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

Stochastic gradient descent (SGD) is commonly used for optimization inlarge-scale machine learning problems. Langford et al. (2009) introduce asparse online learning method to induce sparsity via truncated gradient. Withhigh-dimensional sparse data, however, the method suffers from slow convergenceand high variance due to the heterogeneity in feature sparsity. To mitigatethis issue, we introduce a stabilized truncated stochastic gradient descentalgorithm. We employ a soft-thresholding scheme on the weight vector where theimposed shrinkage is adaptive to the amount of information available in eachfeature. The variability in the resulted sparse weight vector is furthercontrolled by stability selection integrated with the informative truncation.To facilitate better convergence, we adopt an annealing strategy on thetruncation rate, which leads to a balanced trade-off between exploration andexploitation in learning a sparse weight vector. Numerical experiments showthat our algorithm compares favorably with the original algorithm in terms ofprediction accuracy, achieved sparsity and stability.

Stabilized Sparse Online Learning for Sparse Data