Island Transpeciation: A Co-Evolutionary Neural Architecture Search, Applied to Country-Scale Air-Quality Forecasting

Authors:
Publisher website
Google Scholar
Abstract
Air pollution causes around 400,000 premature deaths per year in Europe due to Particulate Matter, nitrogen oxides and ground-level ozone pollutants. Multiple-Input Multiple-Output Nonlinear Auto-Regressive eXogenous Deep Neural Networks are frequently used to predict a day before, air-quality pollution incidents, at a country-scale. With complexity and data sizes increasing, finding performant models becomes harder. We propose island transpeciation to optimize hyperparameters and architectures. Unlike using a single optimizer, island transpeciation combines results from multiple optimizers, to consistently provide excellent performance. Moreover, we show that island transpeciation outperforms random model search and other previous modelling efforts. Island transpeciation is a Neural Architecture Search that uses co-evolution (genes), to combine (transpeciation) populations of incompatible optimizers (species) organized in island formations. In island transpeciation, architecture search is parallelized and utilizes a distributed pool of hardware resources. We have successfully used these techniques to predict next-day ozone concentrations across the Belgian territory.
Funding Information
  • Vlaamse regering (AI Research Program, Leuven.AI - KU Leuven institute for AI)
  • European Commission: European Research Council under the European Union?s Horizon 2020 research and innovation programme (ERC Adv. Grant grant agreement No 885682)
  • Research Council KUL: Optimization frameworks for deep kernel machines (C14/18/068)
  • KU Leuven: Research Fund (C16/15/059, C24/18/022, C3/19/053, C3/20/117, C3I-21-00316)
  • Flemish Government Agencies: VLAIO (Baekeland PhD (HBC.20192204)., CSBO (HBC.2021.0076))
  • Ford KU Leuven Research Alliance Project KUL0076
  • Flemish Government Agencies: EWI (the Flanders AI Research Program)
  • KU Leuven: Industrial Research Fund (13-0260, Development bilateral industrial projects, IOFm/16/004)
  • Flemish Government Agencies: FWO (Deep Restricted Kernel Machines GOA4917N, EOS Project no G0F6718N (SeLMA), GOA4917N (Deep Restricted Kernel Machines), Infrastructure project I013218N, PhD Grants (SB/1SA1319N), PhD/Postdoc grant, SBO project S005319N, TBM Project T001919N)
  • Foundation ?Kom op tegen Kanker?, CM
  • European Research Council under the European Union's Horizon 2020 Research and Innovation Program (ERC Advanced Grant E-DUALITY (787960))

This publication has 35 references indexed in Scilit: