Replace all ML.Transforms SubComponent usages with IComponentFactory. (#700)

* Replace all ML.Transforms SubComponent usages with IComponentFactory.

Working towards #585

* PR feedback

Author: eerhardt

src/Microsoft.ML.Core/EntryPoints/ComponentFactory.cs

@@ -76,6 +76,15 @@ public static IComponentFactory<TArg1, TComponent> CreateFromFunction<TArg1, TCo
             return new SimpleComponentFactory<TArg1, TComponent>(factory);
         }
 
+        /// <summary>
+        /// Creates a component factory when we take two extra parameters (and an
+        /// <see cref="IHostEnvironment"/>) that simply wraps a delegate which creates the component.
+        /// </summary>
+        public static IComponentFactory<TArg1, TArg2, TComponent> CreateFromFunction<TArg1, TArg2, TComponent>(Func<IHostEnvironment, TArg1, TArg2, TComponent> factory)
+        {
+            return new SimpleComponentFactory<TArg1, TArg2, TComponent>(factory);
+        }
+
         /// <summary>
         /// Creates a component factory when we take three extra parameters (and an
         /// <see cref="IHostEnvironment"/>) that simply wraps a delegate which creates the component.
@@ -124,6 +133,26 @@ public TComponent CreateComponent(IHostEnvironment env, TArg1 argument1)
             }
         }
 
+        /// <summary>
+        /// A class for creating a component when we take one extra parameter
+        /// (and an <see cref="IHostEnvironment"/>) that simply wraps a delegate which
+        /// creates the component.
+        /// </summary>
+        private sealed class SimpleComponentFactory<TArg1, TArg2, TComponent> : IComponentFactory<TArg1, TArg2, TComponent>
+        {
+            private readonly Func<IHostEnvironment, TArg1, TArg2, TComponent> _factory;
+
+            public SimpleComponentFactory(Func<IHostEnvironment, TArg1, TArg2, TComponent> factory)
+            {
+                _factory = factory;
+            }
+
+            public TComponent CreateComponent(IHostEnvironment env, TArg1 argument1, TArg2 argument2)
+            {
+                return _factory(env, argument1, argument2);
+            }
+        }
+
         /// <summary>
         /// A class for creating a component when we take three extra parameters
         /// (and an <see cref="IHostEnvironment"/>) that simply wraps a delegate which

src/Microsoft.ML.Data/Commands/CrossValidationCommand.cs

@@ -554,7 +554,7 @@ private FoldResult RunFold(int fold)
                     }
 
                     // Train.
-                    var predictor = TrainUtils.Train(host, ch, trainData, trainer, _trainer.Kind, validData,
+                    var predictor = TrainUtils.Train(host, ch, trainData, trainer, validData,
                         _calibrator, _maxCalibrationExamples, _cacheData, _inputPredictor);
 
                     // Score.

src/Microsoft.ML.Data/Commands/TrainCommand.cs

@@ -177,7 +177,7 @@ private void RunCore(IChannel ch, string cmd)
                 }
             }
 
-            var predictor = TrainUtils.Train(Host, ch, data, trainer, _info.LoadNames[0], validData,
+            var predictor = TrainUtils.Train(Host, ch, data, trainer, validData,
                 Args.Calibrator, Args.MaxCalibrationExamples, Args.CacheData, inputPredictor);
 
             using (var file = Host.CreateOutputFile(Args.OutputModelFile))
@@ -228,28 +228,27 @@ public static string MatchNameOrDefaultOrNull(IExceptionContext ectx, ISchema sc
 #pragma warning restore MSML_ContractsNameUsesNameof
         }
 
-        public static IPredictor Train(IHostEnvironment env, IChannel ch, RoleMappedData data, ITrainer trainer, string name,
+        public static IPredictor Train(IHostEnvironment env, IChannel ch, RoleMappedData data, ITrainer trainer,
             ICalibratorTrainerFactory calibrator, int maxCalibrationExamples)
         {
             var caliTrainer = calibrator?.CreateComponent(env);
-            return TrainCore(env, ch, data, trainer, name, null, caliTrainer, maxCalibrationExamples, false);
+            return TrainCore(env, ch, data, trainer, null, caliTrainer, maxCalibrationExamples, false);
         }
 
