Skip to content
Browse files
Implementing the t-SNE algorithm (#117)
* Added "Hbeta" function for PMTSNE with test

It is named as `entropyOf: andPRow: withBeta:`.
Added a test for the method ensuring:
- Original pVector reference is maintained
- entropy is calculated correctly
- pVector is calculated correctly

* Added `reduceXToInputDims:` method for PMTSNE

- Added corresponding test for new method.
- Renamed `ndims` to `outputDims`
- Renamed `initialdims` to `initialDims`
- Added accessor `x`

* Added method to get pairwise affinities in PMTSNE

- The new function is called `computePairwiseAffinities` (as in original paper)
- The old one is renamed to `computeLowDimentionalAffinities`
- Removed x2p function (translated it's code to `computePairwiseAffinities`)
- Also added tests

* Added progress bar(Job) for PMTSNE

Moved contents of the `start` method to `initializeJob`.

* Added gradient descent in PMTSNE

- Also added various accessors and instance vars
- Classified existing methods
- Added a test for `computePValues`
  • Loading branch information
AtharvaKhare authored and SergeStinckwich committed May 26, 2019
1 parent e63cc43 commit 7f3a8121300b90b97956931d7e0ac4294001a66e
Showing with 415 additions and 95 deletions.
  1. +356 −95 src/Math-TSNE/
  2. +59 −0 src/Math-Tests-TSNE/

0 comments on commit 7f3a812

Please sign in to comment.