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 Plus annual

Presents the access of premium content as premium feature

Premium Content

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 Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Unsupervised feature learning algorithms based on convolutional formulationsof independent components analysis (ICA) have been demonstrated to yieldstate-of-the-art results in several action recognition benchmarks. However,existing approaches do not allow for the number of latent components (features)to be automatically inferred from the data in an unsupervised manner. This is asignificant disadvantage of the state-of-the-art, as it results in considerableburden imposed on researchers and practitioners, who must resort to tediouscross-validation procedures to obtain the optimal number of latent features. Toresolve these issues, in this paper we introduce a convolutional nonparametricBayesian sparse ICA architecture for overcomplete feature learning fromhigh-dimensional data. Our method utilizes an Indian buffet process prior tofacilitate inference of the appropriate number of latent features under ahybrid variational inference algorithm, scalable to massive datasets. As weshow, our model can be naturally used to obtain deep unsupervised hierarchicalfeature extractors, by greedily stacking successive model layers, similar toexisting approaches. In addition, inference for this model is completelyheuristics-free; thus, it obviates the need of tedious parameter tuning, whichis a major challenge most deep learning approaches are faced with. We evaluateour method on several action recognition benchmarks, and exhibit its advantagesover the state-of-the-art.

A Nonparametric Bayesian Approach Toward Stacked Convolutional Independent Component Analysis