-        public static IPredictor Train(IHostEnvironment env, IChannel ch, RoleMappedData data, ITrainer trainer, string name, RoleMappedData validData,
+        public static IPredictor Train(IHostEnvironment env, IChannel ch, RoleMappedData data, ITrainer trainer, RoleMappedData validData,
             IComponentFactory<ICalibratorTrainer> calibrator, int maxCalibrationExamples, bool? cacheData, IPredictor inputPredictor = null)
         {
             ICalibratorTrainer caliTrainer = calibrator?.CreateComponent(env);
-            return TrainCore(env, ch, data, trainer, name, validData, caliTrainer, maxCalibrationExamples, cacheData, inputPredictor);
+            return TrainCore(env, ch, data, trainer, validData, caliTrainer, maxCalibrationExamples, cacheData, inputPredictor);
         }
 
-        private static IPredictor TrainCore(IHostEnvironment env, IChannel ch, RoleMappedData data, ITrainer trainer, string name, RoleMappedData validData,
+        private static IPredictor TrainCore(IHostEnvironment env, IChannel ch, RoleMappedData data, ITrainer trainer, RoleMappedData validData,
             ICalibratorTrainer calibrator, int maxCalibrationExamples, bool? cacheData, IPredictor inputPredictor = null)
         {
             Contracts.CheckValue(env, nameof(env));
             env.CheckValue(ch, nameof(ch));
             ch.CheckValue(data, nameof(data));
             ch.CheckValue(trainer, nameof(trainer));
-            ch.CheckNonEmpty(name, nameof(name));
             ch.CheckValueOrNull(validData);
             ch.CheckValueOrNull(inputPredictor);
 

src/Microsoft.ML.Data/Commands/TrainTestCommand.cs

@@ -166,7 +166,7 @@ private void RunCore(IChannel ch, string cmd)
                 }
             }
 
-            var predictor = TrainUtils.Train(Host, ch, data, trainer, _info.LoadNames[0], validData,
+            var predictor = TrainUtils.Train(Host, ch, data, trainer, validData,
                 Args.Calibrator, Args.MaxCalibrationExamples, Args.CacheData, inputPredictor);
 
             IDataLoader testPipe;

src/Microsoft.ML.Data/EntryPoints/InputBase.cs

@@ -186,7 +186,7 @@ public static TOut Train<TArg, TOut>(IHost host, TArg input,
                     cachedRoleMappedData = new RoleMappedData(cacheView, roleMappedData.Schema.GetColumnRoleNames());
                 }
 
-                var predictor = TrainUtils.Train(host, ch, cachedRoleMappedData, trainer, "Train", calibrator, maxCalibrationExamples);
+                var predictor = TrainUtils.Train(host, ch, cachedRoleMappedData, trainer, calibrator, maxCalibrationExamples);
                 var output = new TOut() { PredictorModel = new PredictorModel(host, roleMappedData, input.TrainingData, predictor) };
 
                 ch.Done();

src/Microsoft.ML.Data/Transforms/TrainAndScoreTransform.cs

@@ -233,7 +233,7 @@ private static IDataTransform Create(IHostEnvironment env, Arguments args, ITrai
                 string feat;
                 string group;
                 var data = CreateDataFromArgs(ch, input, args, out feat, out group);
-                var predictor = TrainUtils.Train(host, ch, data, trainer, args.Trainer.Kind, null,
+                var predictor = TrainUtils.Train(host, ch, data, trainer, null,
                     args.Calibrator, args.MaxCalibrationExamples, null);
 
                 ch.Done();

src/Microsoft.ML.Data/Utilities/ModelFileUtils.cs

@@ -282,8 +282,7 @@ public static IEnumerable<KeyValuePair<ColumnRole, string>> LoadRoleMappingsOrNu
             {
                 // REVIEW: Should really validate the schema here, and consider
                 // ignoring this stream if it isn't as expected.
-                var loaderSub = new SubComponent<IDataLoader, SignatureDataLoader>("Text");
-                var loader = loaderSub.CreateInstance(env,
+                var loader = new TextLoader(env, new TextLoader.Arguments(),
                     new RepositoryStreamWrapper(rep, DirTrainingInfo, RoleMappingFile));
 
                 using (var cursor = loader.GetRowCursor(c => true))

src/Microsoft.ML.Transforms/LearnerFeatureSelection.cs

@@ -7,8 +7,9 @@
 using Microsoft.ML.Runtime;
 using Microsoft.ML.Runtime.CommandLine;
 using Microsoft.ML.Runtime.Data;
-using Microsoft.ML.Runtime.Internal.Utilities;
+using Microsoft.ML.Runtime.EntryPoints;
 using Microsoft.ML.Runtime.Internal.Internallearn;
+using Microsoft.ML.Runtime.Internal.Utilities;
 
