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

An automatic machine learning (AutoML) task is to select the best algorithmand its hyper-parameters simultaneously. Previously, the hyper-parameters ofall algorithms are joint as a single search space, which is not only huge butalso redundant, because many dimensions of hyper-parameters are irrelevant withthe selected algorithms. In this paper, we propose a cascaded approach foralgorithm selection and hyper-parameter optimization. While a search procedureis employed at the level of hyper-parameter optimization, a bandit strategyruns at the level of algorithm selection to allocate the budget based on thesearch feedbacks. Since the bandit is required to select the algorithm with themaximum performance, instead of the average performance, we thus propose theextreme-region upper confidence bound (ER-UCB) strategy, which focuses on theextreme region of the underlying feedback distribution. We show theoreticallythat the ER-UCB has a regret upper bound $O\left(K \ln n\right)$ withindependent feedbacks, which is as efficient as the classical UCB bandit. Wealso conduct experiments on a synthetic problem as well as a set of AutoMLtasks. The results verify the effectiveness of the proposed method.

Cascaded Algorithm-Selection and Hyper-Parameter Optimization with Extreme-Region Upper Confidence Bound Bandit