From c4da115414d42712814b7b2733d761d9b03021ae Mon Sep 17 00:00:00 2001 From: Zhuyi Xue Date: Sat, 29 Oct 2016 06:31:01 +0000 Subject: [PATCH] improved comments --- scramble-experiment.ipynb | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/scramble-experiment.ipynb b/scramble-experiment.ipynb index b338fed..0103587 100644 --- a/scramble-experiment.ipynb +++ b/scramble-experiment.ipynb @@ -4,7 +4,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "This experiment shows that even after the pixels in hand-written digits images are scrambled, they can still be clustered effectively. It also shows that t-SNE takes no information about the translation/rotaional/illumination invariance in a image, which is very different from typical image recognition tasks." + "This experiment shows that even after the pixels in hand-written digits images are scrambled, they can still be clustered effectively by t-SNE. It means that t-SNE doesn't take information about the translation/rotaional/illumination invariance in an image for clustering, which is very different from typical image recognition tasks. t-SNE itself doesn't have much to do with images, either." ] }, { @@ -728,7 +728,7 @@ } ], "source": [ - "# Make sure columns in df_X_scrambled is scrambled\n", + "# Make sure columns in df_X_scrambled is indeed scrambled\n", "nrows, ncols = 2, 5\n", "plt.figure(figsize=(6,3))\n", "# plt.gray()\n",