Commits · 6380205cea3b3d29fa6b7df3d9ca988b1050d9c0 · Eric.Schanet / KerasROOTClassification · GitLab

Snippets Groups Projects

Dec 05, 2018
- make adv learning rate tunable · 6380205c
  Nikolai.Hartmann authored 6 years ago
  
  6380205c
Dec 04, 2018
- keyboard interrupt for adversarial training · 6c1b3d00
  Nikolai.Hartmann authored 6 years ago
  
  6c1b3d00
- revert adam optimizer for adversarial to default lr · b049f3d7
  Nikolai.Hartmann authored 6 years ago
  
  b049f3d7
Nov 30, 2018
- apply class weights for adversarial targets · 53ba3604
  Nikolai.Hartmann authored 6 years ago
  
  53ba3604
- adding class_weight_target property · 0bd404e7
  Nikolai.Hartmann authored 6 years ago
  
  0bd404e7
- allow arbitrary binning for decorr dist · 1e44f56c
  Nikolai.Hartmann authored 6 years ago
  
  1e44f56c
- number of decorr_bins as a parameter · 7853cb06
  Nikolai.Hartmann authored 6 years ago
  
  7853cb06
Nov 29, 2018
- fixing loss plot · a741cd01
  Nikolai.Hartmann authored 6 years ago
  
  a741cd01
- adding BaseLogger and History callbacks · 0a377f2a
  Nikolai.Hartmann authored 6 years ago
  
  0a377f2a
- ensure fixed number of labels for adv target · 37280c34
  Nikolai Hartmann authored 6 years ago
  
  37280c34
- name target layers in adversarial setup · 7c1e95cb
  Nikolai Hartmann authored 6 years ago
  
  7c1e95cb
- use keras callbacks in adversarial training · 34783571
  Nikolai Hartmann authored 6 years ago
  
  34783571
- some fixes · 9a1b3b5e
  Nikolai.Hartmann authored 6 years ago
  
  9a1b3b5e
Nov 28, 2018
- Adversarial setup proof-of-concept · 24e51c50
  Nikolai.Hartmann authored 6 years ago
  
  24e51c50
- weights and layers · db706fce
  Nikolai.Hartmann authored 6 years ago
  
  db706fce
- starting adversarial setup for decorrelation · 80673f0e
  Nikolai Hartmann authored 6 years ago
  
  80673f0e
Nov 20, 2018
- add another hidden layer on top of regression targets · be2eda46
  Nikolai Hartmann authored 6 years ago
  
  be2eda46
- adjust compare.py · 2267c3cf
  Nikolai Hartmann authored 6 years ago
  
  2267c3cf
- fixing evaluate_train_test for regression targets · 9ed6b0a7
  Nikolai Hartmann authored 6 years ago
  
  9ed6b0a7
- remove duplicate function · 28f14b38
  Nikolai Hartmann authored 6 years ago
  
  28f14b38
- transform target for regression · 3b5bb93e
  Nikolai Hartmann authored 6 years ago
  
  3b5bb93e
Nov 19, 2018
- weight individual losses · 3c3d6c50
  Nikolai.Hartmann authored 6 years ago
  
  3c3d6c50
- use mse by default for regression · e67e13f2
  Nikolai.Hartmann authored 6 years ago
  
  e67e13f2
- add regression targets to models · e4212c7c
  Nikolai.Hartmann authored 6 years ago
  
  e4212c7c
- introduce l_train/test to indicate the labels in case of output with regression targets · 3e73877f
  Nikolai.Hartmann authored 6 years ago
  
  3e73877f
- fill regression targets optionally · 4ed198f6
  Nikolai Hartmann authored 6 years ago
  
  4ed198f6
Nov 01, 2018
- fixing evaluate_train_test (includes transformation now) · dca187ae
  Nikolai Hartmann authored 6 years ago
  
  dca187ae
Oct 22, 2018
- some comments · f2b50881
  Nikolai Hartmann authored 6 years ago
  
  f2b50881
- Do transformations only in batch generator - not globally · 9dbc0793
  Nikolai Hartmann authored 6 years ago
  
  9dbc0793
- scale and shift example data for tests randomly · 778a981d
  Nikolai Hartmann authored 6 years ago
  
  778a981d
- got rid of all global shuffling · 7d32696b
  Nikolai Hartmann authored 6 years ago
  
  7d32696b
- adding tests · 649d4d96
  Nikolai Hartmann authored 6 years ago
  
  649d4d96
Oct 21, 2018
- call _load_data in generate_actmax.py · e7a75351
  Nikolai.Hartmann authored 6 years ago
  
  e7a75351
Oct 17, 2018
- option to fix parameters when writing friends (useful for parametrized classifiers) · 17f96ce6
  Nikolai.Hartmann authored 6 years ago
  
  17f96ce6
Oct 05, 2018
- support for score_mode in write_friend_tree · 4f7d3f1f
  Nikolai.Hartmann authored 6 years ago
  
  4f7d3f1f
- write training weight in write_parametrized · 3dbe82d9
  Nikolai.Hartmann authored 6 years ago
  
  3dbe82d9
Sep 19, 2018
- info message about traning stats · d8abe157
  Nikolai.Hartmann authored 6 years ago
  
  d8abe157
- catch KeyError as well as IOError in _load_data · b2336145
  Nikolai.Hartmann authored 6 years ago
  
  b2336145
- adding script to write ROOT trees for parametrized learning · 9f5de8d3
  Nikolai.Hartmann authored 6 years ago
  
  9f5de8d3
- Switch to functional interface and make usage of bias vectors configurable · ab3785ad
  Nikolai.Hartmann authored 6 years ago
  
  ab3785ad