 [assembly: LoadableClass(LearnerFeatureSelectionTransform.Summary, typeof(IDataTransform), typeof(LearnerFeatureSelectionTransform), typeof(LearnerFeatureSelectionTransform.Arguments), typeof(SignatureDataTransform),
     "Learner Feature Selection Transform", "LearnerFeatureSelectionTransform", "LearnerFeatureSelection")]
@@ -32,9 +33,11 @@ public sealed class Arguments
             [Argument(ArgumentType.AtMostOnce, HelpText = "The number of slots to preserve", ShortName = "topk", SortOrder = 1)]
             public int? NumSlotsToKeep;
 
-            [Argument(ArgumentType.Multiple, HelpText = "Filter", ShortName = "f", SortOrder = 1)]
-            public SubComponent<ITrainer<IPredictorWithFeatureWeights<Single>>, SignatureFeatureScorerTrainer> Filter =
-                new SubComponent<ITrainer<IPredictorWithFeatureWeights<Single>>, SignatureFeatureScorerTrainer>("SDCA");
+            [Argument(ArgumentType.Multiple, HelpText = "Filter", ShortName = "f", SortOrder = 1, SignatureType = typeof(SignatureFeatureScorerTrainer))]
+            public IComponentFactory<ITrainer<IPredictorWithFeatureWeights<Single>>> Filter =
+                ComponentFactoryUtils.CreateFromFunction(env =>
+                    // ML.Transforms doesn't have a direct reference to ML.StandardLearners, so use ComponentCatalog to create the Filter
+                    ComponentCatalog.CreateInstance<ITrainer<IPredictorWithFeatureWeights<Single>>>(env, typeof(SignatureFeatureScorerTrainer), "SDCA", options: null));
 
             [Argument(ArgumentType.LastOccurenceWins, HelpText = "Column to use for features", ShortName = "feat,col", SortOrder = 3, Purpose = SpecialPurpose.ColumnName)]
             public string FeatureColumn = DefaultColumnNames.Features;
@@ -283,7 +286,7 @@ private static void TrainCore(IHost host, IDataView input, Arguments args, ref V
             using (var ch = host.Start("Train"))
             {
                 ch.Trace("Constructing trainer");
-                ITrainer trainer = args.Filter.CreateInstance(host);
+                ITrainer trainer = args.Filter.CreateComponent(host);
 
                 IDataView view = input;
 
@@ -301,7 +304,7 @@ private static void TrainCore(IHost host, IDataView input, Arguments args, ref V
                 var customCols = TrainUtils.CheckAndGenerateCustomColumns(ch, args.CustomColumn);
                 var data = new RoleMappedData(view, label, feature, group, weight, name, customCols);
 
-                var predictor = TrainUtils.Train(host, ch, data, trainer, args.Filter.Kind, null,
+                var predictor = TrainUtils.Train(host, ch, data, trainer, null,
                     null, 0, args.CacheData);
 
                 var rfs = predictor as IPredictorWithFeatureWeights<Single>;

src/Microsoft.ML.Transforms/RffTransform.cs

@@ -10,6 +10,7 @@
 using Microsoft.ML.Runtime;
 using Microsoft.ML.Runtime.CommandLine;
 using Microsoft.ML.Runtime.Data;
+using Microsoft.ML.Runtime.EntryPoints;
 using Microsoft.ML.Runtime.Internal.CpuMath;
 using Microsoft.ML.Runtime.Internal.Utilities;
 using Microsoft.ML.Runtime.Model;
@@ -39,11 +40,12 @@ public sealed class Arguments
             [Argument(ArgumentType.AtMostOnce, HelpText = "The number of random Fourier features to create", ShortName = "dim")]
             public int NewDim = Defaults.NewDim;
 
