eduzhai > Applied Sciences > Engineering >

dynoNet a neural network architecture for learning dynamical systems

  • Save

... pages left unread,continue reading

Document pages: 19 pages

Abstract: This paper introduces a network architecture, called dynoNet, utilizinglinear dynamical operators as elementary building blocks. Owing to thedynamical nature of these blocks, dynoNet networks are tailored for sequencemodeling and system identification purposes. The back-propagation behavior ofthe linear dynamical operator with respect to both its parameters and its inputsequence is defined. This enables end-to-end training of structured networkscontaining linear dynamical operators and other differentiable units,exploiting existing deep learning software. Examples show the effectiveness ofthe proposed approach on well-known system identification benchmarks.Examples show the effectiveness of the proposed approach against well-knownsystem identification benchmarks.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×