40m QIL Cryo_Lab CTN SUS_Lab TCS_Lab OMC_Lab CRIME_Lab FEA ENG_Labs OptContFac Mariner WBEEShop
  40m Log  Not logged in ELOG logo
Entry  Sun Jul 22 14:01:07 2018, pooja, Update, Cameras, Developing neural networks on simulated video nn_simulation_2_normalized_mult_sin_nodes8_128epochs_lr0p00001_beta1_0p8_beta2_0p85_0p4train_0p1valid_marked.pdfnn_simulation_normalizedtarget_128epochs_mult_sin_load_wt_varyingtest_nodes8_lr0p00001_beta1_0p8_beta2_0p85_0p4train_0p1valid_marked.pdfnn_simulation_2_normalized_varying_mult_sin_nodes8_128epochs_lr0p00001_beta1_0p8_beta2_0p85_0p4train_0p1valid_marked.pdf
    Reply  Tue Jul 24 06:11:50 2018, rana, Update, Cameras, Developing neural networks on simulated video 
       Reply  Tue Jul 24 09:47:51 2018, gautam, Update, Cameras, Developing neural networks on simulated video 
Message ID: 14100     Entry time: Tue Jul 24 06:11:50 2018     In reply to: 14097     Reply to this: 14101
Author: rana 
Type: Update 
Category: Cameras 
Subject: Developing neural networks on simulated video 

This looks like good progress. Instead of fixed sines or random noise, you should generate now a time series for the motion which is random noise but with a power spectrum similar to what we see for the ETM pitch motion in lock. You can use inverse FFT to get the time series from the open loop OL spectra (being careful about edge effects).

Quote:

Aim: To develop a neural network that resolves mirror motion from video.

ELOG V3.1.3-