Projection documentation #3232

Ivanidzo4ka · 2019-04-08T17:03:20Z

Towards #1209

wschin · 2019-04-08T17:14:27Z

docs/samples/Microsoft.ML.Samples/Dynamic/Transforms/NormalizeLpNorm.cs

+            // Convert training data to IDataView, the general data type used in ML.NET.
+            var data = mlContext.Data.LoadFromEnumerable(samples);
+            // NormalizeLpNorm normalize rows individually by rescaling them to unit norm.
+            // Performs the following operaion on a row X:  Y = (X - M) / D where M is mean, and D is selected norm.


What is the selected norm? Is it norm of the feature vector in a row being processed? Also, what are the shapes of X, Y, M, and D? #Pending

mean -> mean vector
D is selected norm -> D is calculated value of selected norm parameter
Does that sound better?

In reply to: 273152443 [](ancestors = 273152443)

Norm on what? A column? A row? Is it a scalar? In tensor computation, norm operation can produce another tensor.

Say, if I have rows, x_1, x_2, x_3. Is M=1/3 (x_1 + x_2 + x_3) true? Or M=ReduceSum(x_i) for the x subscripted by i? In addition, is D=||x_i||_2 for the x subscripted by i?

In reply to: 273153905 [](ancestors = 273153905,273152443)

Same here. Please move the final answer to what NormalizeLpNorm does' to the section for the estimator.

In reply to: 273181155 [](ancestors = 273181155,273153905,273152443)

wschin · 2019-04-08T17:21:09Z

docs/samples/Microsoft.ML.Samples/Dynamic/Transforms/ApproximatedKernelMap.cs

+{
+    public static class ApproximatedKernelMap
+    {
+        public static void Example()


Suggested change

public static void Example()

// Transform feature vector to another non-linear space. See https://people.eecs.berkeley.edu/~brecht/papers/07.rah.rec.nips.pdf.

public static void Example()

This transform is non-trivial, so some references are required. #Resolved

Ivanidzo4ka · 2019-04-08T18:01:22Z

docs/samples/Microsoft.ML.Samples/Dynamic/Transforms/NormalizeGlobalContrast.cs

+
+        private class DataPoint
+        {
+            [VectorType(7)]


7 [](start = 24, length = 1)

It shouldn't work! #Resolved

Ivanidzo4ka · 2019-04-08T18:01:31Z

docs/samples/Microsoft.ML.Samples/Dynamic/Transforms/NormalizeLpNorm.cs

+
+        private class DataPoint
+        {
+            [VectorType(7)]


it shouldn't work! #Resolved

artidoro · 2019-04-08T23:46:11Z

docs/samples/Microsoft.ML.Samples/Dynamic/Transforms/ApproximatedKernelMap.cs

+            //-0.0119, 0.5867, 0.4942,  0.7041
+            // 0.4720, 0.5639, 0.4346,  0.2671
+            //-0.2243, 0.7071, 0.7053, -0.1681
+            // 0.0846, 0.5836, 0.6575,  0.0581


Could you move these lines below the foreach loop and use:
// Expected output:

Could you do the same for the other files? #Resolved

codecov · 2019-04-09T19:45:32Z

Codecov Report

Merging #3232 into master will decrease coverage by <.01%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #3232      +/-   ##
==========================================
- Coverage   72.62%   72.62%   -0.01%     
==========================================
  Files         807      807              
  Lines      145080   145080              
  Branches    16213    16213              
==========================================
- Hits       105369   105365       -4     
- Misses      35294    35297       +3     
- Partials     4417     4418       +1

Flag	Coverage Δ
#Debug	`72.62% <ø> (-0.01%)`	⬇️
#production	`68.17% <ø> (-0.01%)`	⬇️
#test	`88.92% <ø> (-0.01%)`	⬇️

