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

Hyperparameter optimization (HPO) is a vital step in improving performance indeep learning (DL). Practitioners are often faced with the trade-off betweenmultiple criteria, such as accuracy and latency. Given the high computationalneeds of DL and the growing demand for efficient HPO, the acceleration ofmulti-objective (MO) optimization becomes ever more important. Despite thesignificant body of work on meta-learning for HPO, existing methods areinapplicable to MO tree-structured Parzen estimator (MO-TPE), a simple yetpowerful MO-HPO algorithm. In this paper, we extend TPE's acquisition functionto the meta-learning setting using a task similarity defined by the overlap oftop domains between tasks. We also theoretically analyze and address thelimitations of our task similarity. In the experiments, we demonstrate that ourmethod speeds up MO-TPE on tabular HPO benchmarks and attains state-of-the-artperformance. Our method was also validated externally by winning the AutoML2022 competition on "Multiobjective Hyperparameter Optimization forTransformers".

Speeding Up Multi-Objective Hyperparameter Optimization by Task Similarity-Based Meta-Learning for the Tree-Structured Parzen Estimator