Generative versus discriminative training of RBMs for classification of fMRI images

By Tanya Schmah, Geoffrey E. Hinton, Richard S. Zemel, S. Small, Stephen C. Strother. 2008

83 citations recorded by the source corpus.

Subfields: Generative Models, Representation & Self-Supervised Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Generative Models

1. Yoshua Bengio
2. Aaron C. Courville
3. Stefano Ermon
4. Pascal Vincent
5. Iain Murray
6. Diederik P. Kingma
7. Max Welling
8. Asja Fischer

In Representation & Self-Supervised Learning

1. Yoshua Bengio
2. Pascal Vincent
3. Aaron C. Courville
4. Hugo Larochelle
5. Asja Fischer
6. Razvan Pascanu
7. Yann LeCun
8. Kyunghyun Cho

Citations in the corpus

• MADE: Masked Autoencoder for Distribution Estimation (2015, methodology): “Thus a good distribution estimator can be used in many scenarios, including classification (Schmah et al., 2009), denoising or missing input imputation (Poon & Domingos, 2011; Dinh et al., 2014), data (e.g. speech) synthesis (Uria et al., 2015) and many others.”
• Training Restricted Boltzmann Machines (2015, methodology): “Furthermore, other techniques for using RBMs as classifiers exist, where several RBMs are trained, each modeling the input data from one class (Schmah et al., 2009).”
• The Neural Autoregressive Distribution Estimator (2011, background): “However, in this particular case the relative partition functions of the RBMs have successfully been approximated to build a classifier (Hinton, 2002; Schmah et al., 2009).”
• Learning Algorithms for the Classification Restricted Boltzmann Machine (2012, methodology): “Schmah et al. (2009) proposed a different approach to discriminative training of RBMs, where each class is associated with its own individual RBM, as in a Bayes classifier.”
• Capturing Global and Local Dynamics for Human Action Recognition (2014, methodology): “[17] propose a method to discriminatively estimate the difference of log-partition functions of two RBMs.”
• A generative restricted Boltzmann machine based method for high-dimensional motion data modeling (2015, methodology): “[30] propose a method to discriminatively estimate the difference of log-partition functions of two RBMs:”
• Unsupervised Feature Learning and Deep Learning: A Review and New Perspectives (2012, methodology): “…of applications, including fMRI image classification (Schmah et al. , 2009), motion and spatial transformations (Taylor and Hinton, 2009; Memisevic and Hinton, 2010), collaborative filtering (Salakhutdinov et al. , 2007) and natural image modeling (Ranzato and Hinton, 2010; Courville et al. , 2011b).”
• Training restricted Boltzmann machines: An introduction (2014, background): “1 Introduction In the last years, models extending or borrowing concepts from restricted Boltzmann machines (RBMs, [49]) have enjoyed much popularity for pattern analysis and generation, with applications including image classification, processing, and generation [25, 51, 34, 28, 48, 31]; learning m”
• Stacked Autoencoders for Unsupervised Feature Learning and Multiple Organ Detection in a Pilot Study Using 4D Patient Data (2013, methodology): “Deep neural networks for classification of fMRI brain images were studied in [26], where RBMs were used to classify the stage and action of a volume while the images were taken.”