inference. Critically, this then allows us to automatically vary the number of autoencoders in the mixture based on the data. Experiments show the flexibility of our method, particularly for semi-supervised learning, where only a small number of training samples are available.Keywords
All Related Versions
This publication has 11 references indexed in Scilit:
- Semi-supervised Vocabulary-Informed Learning2016
- DeepPano: Deep Panoramic Representation for 3-D Shape RecognitionIEEE Signal Processing Letters, 2015
- VoxNet: A 3D Convolutional Neural Network for real-time object recognition2015
- Model recommendation: Generating object detectors from few samples2015
- Ensemble Projection for Semi-supervised Image Classification2013
- When semi-supervised learning meets ensemble learningFrontiers of Electrical and Electronic Engineering in China, 2011
- Semi-Supervised Random Forests2009
- Multi-label sparse coding for automatic image annotation2009
- SemiBoost: Boosting for Semi-Supervised LearningIEEE Transactions on Pattern Analysis and Machine Intelligence, 2008
- Correlated Label Propagation with Application to Multi-label Learning2006