Impacted Files	Coverage Δ
src/Microsoft.ML.Transforms/NormalizerCatalog.cs	`84.78% <ø> (ø)`	⬆️
src/Microsoft.ML.Transforms/KernelCatalog.cs	`33.33% <ø> (ø)`	⬆️
...rosoft.ML.Transforms/FourierDistributionSampler.cs	`84.16% <ø> (ø)`	⬆️
...soft.ML.TestFramework/DataPipe/TestDataPipeBase.cs	`73.7% <0%> (-0.34%)`	⬇️
...StandardTrainers/Standard/LinearModelParameters.cs	`60.05% <0%> (-0.27%)`	⬇️

rogancarr · 2019-04-09T23:17:00Z

src/Microsoft.ML.Transforms/NormalizerCatalog.cs

@@ -279,7 +279,7 @@ internal static LpNormNormalizingEstimator NormalizeLpNorm(this TransformsCatalo
        /// <example>
        /// <format type="text/markdown">
        /// <![CDATA[
-        /// [!code-csharp[GlobalContrastNormalize](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/ProjectionTransforms.cs?range=1-6,12-112)]
+        /// [!code-csharp[GlobalContrastNormalize](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/Transforms/NormalizeGlobalContrast.cs)]


GlobalContrastNormalize [](start = 26, length = 23)

NormalizeGlobalContrast #Resolved

rogancarr · 2019-04-09T23:17:10Z

src/Microsoft.ML.Transforms/NormalizerCatalog.cs

@@ -249,7 +249,7 @@ public static class NormalizationCatalog
        /// <example>
        /// <format type="text/markdown">
        /// <![CDATA[
-        /// [!code-csharp[LpNormalize](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/ProjectionTransforms.cs?range=1-6,12-112)]
+        /// [!code-csharp[LpNormalize](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/Transforms/NormalizeLpNorm.cs)]


LpNormalize [](start = 26, length = 11)

NormalizeLpNorm #Resolved

rogancarr · 2019-04-09T23:18:35Z

docs/samples/Microsoft.ML.Samples/Dynamic/Transforms/NormalizeLpNorm.cs

+            // Performs the following operaion on a row X:  Y = (X - M(X)) / D(X) 
+            // where M(X) is scalar value of mean for current row,
+            // and D(X) is scalar value of selected `norm` parameter .
+            var approximation = mlContext.Transforms.NormalizeLpNorm("Features", norm: LpNormNormalizingEstimatorBase.NormFunction.L1, ensureZeroMean: true);


ensureZeroMean [](start = 135, length = 14)

What does EnsureZeroMean do? Subtract the mean? #Resolved

yes, added it to comment above.

In reply to: 273740225 [](ancestors = 273740225)

Let's move parameter details to xml docstring.

In reply to: 273741392 [](ancestors = 273741392,273740225)

rogancarr · 2019-04-09T23:18:56Z

src/Microsoft.ML.Transforms/KernelCatalog.cs

@@ -26,7 +26,7 @@ public static class KernelExpansionCatalog
        /// <example>
        /// <format type="text/markdown">
        /// <![CDATA[
-        /// [!code-csharp[CreateRandomFourierFeatures](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/ProjectionTransforms.cs?range=1-6,12-112)]
+        /// [!code-csharp[CreateRandomFourierFeatures](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/Transforms/ApproximatedKernelMap.cs)]


CreateRandomFourierFeatures [](start = 26, length = 27)

ApproximatedKernelMap #Resolved

rogancarr · 2019-04-09T23:19:51Z

docs/samples/Microsoft.ML.Samples/Dynamic/Transforms/ApproximatedKernelMap.cs

+            foreach (var row in column)
+                Console.WriteLine(string.Join(", ", row.Select(x => x.ToString("f4"))));
+            // Expected output:
+            // -0.0119, 0.5867, 0.4942,  0.7041


[](start = 14, length = 1)

Space Space #ByDesign

I prefer to align numbers, so one space was taken by minus sign.

In reply to: 273740487 [](ancestors = 273740487)

rogancarr · 2019-04-09T23:20:56Z

docs/samples/Microsoft.ML.Samples/Dynamic/Transforms/NormalizeGlobalContrast.cs

+            // NormalizeLpNorm normalize rows individually by rescaling them to unit norm.
+            // Performs the following operaion on a row X:  Y = scale *(X - M(X)) / D(X)
+            // where M(X) is scalar value of mean for current row,
+            // and D(X) is scalar value of either Standard deviation or L2 norm.


