Modeling Human Motion Using Binary Latent Variables

By Graham W. Taylor, Geoffrey E. Hinton, Sam T. Roweis. 2006

732 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Control and Systems Engineering

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 Control and Systems Engineering

Citations in the corpus

Differential Privacy Preservation for Deep Auto-Encoders: an Application of Human Behavior Prediction (2016, background): “The CRBM (Taylor, Hinton, and Roweis 2006) and SctRBM (Li et al. 2014; Phan et al. 2015) are competitive models.”
How to Represent Part-Whole Hierarchies in a Neural Network (2021, methodology): “When neural networks are used to represent shape, they generally work much better if they represent the details of a shape relative to its intrinsic coordinate frame rather than relative to a frame based on the camera or the world [Taylor et al., 2007, Deng et al., 2020].”
Deep Learning: Methods and Applications (2014, methodology): “Further, the use of a temporally conditional DBN for video sequence and human motion synthesis were reported in [361].”
Structural-RNN: Deep Learning on Spatio-Temporal Graphs (2015, methodology): “Several approaches based on Gaussian processes [59, 63], Restricted Boltzmann Machines (RBMs) [57, 56, 51], and RNNs [14] have been proposed to model human motion.”
Learning Stochastic Recurrent Networks (2014, methodology): “We will also treat the hidden states of the generating models, h 1: T as random variables following a Dirac distribution with inﬁnite mass at a single point.”
Learning with Hierarchical-Deep Models (2013, background): “Index Terms—Deep networks, deep Boltzmann machines, hierarchical Bayesian models, one-shot learning Ç”
Temporal Generative Adversarial Nets with Singular Value Clipping (2016, background): “Temporal RBMs have also been proposed to explicitly capture the temporal correlations in videos [40, 38, 37].”
A tutorial survey of architectures, algorithms, and applications for deep learning (2014, methodology): “The use of a temporally conditional DBN for video sequence and human motion synthesis is reported in [85].”
Exploiting compositionality to explore a large space of model structures (2012, background): “We ﬁrst consider a human motion capture dataset (Hsu et al., 2005; Taylor et al., 2007) consisting of a person walking in a variety of styles.”
Variational Gaussian Process Dynamical Systems (2011, methodology): “We assess the performance using the cumulative error per joint in the scaled space defined in [14] and by the root mean square error in the angle space suggested by [15].”
Deep learning for cyber security intrusion detection: Approaches, datasets, and comparative study (2020, result): “The proposed system can reduce the complexity of training and execution time of the deep learning architecture compared to work proposed by Taylor et al. [72].”
Training restricted Boltzmann machines: An introduction (2014, methodology): “for small models and it was shown in this work that the reconstruction error, which has been suggested for monitoring the training progress (Bengio et al., 2007; Taylor et al., 2007), may further decrease in spite of a divergence of the likelihood and thus is not reliable.”
Neural conditional random fields (2010, methodology): “We used an extension of RBM for dealing with continuous variables that have Gaussian noise (In our implementation we consider a Gaussian noise with standard derivation 0.2) (Taylor et al., 2007).”
Sequence of the Most Informative Joints (SMIJ): A new representation for human skeletal action recognition (2012, methodology): “[13,14] proposed using conditional restricted Boltzman machines (CRBM) to model the temporal evolution of human actions.”
Aspect extraction for opinion mining with a deep convolutional neural network (2016, background): “Each visible neuron is assumed to be a sample from a normal distribution (Taylor et al., 2007).”
Modeling Deep Temporal Dependencies with Recurrent "Grammar Cells" (2014, methodology): “Chirps Performances of the PGP were compared with that of a standard RNN (trained with BPTT) and a CRBM (trained with contrastive divergence) [19] on a dataset containing chirps (sinusoidal waves that increase or decrease in frequency over time).”
Deep Belief Networks for phone recognition (2009, unknown): “D object recognition [16], information retrieval [17] [18], motion capture data modeling [19] [20], and machine transliteration [21].”
Leveraging Hierarchical Parametric Networks for Skeletal Joints Based Action Segmentation and Recognition (2014, methodology): “Our approach can be seen as an extension to [26] in that instead of using the conditional Restricted Boltzmann Machine, a type of shallow model, to model human motion, we add layers to learn higher level features justified by a variational bound [10]; for modeling temporal information, rather than…”
Factored conditional restricted Boltzmann Machines for modeling motion style (2009, methodology): “As in (Taylor et al., 2007), the root segment was expressed in a body-centred coordinate system which is invariant to ground-plane translations and rotations about the gravitational vertical.”