Extended contexts to regression and multiclass, added FFM pigstension #993

Zruty0 · 2018-09-22T01:50:35Z

Part of #754, extends #949.
Added MulticlassClassification context and RegressionContext, with corresponding Evaluate methods. Also added FFM binary trainer to the context extensions.

Zruty0 · 2018-09-22T01:52:41Z

src/Microsoft.ML.Data/Evaluators/MulticlassClassifierEvaluator.cs

+        /// <param name="label">The name of the label column in <paramref name="data"/>.</param>
+        /// <param name="score">The name of the score column in <paramref name="data"/>.</param>
+        /// <param name="predictedLabel">The name of the predicted label column in <paramref name="data"/>.</param>
+        /// <returns>The evaluation results for these uncalibrated outputs.</returns>


returns [](start = 13, length = 7)

remove #Closed

Zruty0 · 2018-09-22T01:54:43Z

test/Microsoft.ML.StaticPipelineTesting/Training.cs

 using Microsoft.ML.Runtime.Training;
 using System;
 using Xunit;
 using Xunit.Abstractions;
+using Microsoft.ML.Runtime.FactorizationMachine;


using [](start = 0, length = 5)

(placeholder) #Closed

sfilipi · 2018-09-22T03:35:06Z

src/Microsoft.ML.StandardLearners/FactorizationMachine/FactorizationMachineStatic.cs

+namespace Microsoft.ML.Trainers
+{
+    /// <summary>
+    /// Extension methods and utilities for instantiating FFM trainer estimators inside statically typed pipelines.


FFM trai [](start = 58, length = 8)

#Resolved

sfilipi · 2018-09-22T03:41:45Z

src/Microsoft.ML.Data/Evaluators/EvaluatorStaticExtensions.cs

+            var indexer = StaticPipeUtils.GetIndexer(data);
+            string labelName = indexer.Get(label(indexer.Indices));
+            (var scoreCol, var predCol) = pred(indexer.Indices);
+            Contracts.CheckParam(scoreCol != null, nameof(pred), "Indexing delegate resulted in null score column.");


Indexing delegate res [](start = 66, length = 21)

is it time to have a Contracts.CheckIndexingDelegate() :) #Resolved

I'm not sure it is :)

In reply to: 219660357 [](ancestors = 219660357)

sfilipi · 2018-09-22T03:42:42Z

src/Microsoft.ML.Data/Evaluators/EvaluatorStaticExtensions.cs

+        }
+
+        /// <summary>
+        /// Evaluates scored multiclass classification data.


multiclass [](start = 29, length = 10)

regression #Resolved

sfilipi · 2018-09-22T03:43:53Z

src/Microsoft.ML.Data/Evaluators/EvaluatorStaticExtensions.cs

+        /// <param name="data">The data to evaluate.</param>
+        /// <param name="label">The index delegate for the label column.</param>
+        /// <param name="score">The index delegate for predicted score column.</param>
+        /// <param name="loss">Potentially custom loss function. If left unspecified defaults to <see cref="SquaredLoss"/>.</param>


Potentially custom loss function [](start = 31, length = 32)

maybe just simpler: The loss function? #Resolved

sfilipi · 2018-09-22T04:02:43Z

src/Microsoft.ML.Data/Training/TrainContext.cs

+    public sealed class MulticlassClassificationContext : TrainContextBase
+    {
+        /// <summary>
+        /// For trainers for performing multiclass classification.