This comment looks like a copy/paste holdover. #Resolved

Can you come up with better one?

In reply to: 273740748 [](ancestors = 273740748)

shmoradims · 2019-04-12T14:36:37Z

docs/samples/Microsoft.ML.Samples/Dynamic/Transforms/NormalizeGlobalContrast.cs

+            };
+            // Convert training data to IDataView, the general data type used in ML.NET.
+            var data = mlContext.Data.LoadFromEnumerable(samples);
+            // NormalizeLpNorm normalize rows individually by rescaling them to unit norm.


NormalizeLpNorm [](start = 15, length = 15)

old name? #Resolved

shmoradims · 2019-04-12T14:37:04Z

docs/samples/Microsoft.ML.Samples/Dynamic/Transforms/NormalizeGlobalContrast.cs

+            };
+            // Convert training data to IDataView, the general data type used in ML.NET.
+            var data = mlContext.Data.LoadFromEnumerable(samples);
+            // NormalizeLpNorm normalize rows individually by rescaling them to unit norm.


normalize [](start = 31, length = 9)

normalizes #Resolved

shmoradims · 2019-04-12T14:38:28Z

docs/samples/Microsoft.ML.Samples/Dynamic/Transforms/NormalizeGlobalContrast.cs

+            // NormalizeLpNorm normalize rows individually by rescaling them to unit norm.
+            // Performs the following operaion on a row X:  Y = scale *(X - M(X)) / D(X)
+            // where M(X) is scalar value of mean for current row if ensureZeroMean = true or 0 othewise
+            // and D(X) is scalar value of either Standard deviation or L2 norm.


let's actually drop detailed algorithm descriptions inside examples. such details belong to the section and we don't want to repeat them again here. #Resolved

wschin · 2019-04-12T17:06:35Z

src/Microsoft.ML.Transforms/NormalizerCatalog.cs

@@ -276,10 +281,16 @@ internal static LpNormNormalizingEstimator NormalizeLpNorm(this TransformsCatalo
        /// <param name="ensureZeroMean">If <see langword="true"/>, subtract mean from each value before normalizing and use the raw input otherwise.</param>
        /// <param name="ensureUnitStandardDeviation">If <see langword="true"/>, resulted vector's standard deviation would be one. Otherwise, resulted vector's L2-norm would be one.</param>
        /// <param name="scale">Scale features by this value.</param>
+        /// <remarks>
+        /// This transform performs the following operation on a row X: Y = scale * (X - M(X)) / D(X)
+        /// where M(X) is scalar value of mean for current row if <paramref name="ensureZeroMean"/>set to <see langword="true"/> or <value>0</value> othewise


Suggested change

/// where M(X) is scalar value of mean for current row if <paramref name="ensureZeroMean"/>set to <see langword="true"/> or <value>0</value> othewise

/// where M(X) is scalar value of mean for all elements in the current row if <paramref name="ensureZeroMean"/>set to <see langword="true"/> or <value>0</value> othewise

``` #Resolved

wschin · 2019-04-12T17:07:18Z

src/Microsoft.ML.Transforms/NormalizerCatalog.cs

+        /// This transform performs the following operation on a row X: Y = scale * (X - M(X)) / D(X)
+        /// where M(X) is scalar value of mean for current row if <paramref name="ensureZeroMean"/>set to <see langword="true"/> or <value>0</value> othewise
+        /// D(X) is scalar value of standard deviation for row if <paramref name="ensureUnitStandardDeviation"/> set to <see langword="true"/> or
+        /// L2 norm value for this row if it set to <see langword="false"/> and scale is <paramref name="scale"/>.


Suggested change

/// L2 norm value for this row if it set to <see langword="false"/> and scale is <paramref name="scale"/>.

/// L2 norm of this row vector if <paramref name="ensureUnitStandardDeviation"/> set to <see langword="false"/>. "scale" is defined by <paramref name="scale"/>.

