Open blasscoc opened 5 years ago
@blasscoc, did Siddhant run these tiles through scatnet.m and get these coefficients?
Tommy, should be getting the test data put into the GitLab project today. The 2D data is shown on slide 10 of this presentation: http://qitech.biz/tech_papers/presentation_aps_dpp_18a.pdf and the 1D data is shown on slide 17 of this presentation: http://qitech.biz/tech_papers/presentation_aps_dpp_HEDSA_v1.pdf (PDF) https://www.dropbox.com/s/dkzqph5uwzfxgp6/presentation_aps_dpp_HEDSA_v1.mp4 (video)
You can find the work that I did with the MST, code and data at: https://www.dropbox.com/s/ot79hpf5x6uy5ry/mst.tgz (code) https://www.dropbox.com/s/0kfu1tmg3tuhahj/matlab_data.zip (data) The key code files are: mst/snl_scatnet/genGausFamily.m (generates the data) mst/snl_scatnet/runMallat_onSimp_meg.m (does the calculation) This presentation that has a lot of the results from this analysis is: http://qitech.biz/tech_papers/presentation_aps_dpp_17.pdf (PDF) https://www.dropbox.com/s/hpwbqe951ettn1j/presentation_aps_dpp_17.mp4 (video)
I have written a more comprehensive report on the 3 year research project. I am awaiting the formal release and some edits from Will. Will send you the link this week, when the report is finalized. It includes a better display of the 2D data.
Tommy needs to be getting MATLAB plots and MAT data of 2D transforms of 3 Gorgon simulations and 2 experimental data.
Ben should put data for the 3 Gorgon simulations on Github.
Done. 1) Replaced previous images in blusky/datasets/2D_maglif/ 2) Added rose plots and scattering coefficient data to blusky/datasets/2D_maglif/rose_plots
With the MATLAB scatnet.m, go take the tiles in the test dataset directory and run the 2-d morlet and gabor wavelets on them, output the result for those two functions. Then run the scattering transform to second order an output the coefficients.
We'll test the two wavelets and the scattering transform end-to-end with the result in 2-d.