* Representation Learning: A Review and New Perspectives - (Bengio) Reviews representation learning in probabalistic models (e.g. BMs), reconstruction approaches (e.g. autoencoders), and geometrically motivated manifold-learning approaches (e.g. PCA models a linear manifold)

* Causal Inference on Time Series using Restricted Structural Equation Models] - Not familiar enough with concepts

* What Are the Invariant Occlusive Components of Image Patches? A Probabilistic Generative Approach - Presents a model that learns features and masks of image patches to model occlusion

* Multi-Prediction Deep Boltzmann Machines - (Bengio) Frank paper that introduces a method of training DBMs in one sweep as apposed to layerwise that does not reduce performance and provides competitive classification ability

* Learning Multi-level Sparse Representations

* Adaptive Multi-Column Deep Neural Networks with Application to Robust Image Denoising

* Understanding Dropout

* On the Expressive Power of Restricted Boltzmann Machines

* Adaptive dropout for training deep neural networks

* Deep Generative Stochastic Networks Trainable by Backprop

Also, lets just pick a few talks from

International Conference on Learning Representations 2013

At a glance, these may be intersting

• The Neural Representation Benchmark and its Evaluation on Brain and Machine
• Efficient Learning of Domain-invariant Image Representations
• Deep Learning of Recursive Structure: Grammar Induction
• Feature Learning in Deep Neural Networks - A Study on Speech Recognition Tasks
• Joint Training Deep Boltzmann Machines for Classification
• What Regularized Auto-Encoders Learn from the Data Generating Distribution

-- AmosStorkey - 17 Dec 2013

List of papers that might have some use, haven't fully read and needs pruning:

* Stochastic Ratio Matching of RBMs for Sparse High-Dimensional Inputs - (Bengio) Applied to text classification, attempts to keep RBM computational training cost linear in the number of non-zeros

* Dropout Training as Adaptive Regularization

* Fast dropout training

* Sparse coding for multitask and transfer learning

* Smooth Sparse Coding via Marginal Regression for Learning Sparse Representations

* Simple Sparsification Improves Sparse Denoising Autoencoders in Denoising Highly Noisy Images

* On A Nonlinear Generalization of Sparse Coding and Dictionary Learning