For trainers for performing multiclass classification. [](start = 12, length = 54)

clarify, maybe. #Resolved

sfilipi · 2018-09-22T04:12:57Z

src/Microsoft.ML.StandardLearners/FactorizationMachine/FactorizationMachineStatic.cs

+        /// the model that was trained.  Note that this action cannot change the result in any way; it is only a way for the caller to
+        /// be informed about what was learnt.</param>
+        /// <returns>The predicted output.</returns>
+        public static (Scalar<float> score, Scalar<bool> predictedLabel) FieldAwareFactorizationMachine(this BinaryClassificationContext.BinaryClassificationTrainers ctx,


FieldAwareFactorizationMachine [](start = 73, length = 30)

Shall we go with FAFM, if we are doing the acronym for SDCA? #Closed

If the choice is to go with acronyms, we should use FFM to follow the relevant papers (such as this one). #Resolved

Actually, I would prefer NOT to have acronyms, so fix SDCA instead

In reply to: 219684257 [](ancestors = 219684257)

sfilipi · 2018-09-22T04:23:39Z

src/Microsoft.ML.StandardLearners/FactorizationMachine/FactorizationMachineStatic.cs

+
+        private sealed class CustomReconciler : TrainerEstimatorReconciler
+        {
+            private static readonly string[] _fixedOutputNames = new[] { DefaultColumnNames.Score, DefaultColumnNames.PredictedLabel };


Score [](start = 92, length = 5)

is probability missing because FAFM doesn't calibrate? #Resolved

Yes

In reply to: 219661105 [](ancestors = 219661105)

sfilipi · 2018-09-22T04:26:44Z

test/Microsoft.ML.StaticPipelineTesting/Microsoft.ML.StaticPipelineTesting.csproj

@@ -7,6 +7,7 @@
    <ProjectReference Include="..\..\src\Microsoft.ML.ImageAnalytics\Microsoft.ML.ImageAnalytics.csproj" />
    <ProjectReference Include="..\..\src\Microsoft.ML.StandardLearners\Microsoft.ML.StandardLearners.csproj" />
    <ProjectReference Include="..\Microsoft.ML.TestFramework\Microsoft.ML.TestFramework.csproj" />
+    <NativeAssemblyReference Include="FactorizationMachineNative" />


NativeAssemblyReference [](start = 5, length = 23)

want to move it close to line 17? #Resolved

sfilipi

TomFinley · 2018-09-24T16:29:05Z

test/Microsoft.ML.StaticPipelineTesting/Microsoft.ML.StaticPipelineTesting.csproj

@@ -7,6 +7,7 @@
    <ProjectReference Include="..\..\src\Microsoft.ML.ImageAnalytics\Microsoft.ML.ImageAnalytics.csproj" />
    <ProjectReference Include="..\..\src\Microsoft.ML.StandardLearners\Microsoft.ML.StandardLearners.csproj" />
    <ProjectReference Include="..\Microsoft.ML.TestFramework\Microsoft.ML.TestFramework.csproj" />
+    <NativeAssemblyReference Include="FactorizationMachineNative" />


FactorizationMachineNative [](start = 38, length = 26)

Hmmm. Opportunistic fix for some other issue? #Resolved

No, it's that FFM requires FFMNative, and I added a test for it in static training

In reply to: 219902222 [](ancestors = 219902222)

TomFinley

Thanks @Zruty0 !

Extended contexts to regression and multiclass, added FFM pigstension

494e6a7

Zruty0 added the API Issues pertaining the friendly API label Sep 22, 2018

Zruty0 self-assigned this Sep 22, 2018

Zruty0 requested review from GalOshri, TomFinley and sfilipi September 22, 2018 01:50

Zruty0 commented Sep 22, 2018

View reviewed changes

Minor fixes

23c16bb

sfilipi reviewed Sep 22, 2018

View reviewed changes

sfilipi approved these changes Sep 22, 2018

View reviewed changes

TomFinley reviewed Sep 24, 2018

View reviewed changes

TomFinley approved these changes Sep 24, 2018

View reviewed changes

TomFinley merged commit eb26489 into dotnet:master Sep 24, 2018

TomFinley mentioned this pull request Jan 3, 2019

Load entry point models #1951

Merged

ghost locked as resolved and limited conversation to collaborators Mar 28, 2022

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Extended contexts to regression and multiclass, added FFM pigstension #993

