[scikit-learn/scikit-learn] 6c99d7: [MRG+1] Make TSNE trustworthiness test more robust...

Sept. 24, 2017
11:33 a.m.
Branch: refs/heads/master Home: https://github.com/scikit-learn/scikit-learn Commit: 6c99d797d7c71d216503612a242bffb8d006582d https://github.com/scikit-learn/scikit-learn/commit/6c99d797d7c71d216503612a... Author: Olivier Grisel <olivier.grisel@ensta.org> Date: 2017-09-24 (Sun, 24 Sep 2017) Changed paths: M sklearn/manifold/tests/test_t_sne.py Log Message: ----------- [MRG+1] Make TSNE trustworthiness test more robust (#9808) Platform specific rounding errors can make the t-SNE algorithm converge to varying quality results especially on small datasets as done in this test. We therefore need a lower threshold to account for that variability.
2707
Age (days ago)
2707
Last active (days ago)
0 comments
1 participants
participants (1)
-
GitHub