eduzhai > Applied Sciences > Engineering >

Interpretable Super-Resolution via a Learned Time-Series Representation

  • Save

... pages left unread,continue reading

Document pages: 19 pages

Abstract: We develop an interpretable and learnable Wigner-Ville distribution thatproduces a super-resolved quadratic signal representation for time-seriesanalysis. Our approach has two main hallmarks. First, it interpolates betweenknown time-frequency representations (TFRs) in that it can reachsuper-resolution with increased time and frequency resolution beyond what theHeisenberg uncertainty principle prescribes and thus beyond commonly employedTFRs, Second, it is interpretable thanks to an explicit low-dimensional andphysical parameterization of the Wigner-Ville distribution. We demonstrate thatour approach is able to learn highly adapted TFRs and is ready and able totackle various large-scale classification tasks, where we reachstate-of-the-art performance compared to baseline and learned TFRs.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×