Commits · 72bf0bf3831495cbe59fff4e8f4c62024a2176f0 · Jon Lachmann / sgd

Browse files

May 25, 2015
- temporary fix for implicit update · 72bf0bf3
  Tian Lan authored May 25, 2015
  
  72bf0bf3
May 24, 2015
- change structure for experiment · 545cf0ee
  Tian Lan authored May 24, 2015
  
  545cf0ee
- add plotting for mse; add plotting for error rate · ee4271f1
  Tian Lan authored May 24, 2015
  
  ee4271f1
May 23, 2015
- change file structure · 7be646dc
  Tian Lan authored May 23, 2015
  
  7be646dc
- mse plot: add support for multiple sgd objects · e7b2fddf
  Tian Lan authored May 23, 2015
  
  e7b2fddf
- add npasses to sgd.control for multiple passes · c6ed4f24
  Tian Lan authored May 22, 2015
  
  c6ed4f24
May 22, 2015
- change epsilon arg to delta · 7f9e4ec1
  Dustin Tran authored May 22, 2015
  
  7f9e4ec1
- add test for sgd update · 8e609db7
  Tian Lan authored May 22, 2015
  
  8e609db7
- bug fix for glm: df, aic · 06c56ed7
  Tian Lan authored May 21, 2015
  
  06c56ed7
May 21, 2015
- follow function naming convention · 257748ac
  Dustin Tran authored May 21, 2015
  
  257748ac
- add log-log scale to mse plot · 3b61fdcb
  Dustin Tran authored May 20, 2015
  
  3b61fdcb
- add rmsprop · 892ec9b8
  Dustin Tran authored May 20, 2015
  
  892ec9b8
May 18, 2015
- add wmatrix to ee model arguments · 6a17c24d
  Dustin Tran authored May 18, 2015
  
  6a17c24d
- change interface to be more readaable · 642465f2
  Dustin Tran authored May 18, 2015
  
  642465f2
- add different gmm types to interface · b0c30a4d
  Dustin Tran authored May 17, 2015
  
  b0c30a4d
- overhaul base code for extending beyond glms · f52712a8
  Dustin Tran authored May 17, 2015
  
  f52712a8
May 17, 2015
- fix a bug in storing estimates. add mse plotting for glm · 389d645a
  Tian Lan authored May 16, 2015
  
  389d645a
May 16, 2015
- store less estimated coefficients. change sgd update step to adapt to current data structure. · 91ec8be6
  Tian Lan authored May 15, 2015
  
  91ec8be6
May 15, 2015
- update doc · bf803789
  Dustin Tran authored May 15, 2015
  
  bf803789
- move model specific output from C to model_out. move model specific post... · 7118b959
  Tian Lan authored May 15, 2015
  7118b959
- rename family and transfer.h to be glm-specific · 287a777a
  Dustin Tran authored May 15, 2015
  
  287a777a
- add all methods and lr combinations to unit test · 1c6b776d
  Dustin Tran authored May 15, 2015
  
  1c6b776d
- add benchmark code for comparing runtime · 4cd164e7
  Dustin Tran authored May 15, 2015
  
  4cd164e7
- add unit test for mse · 991b10a4
  Tian Lan authored May 15, 2015
  
  991b10a4
May 13, 2015
- store less; set Idiag to vector instead of diagonal matrix; #38 · 26eba14a
  Dustin Tran authored May 13, 2015
  
  26eba14a
- add rank as a flag for glm model · af737ea9
  Tian Lan authored May 13, 2015
  
  af737ea9
- fix a bug for d-dim learning rate · 630e98d8
  Tian Lan authored May 13, 2015
  
  630e98d8
- removed rank check. initialize one-dim learning rate lr_alpha with 1. fixed performance issue · 05534291
  Tian Lan authored May 13, 2015
  
  05534291
May 12, 2015
- add documentation for ai-sgd · e7b35676
  Dustin Tran authored May 11, 2015
  
  e7b35676
May 11, 2015
- temporary fix to one-dim; #41 · 84b00821
  Dustin Tran authored May 11, 2015
  
  84b00821
- fix #define DEBUG 1 so that it works · 8c801405
  Dustin Tran authored May 11, 2015
  
  8c801405
- reformat curly braces · 582f0d0d
  Dustin Tran authored May 11, 2015
  
  582f0d0d
- rename fitting-glm.R ->glm-poisson-regression.R · 91efa598
  Dustin Tran authored May 11, 2015
  
  91efa598
- reuse old score_function() code · 6c00daec
  Dustin Tran authored May 11, 2015
  
  6c00daec
- rename src_*.h files · 5e658228
  Dustin Tran authored May 11, 2015
  
  5e658228
- add c=1 for non-averaging in one-dim learning rate; update documentation · b30d1723
  Dustin Tran authored May 10, 2015
  
  b30d1723
- update documentation on learning rates · f15f788b
  Dustin Tran authored May 10, 2015
  
  f15f788b
- flip alpha; #50 · fed55797
  Dustin Tran authored May 10, 2015
  
  fed55797
- speed up learning rate calculation; remove d-dim weighted · 12dafcc4
  Dustin Tran authored May 10, 2015
  
  12dafcc4
- generalize the d-dim learning rate; #50 · dac45c0a
  Dustin Tran authored May 10, 2015
  
  dac45c0a