``` #Resolved

shmoradims

Ivan Matantsev added 2 commits April 8, 2019 10:02

Separate documentaion for projection transforms

4e39a92

remove old file

1701698

Ivanidzo4ka requested review from wschin, shmoradims and artidoro April 8, 2019 17:03

wschin reviewed Apr 8, 2019

View reviewed changes

Address comments

8badc77

Ivanidzo4ka commented Apr 8, 2019

View reviewed changes

artidoro reviewed Apr 8, 2019

View reviewed changes

Ivan Matantsev added 2 commits April 9, 2019 09:13

address comments

9b0a42a

Switch order of comments and console.writeline

71e83a0

shift output by one

70bb363

rogancarr reviewed Apr 9, 2019

View reviewed changes

Merge with master

dc92c57

rogancarr reviewed Apr 9, 2019

View reviewed changes

Ivan Matantsev added 2 commits April 9, 2019 16:22

ensure zero mean

8d29045

ApproximatedKernelMap

dfa6308

shmoradims reviewed Apr 12, 2019

View reviewed changes

Let's see can I make people happier?

f53218c

wschin approved these changes Apr 12, 2019

View reviewed changes

wschin reviewed Apr 12, 2019

View reviewed changes

update for Wschin comments

0355151

shmoradims approved these changes Apr 12, 2019

View reviewed changes

Ivanidzo4ka merged commit 9ca5a5a into dotnet:master Apr 12, 2019

Ivanidzo4ka mentioned this pull request Apr 15, 2019

Cherry pick Projection documentation and Normalize changes to 1.0 #3344

Closed

sfilipi mentioned this pull request Apr 16, 2019

API reference - Samples for Transforms #1209

Closed

ghost locked as resolved and limited conversation to collaborators Mar 23, 2022

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Projection documentation #3232

Projection documentation #3232

Ivanidzo4ka commented Apr 8, 2019

wschin Apr 8, 2019 •

edited

Loading

Ivanidzo4ka Apr 8, 2019

wschin Apr 8, 2019 •

edited

Loading

shmoradims Apr 12, 2019

wschin Apr 8, 2019 •

edited by Ivanidzo4ka

Loading

Ivanidzo4ka Apr 8, 2019 •

edited

Loading

Ivanidzo4ka Apr 8, 2019 •

edited

Loading

artidoro Apr 8, 2019 •

edited by Ivanidzo4ka

Loading

codecov bot commented Apr 9, 2019 •

edited

Loading

rogancarr Apr 9, 2019 •

edited by Ivanidzo4ka

Loading

rogancarr Apr 9, 2019 •

edited by Ivanidzo4ka

Loading

rogancarr Apr 9, 2019 •

edited by Ivanidzo4ka

Loading

Ivanidzo4ka Apr 9, 2019

shmoradims Apr 12, 2019

rogancarr Apr 9, 2019 •

edited by Ivanidzo4ka

Loading

rogancarr Apr 9, 2019 •

edited by Ivanidzo4ka

Loading

Ivanidzo4ka Apr 9, 2019

rogancarr Apr 9, 2019 •

edited by Ivanidzo4ka

Loading

Ivanidzo4ka Apr 9, 2019

shmoradims Apr 12, 2019 •

edited by Ivanidzo4ka

Loading

shmoradims Apr 12, 2019 •

edited by Ivanidzo4ka

Loading

shmoradims Apr 12, 2019 •

edited by Ivanidzo4ka

Loading

wschin Apr 12, 2019 •

edited by Ivanidzo4ka

Loading

wschin Apr 12, 2019 •

edited by Ivanidzo4ka

Loading

shmoradims left a comment

	public static void Example()
	// Transform feature vector to another non-linear space. See https://people.eecs.berkeley.edu/~brecht/papers/07.rah.rec.nips.pdf.
	public static void Example()

	/// where M(X) is scalar value of mean for current row if <paramref name="ensureZeroMean"/>set to <see langword="true"/> or <value>0</value> othewise
	/// where M(X) is scalar value of mean for all elements in the current row if <paramref name="ensureZeroMean"/>set to <see langword="true"/> or <value>0</value> othewise
	``` #Resolved

	/// L2 norm value for this row if it set to <see langword="false"/> and scale is <paramref name="scale"/>.
	/// L2 norm of this row vector if <paramref name="ensureUnitStandardDeviation"/> set to <see langword="false"/>. "scale" is defined by <paramref name="scale"/>.
	``` #Resolved

