Two Distributed-State Models For Generating High-Dimensional Time Series

By Graham W. Taylor, Geoffrey E. Hinton, Sam T. Roweis. 2011

125 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Statistical and Nonlinear Physics

How researchers see this paper

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

In Computer Vision and Pattern Recognition

In Statistical and Nonlinear Physics

Citations in the corpus

Online Contrastive Divergence with Generative Replay: Experience Replay without Storing Data (2016, background): “} Functionally, the visible layer encodes the data, while the hidden layer increases the learning capacity of the RBM model by enlarging the class of distributions that can be represented to an arbitrary complexity (Taylor et al, 2011).”
Deep learning for objective quality assessment of 3D images (2014, methodology): “Different variants of them were used with success in different research areas such for modeling human motion styles [15], for automatically mapping reinforcement learning tasks [16], and pattern recognition [17].”
Building energy load forecasting using Deep Neural Networks (2016, methodology): “In previous work for load forecasting using deep learning, authors of [6], explore Conditional Restricted Boltzmann Machines (CRBM) [19] and Factored Conditional Restricted Boltzmann Machines (FCBRM) [20] for building level load forecasting.”
Analyzing Drum Patterns Using Conditional Deep Belief Networks (2012, background): “In addition, the generative nature of DBNs makes them especially well-suited for stochastic generation of images or sequences [5, 11].”
Modeling Deep Temporal Dependencies with Recurrent "Grammar Cells" (2014, methodology): “This suggests modeling x (2) using a feature learning model whose parameters are a function of x (1) [14], giving rise to bi-linear models of transformations , such as the Gated Boltzmann Machine [15] [3], Gated Autoencoder [16], and similar models (see [14] for an overview).”
Learning Human Identity From Motion Patterns (2015, background): “Deep learning techniques, which achieved early success modeling sequential data such as motion capture [31] and video [16] have shown promise in multi-modal feature learning [22], [30], [15], [21].”
Video (language) modeling: a baseline for generative models of natural videos (2014, methodology): “Bilinear models of transformations between nearby frames have also been investigated by Memisevic & Hinton (2009); Sutskever et al. (2009); Taylor et al. (2011); Michalski et al. (2014) as well as Miao & Rao (2007) via Lie group theory.”
A survey of deep neural network architectures and their applications (2017, background): “In [153], an undirected model is established with real-valued visible variables and binary latent ones.”
Deep neural networks for energy load forecasting (2017, unknown): “In previous work for load forecasting using deep learning, authors of [2], performed building level load forecasting using Conditional Restricted Boltzmann Machines (CRBM) [20] and Factored Conditional Restricted Boltzmann Machines (FCBRM) [21].”
Automatically Mapped Transfer between Reinforcement Learning Tasks via Three-Way Restricted Boltzmann Machines (2013, background): “These models are stochastic with stochastic nodes and layers, making them less vulnerable to local minima [15].”
Multi-distribution deep belief network for speech synthesis (2013, background): “Previously, DBN has been shown to learn very good generative and discriminative models on high-dimensional data such as handwritten digits [4],facial pictures [5], human motion [6] and large vocabulary speech recognition [7] .”
An automated measure of MDP similarity for transfer in reinforcement learning (2014, background): “These features can then be tested on the target task to assess MDP similarity.”
Using Conceptors to Manage Neural Long-Term Memories for Temporal Patterns (2017, unknown): “The mocap-trained human motion generation system presented by Taylor et al. (2011), which is based on restricted Boltzmann machines, comes close to the conceptor-based demo with respect to number of patterns and body model complexity, and has visually more appealing transitions between gaits.”