Extended contexts to regression and multiclass, added FFM pigstension #993

Zruty0 commented Sep 22, 2018

Zruty0 Sep 22, 2018 •

edited

Loading

Zruty0 Sep 22, 2018 •

edited

Loading

sfilipi Sep 22, 2018 •

edited by Zruty0

Loading

sfilipi Sep 22, 2018 •

edited by Zruty0

Loading

Zruty0 Sep 24, 2018

sfilipi Sep 22, 2018 •

edited by Zruty0

Loading

sfilipi Sep 22, 2018 •

edited by Zruty0

Loading

sfilipi Sep 22, 2018 •

edited by Zruty0

Loading

sfilipi Sep 22, 2018 •

edited by Zruty0

Loading

GalOshri Sep 22, 2018 •

edited by Zruty0

Loading

Zruty0 Sep 24, 2018

sfilipi Sep 22, 2018 •

edited by Zruty0

Loading

Zruty0 Sep 24, 2018

sfilipi Sep 22, 2018 •

edited by Zruty0

Loading

sfilipi left a comment

TomFinley Sep 24, 2018 •

edited by Zruty0

Loading

Zruty0 Sep 24, 2018

TomFinley left a comment

Extended contexts to regression and multiclass, added FFM pigstension #993

Extended contexts to regression and multiclass, added FFM pigstension #993

Conversation

Zruty0 commented Sep 22, 2018

Zruty0 Sep 22, 2018 • edited Loading

Choose a reason for hiding this comment

Zruty0 Sep 22, 2018 • edited Loading

Choose a reason for hiding this comment

sfilipi Sep 22, 2018 • edited by Zruty0 Loading

Choose a reason for hiding this comment

sfilipi Sep 22, 2018 • edited by Zruty0 Loading

Choose a reason for hiding this comment

Zruty0 Sep 24, 2018

Choose a reason for hiding this comment

sfilipi Sep 22, 2018 • edited by Zruty0 Loading

Choose a reason for hiding this comment

sfilipi Sep 22, 2018 • edited by Zruty0 Loading

Choose a reason for hiding this comment

sfilipi Sep 22, 2018 • edited by Zruty0 Loading

Choose a reason for hiding this comment

sfilipi Sep 22, 2018 • edited by Zruty0 Loading

Choose a reason for hiding this comment

GalOshri Sep 22, 2018 • edited by Zruty0 Loading

Choose a reason for hiding this comment

Zruty0 Sep 24, 2018

Choose a reason for hiding this comment

sfilipi Sep 22, 2018 • edited by Zruty0 Loading

Choose a reason for hiding this comment

Zruty0 Sep 24, 2018

Choose a reason for hiding this comment

sfilipi Sep 22, 2018 • edited by Zruty0 Loading

Choose a reason for hiding this comment

sfilipi left a comment

Choose a reason for hiding this comment

TomFinley Sep 24, 2018 • edited by Zruty0 Loading

Choose a reason for hiding this comment

Zruty0 Sep 24, 2018

Choose a reason for hiding this comment

TomFinley left a comment

Choose a reason for hiding this comment

Zruty0 Sep 22, 2018 •

edited

Loading

Zruty0 Sep 22, 2018 •

edited

Loading

sfilipi Sep 22, 2018 •

edited by Zruty0

Loading

sfilipi Sep 22, 2018 •

edited by Zruty0

Loading

sfilipi Sep 22, 2018 •

edited by Zruty0

Loading

sfilipi Sep 22, 2018 •

edited by Zruty0

Loading

sfilipi Sep 22, 2018 •

edited by Zruty0

Loading

sfilipi Sep 22, 2018 •

edited by Zruty0

Loading

GalOshri Sep 22, 2018 •

edited by Zruty0

Loading

sfilipi Sep 22, 2018 •

edited by Zruty0

Loading

sfilipi Sep 22, 2018 •

edited by Zruty0

Loading

TomFinley Sep 24, 2018 •

edited by Zruty0

Loading