Projection documentation #3232

Projection documentation #3232

Conversation

Ivanidzo4ka commented Apr 8, 2019

wschin Apr 8, 2019 • edited Loading

Choose a reason for hiding this comment

Ivanidzo4ka Apr 8, 2019

Choose a reason for hiding this comment

wschin Apr 8, 2019 • edited Loading

Choose a reason for hiding this comment

shmoradims Apr 12, 2019

Choose a reason for hiding this comment

wschin Apr 8, 2019 • edited by Ivanidzo4ka Loading

Choose a reason for hiding this comment

Ivanidzo4ka Apr 8, 2019 • edited Loading

Choose a reason for hiding this comment

Ivanidzo4ka Apr 8, 2019 • edited Loading

Choose a reason for hiding this comment

artidoro Apr 8, 2019 • edited by Ivanidzo4ka Loading

Choose a reason for hiding this comment

codecov bot commented Apr 9, 2019 • edited Loading

Codecov Report

rogancarr Apr 9, 2019 • edited by Ivanidzo4ka Loading

Choose a reason for hiding this comment

rogancarr Apr 9, 2019 • edited by Ivanidzo4ka Loading

Choose a reason for hiding this comment

rogancarr Apr 9, 2019 • edited by Ivanidzo4ka Loading

Choose a reason for hiding this comment

Ivanidzo4ka Apr 9, 2019

Choose a reason for hiding this comment

shmoradims Apr 12, 2019

Choose a reason for hiding this comment

rogancarr Apr 9, 2019 • edited by Ivanidzo4ka Loading

Choose a reason for hiding this comment

rogancarr Apr 9, 2019 • edited by Ivanidzo4ka Loading

Choose a reason for hiding this comment

Ivanidzo4ka Apr 9, 2019

Choose a reason for hiding this comment

rogancarr Apr 9, 2019 • edited by Ivanidzo4ka Loading

Choose a reason for hiding this comment

Ivanidzo4ka Apr 9, 2019

Choose a reason for hiding this comment

shmoradims Apr 12, 2019 • edited by Ivanidzo4ka Loading

Choose a reason for hiding this comment

shmoradims Apr 12, 2019 • edited by Ivanidzo4ka Loading

Choose a reason for hiding this comment

shmoradims Apr 12, 2019 • edited by Ivanidzo4ka Loading

Choose a reason for hiding this comment

wschin Apr 12, 2019 • edited by Ivanidzo4ka Loading

Choose a reason for hiding this comment

wschin Apr 12, 2019 • edited by Ivanidzo4ka Loading

Choose a reason for hiding this comment

shmoradims left a comment

Choose a reason for hiding this comment

wschin Apr 8, 2019 •

edited

Loading

wschin Apr 8, 2019 •

edited

Loading

wschin Apr 8, 2019 •

edited by Ivanidzo4ka

Loading

Ivanidzo4ka Apr 8, 2019 •

edited

Loading

Ivanidzo4ka Apr 8, 2019 •

edited

Loading

artidoro Apr 8, 2019 •

edited by Ivanidzo4ka

Loading

codecov bot commented Apr 9, 2019 •

edited

Loading

rogancarr Apr 9, 2019 •

edited by Ivanidzo4ka

Loading

rogancarr Apr 9, 2019 •

edited by Ivanidzo4ka

Loading

rogancarr Apr 9, 2019 •

edited by Ivanidzo4ka

Loading

rogancarr Apr 9, 2019 •

edited by Ivanidzo4ka

Loading

rogancarr Apr 9, 2019 •

edited by Ivanidzo4ka

Loading

rogancarr Apr 9, 2019 •

edited by Ivanidzo4ka

Loading

shmoradims Apr 12, 2019 •

edited by Ivanidzo4ka

Loading

shmoradims Apr 12, 2019 •

edited by Ivanidzo4ka

Loading

shmoradims Apr 12, 2019 •

edited by Ivanidzo4ka

Loading

wschin Apr 12, 2019 •

edited by Ivanidzo4ka

Loading

wschin Apr 12, 2019 •

edited by Ivanidzo4ka

Loading