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

Neural network controllers have become popular in control tasks thanks totheir flexibility and expressivity. Stability is a crucial property forsafety-critical dynamical systems, while stabilization of partially observedsystems, in many cases, requires controllers to retain and process long-termmemories of the past. We consider the important class of recurrent neuralnetworks (RNN) as dynamic controllers for nonlinear uncertainpartially-observed systems, and derive convex stability conditions based onintegral quadratic constraints, S-lemma and sequential convexification. Toensure stability during the learning and control process, we propose aprojected policy gradient method that iteratively enforces the stabilityconditions in the reparametrized space taking advantage of mild additionalinformation on system dynamics. Numerical experiments show that our methodlearns stabilizing controllers while using fewer samples and achieving higherfinal performance compared with policy gradient.

Recurrent Neural Network Controllers Synthesis with Stability Guarantees for Partially Observed Systems