DeepTime: Studying deep-learning architectures for time series classification

Abstract

This Project aims to develop deep learning architectures to classify time series at scale. This Project is motivated by new-generation satellites that are now imaging our planet frequently, completely, in high-resolution, and at no charge to end-users. This introduces unprecedented opportunities to monitor the flux of our Planet s systems. However, we cannot yet make the most of these opportunities, as state-of-the-art time series classification techniques do not scale tohandle such wealth of data, and are not capable of leveraging the spatial and spectral dimensions of the data. Deep learning technologies have recently revolutionized the research field of machine learning, and particularly for data with spatial or temporal structure. This Project seeks to develop new deep learning technologies that can make the most of the temporal, spatial and spectral dimensions of series of satellite images. This Project’s outputs will be both applicable to general time series and to the satellite application.

Document Details

Document Type
DoD Grant Award
Publication Date
Aug 28, 2018
Source ID
FA23861814030

Entities

People

  • Francois Petitjean

Organizations

  • Air Force Office of Scientific Research
  • Monash University
  • United States Air Force

Tags

Fields of Study

  • Computer science

Readers

  • Neural Network Machine Learning.
  • Systems Analysis and Design

Technology Areas

  • AI & ML
  • AI & ML - Neural Networks
  • Space