English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Restricted Boltzmann Machines (RBMs) are generative models which can learnuseful representations from samples of a dataset in an unsupervised fashion.They have been widely employed as an unsupervised pre-training method inmachine learning. RBMs have been modified to model time series in two mainways: The Temporal RBM stacks a number of RBMs laterally and introducestemporal dependencies between the hidden layer units; The Conditional RBM, onthe other hand, considers past samples of the dataset as a conditional bias andlearns a representation which takes these into account. Here we propose a newtraining method for both the TRBM and the CRBM, which enforces the dynamicstructure of temporal datasets. We do so by treating the temporal models asdenoising autoencoders, considering past frames of the dataset as corruptedversions of the present frame and minimizing the reconstruction error of thepresent data by the model. We call this approach Temporal Autoencoding. Thisleads to a significant improvement in the performance of both models in afilling-in-frames task across a number of datasets. The error reduction formotion capture data is 56\% for the CRBM and 80\% for the TRBM. Taking theposterior mean prediction instead of single samples further improves themodel's estimates, decreasing the error by as much as 91\% for the CRBM onmotion capture data. We also trained the model to perform forecasting on alarge number of datasets and have found TA pretraining to consistently improvethe performance of the forecasts. Furthermore, by looking at the predictionerror across time, we can see that this improvement reflects a betterrepresentation of the dynamics of the data as opposed to a bias towardsreconstructing the observed data on a short time scale.

Temporal Autoencoding Improves Generative Models of Time Series