correct std dev

session-1/session-1.ipynb

@@ -353,9 +353,9 @@
     "<a name=\"instructions-2\"></a>\n",
     "## Instructions\n",
     "\n",
-    "Now use tensorflow to calculate the standard deviation and upload the standard deviation image averaged across color channels as a \"jet\" heatmap of the 100 images.  This will be a little more involved as there is no operation in tensorflow to do this for you.  However, you can do this by calculating the mean image of your dataset as a 4-D array.  To do this, you could write e.g.  `mean_img_4d = tf.reduce_mean(imgs, reduction_indices=0, keep_dims=True)` to give you a `1 x H x W x C` dimension array calculated on the `N x H x W x C` images variable.  The reduction_indices parameter is saying to calculate the mean over the 0th dimension, meaning for every possible `H`, `W`, `C`, or for every pixel, you will have a mean composed over the `N` possible values it could have had, or what that pixel was for every possible image.  This way, you can write `images - mean_img_4d` to give you a `N x H x W x C` dimension variable, with every image in your images array having been subtracted by the `mean_img_4d`.  If you calculate the square root of the sum of the squared differences of this resulting operation, you have your standard deviation!\n",
+    "Now use tensorflow to calculate the standard deviation and upload the standard deviation image averaged across color channels as a \"jet\" heatmap of the 100 images.  This will be a little more involved as there is no operation in tensorflow to do this for you.  However, you can do this by calculating the mean image of your dataset as a 4-D array.  To do this, you could write e.g.  `mean_img_4d = tf.reduce_mean(imgs, reduction_indices=0, keep_dims=True)` to give you a `1 x H x W x C` dimension array calculated on the `N x H x W x C` images variable.  The reduction_indices parameter is saying to calculate the mean over the 0th dimension, meaning for every possible `H`, `W`, `C`, or for every pixel, you will have a mean composed over the `N` possible values it could have had, or what that pixel was for every possible image.  This way, you can write `images - mean_img_4d` to give you a `N x H x W x C` dimension variable, with every image in your images array having been subtracted by the `mean_img_4d`.  If you calculate the square root of the expected squared differences of this resulting operation, you have your standard deviation!\n",
     "\n",
-    "In summary, you'll need to write something like: `subtraction = imgs - tf.reduce_mean(imgs, reduction_indices=0, keep_dims=True)`, then reduce this operation using `tf.sqrt(tf.reduce_sum(subtraction * subtraction, reduction_indices=0))` to get your standard deviation then include this image in your zip file as <b>std.png</b>\n",
+    "In summary, you'll need to write something like: `subtraction = imgs - tf.reduce_mean(imgs, reduction_indices=0, keep_dims=True)`, then reduce this operation using `tf.sqrt(tf.reduce_mean(subtraction * subtraction, reduction_indices=0))` to get your standard deviation then include this image in your zip file as <b>std.png</b>\n",
     "\n",
     "<a name=\"code-2\"></a>\n",
     "## Code\n",
@@ -389,8 +389,8 @@
     "subtraction = imgs - mean_img_4d\n",
     "\n",
     "# Now compute the standard deviation by calculating the\n",
-    "# square root of the sum of squared differences\n",
-    "std_img_op = tf.sqrt(tf.reduce_sum(subtraction * subtraction, reduction_indices=0))\n",
+    "# square root of the expected squared differences\n",
+    "std_img_op = tf.sqrt(tf.reduce_mean(subtraction * subtraction, reduction_indices=0))\n",
     "\n",
     "# Now calculate the standard deviation using your session\n",
     "std_img = sess.run(std_img_op)"