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

The presented work addresses two-stage stochastic programs (2SPs), a broadlyapplicable model to capture optimization problems subject to uncertainparameters with adjustable decision variables. In case the adjustable orsecond-stage variables contain discrete decisions, the corresponding 2SPs areknown to be NP-complete. The standard approach of forming a single-stagedeterministic equivalent problem can be computationally challenging even forsmall instances, as the number of variables and constraints scales with thenumber of scenarios. To avoid forming a potentially huge MILP problem, we buildupon an approach of approximating the expected value of the second-stageproblem by a neural network (NN) and encoding the resulting NN into thefirst-stage problem. The proposed algorithm alternates between optimizing thefirst-stage variables and retraining the NN. We demonstrate the value of ourapproach with the example of computing operating points in power systems byshowing that the alternating approach provides improved first-stage decisionsand a tighter approximation between the expected objective and its neuralnetwork approximation.

Alternating mixed-integer programming and neural network training for approximating stochastic two-stage problems