-            [Argument(ArgumentType.Multiple, HelpText = "which kernel to use?", ShortName = "kernel")]
-            public SubComponent<IFourierDistributionSampler, SignatureFourierDistributionSampler> MatrixGenerator =
-                new SubComponent<IFourierDistributionSampler, SignatureFourierDistributionSampler>(GaussianFourierSampler.LoadName);
+            [Argument(ArgumentType.Multiple, HelpText = "Which kernel to use?", ShortName = "kernel", SignatureType = typeof(SignatureFourierDistributionSampler))]
+            public IComponentFactory<Float, IFourierDistributionSampler> MatrixGenerator =
+                ComponentFactoryUtils.CreateFromFunction<Float, IFourierDistributionSampler>(
+                    (env, avgDist) => new GaussianFourierSampler(env, new GaussianFourierSampler.Arguments(), avgDist));
 
-            [Argument(ArgumentType.AtMostOnce, HelpText = "create two features for every random Fourier frequency? (one for cos and one for sin)")]
+            [Argument(ArgumentType.AtMostOnce, HelpText = "Create two features for every random Fourier frequency? (one for cos and one for sin)")]
             public bool UseSin = Defaults.UseSin;
 
             [Argument(ArgumentType.LastOccurenceWins,
@@ -57,8 +59,8 @@ public sealed class Column : OneToOneColumn
             [Argument(ArgumentType.AtMostOnce, HelpText = "The number of random Fourier features to create", ShortName = "dim")]
             public int? NewDim;
 
-            [Argument(ArgumentType.Multiple, HelpText = "which kernel to use?", ShortName = "kernel")]
-            public SubComponent<IFourierDistributionSampler, SignatureFourierDistributionSampler> MatrixGenerator;
+            [Argument(ArgumentType.Multiple, HelpText = "which kernel to use?", ShortName = "kernel", SignatureType = typeof(SignatureFourierDistributionSampler))]
+            public IComponentFactory<Float, IFourierDistributionSampler> MatrixGenerator;
 
             [Argument(ArgumentType.AtMostOnce, HelpText = "create two features for every random Fourier frequency? (one for cos and one for sin)")]
             public bool? UseSin;
@@ -81,7 +83,7 @@ public static Column Parse(string str)
             public bool TryUnparse(StringBuilder sb)
             {
                 Contracts.AssertValue(sb);
-                if (NewDim != null || MatrixGenerator.IsGood() || UseSin != null || Seed != null)
+                if (NewDim != null || MatrixGenerator != null || UseSin != null || Seed != null)
                     return false;
                 return TryUnparseCore(sb);
             }
@@ -115,10 +117,10 @@ public TransformInfo(IHost host, Column item, Arguments args, int d, Float avgDi
                 _rand = seed.HasValue ? RandomUtils.Create(seed) : RandomUtils.Create(host.Rand);
                 _state = _rand.GetState();
 
-                var sub = item.MatrixGenerator;
-                if (!sub.IsGood())
-                    sub = args.MatrixGenerator;
-                _matrixGenerator = sub.CreateInstance(host, avgDist);
+                var generator = item.MatrixGenerator;
+                if (generator == null)
+                    generator = args.MatrixGenerator;
+                _matrixGenerator = generator.CreateComponent(host, avgDist);
 
                 int roundedUpD = RoundUp(NewDim, CfltAlign);
                 int roundedUpNumFeatures = RoundUp(SrcDim, CfltAlign);
@@ -417,12 +419,13 @@ private static Float[] Train(IHost host, ColInfo[] infos, Arguments args, IDataV
                 else
                 {
                     Float[] distances;
-
                     var sub = args.Column[iinfo].MatrixGenerator;
-                    if (!sub.IsGood())
+                    if (sub == null)
                         sub = args.MatrixGenerator;
-                    var info = ComponentCatalog.GetLoadableClassInfo(sub);
-                    bool gaussian = info != null && info.Type == typeof(GaussianFourierSampler);
+                    // create a dummy generator in order to get its type.
+                    // REVIEW this should be refactored. See https://github.com/dotnet/machinelearning/issues/699
+                    var matrixGenerator = sub.CreateComponent(host, 1);
+                    bool gaussian = matrixGenerator is GaussianFourierSampler;
 
                     // If the number of pairs is at most the maximum reservoir size / 2, go over all the pairs.
                     if (resLength < reservoirSize)