Plots for Conference about Miscellaneous topics of PPTS

deep neural networks with domain adaptation

Discussed at the PPTS meeting on 4th April 2016, contribution

PlotSorted ascending Description
caffe_vs_tmva.png.C .png .pdf A proof of principle that caffe trained deep neural networks do the job of signal/background separation, in the ballpark of what we're used to from TMVA. This shall demonstrate that the caffe test setup works in principle and we can assume trainings have converged and we can go on to the next step
adding_grl_to_network.png.C .png .pdf A by-eye comparison of networks with and without domain adaptation. The signal to background is hardly distinguishable, the network without domain adaptation is even a bit better.

-- SilviaBorghi - 2016-04-07

Topic attachments
I Attachment History Action Size Date Who Comment
C source code filec adding_grl_to_network.C r1 manage 48.8 K 2016-04-08 - 10:34 PaulSeyfert plots for deep learning with domain adaptation
PDFpdf adding_grl_to_network.pdf r1 manage 36.9 K 2016-04-08 - 10:34 PaulSeyfert plots for deep learning with domain adaptation
PNGpng adding_grl_to_network.png r1 manage 18.2 K 2016-04-08 - 10:34 PaulSeyfert plots for deep learning with domain adaptation
C source code filec caffe_vs_tmva.C r1 manage 49.0 K 2016-04-08 - 10:34 PaulSeyfert plots for deep learning with domain adaptation
PDFpdf caffe_vs_tmva.pdf r1 manage 36.8 K 2016-04-08 - 10:34 PaulSeyfert plots for deep learning with domain adaptation
PNGpng caffe_vs_tmva.png r1 manage 16.8 K 2016-04-08 - 10:34 PaulSeyfert plots for deep learning with domain adaptation
Edit | Attach | Watch | Print version | History: r2 < r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r2 - 2016-04-08 - PaulSeyfert
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    LHCb All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback