set trial to unsuccess if trial loss is nan/inf

src/Microsoft.ML.AutoML/CodeGen/estimator-schema.json

@@ -71,7 +71,7 @@
               "DnnFeaturizerImage",
               "Naive",
               "ForecastBySsa",
-              "TextClassifcation"
+              "TextClassification"
             ]
           },
           "nugetDependencies": {

src/Microsoft.ML.AutoML/CodeGen/trainer-estimators.json

@@ -512,7 +512,7 @@
       "searchOption": "image_classification_option"
     },
     {
-      "functionName": "TextClassifcation",
+      "functionName": "TextClassification",
       "estimatorTypes": [ "MultiClassification" ],
       "arguments": [
         {

src/Microsoft.ML.AutoML/SweepableEstimator/Estimators/TextClassification.cs

@@ -10,7 +10,7 @@
 
 namespace Microsoft.ML.AutoML.CodeGen
 {
-    internal partial class TextClassifcationMulti
+    internal partial class TextClassificationMulti
     {
         public override IEstimator<ITransformer> BuildFromOption(MLContext context, TextClassificationOption param)
         {

src/Microsoft.ML.AutoML/Tuner/PipelineProposer.cs

@@ -128,7 +128,7 @@ public void Update(TrialResult result, string schema)
         {
             var loss = result.Loss;
             var duration = result.DurationInMilliseconds / 1000;
-            var isSuccess = duration != 0;
+            var isSuccess = duration != 0 && !double.IsNaN(loss) && !double.IsInfinity(loss);
 
             // if k1 is null, it means this is the first completed trial.
             // in that case, initialize k1, k2, e1, e2 in the following way:

test/Microsoft.ML.AutoML.Tests/TunerTests.cs

@@ -127,6 +127,40 @@ public void CFO_should_start_from_init_point_if_provided()
             (x * x + y * y + z * z).Should().Be(0);
         }
 
+        [Fact]
+        public void EciCfo_should_handle_trial_result_with_nan_value()
+        {
+            // this test verify if tuner can find max value for LSE.
+            var context = new MLContext(1);
+            var pipeline = this.CreateDummySweepablePipeline(context);
+            var searchSpace = new SearchSpace.SearchSpace();
+            searchSpace["_pipeline_"] = pipeline.SearchSpace;
+            var tuner = new EciCostFrugalTuner(pipeline, new AutoMLExperiment.AutoMLExperimentSettings
+            {
+                SearchSpace = searchSpace,
+                Seed = 1,
+            });
+            var invalidLosses = new[] { double.NaN, double.NegativeInfinity, double.PositiveInfinity };
+            var id = 0;
+            foreach (var loss in invalidLosses)
+            {
+                var trialSetting = new TrialSettings
+                {
+                    TrialId = id++,
+                    Parameter = Parameter.CreateNestedParameter(),
+                };
+                var parameter = tuner.Propose(trialSetting);
+                trialSetting.Parameter = parameter;
+                var trialResult = new TrialResult
+                {
+                    TrialSettings = trialSetting,
+                    DurationInMilliseconds = 10000,
+                    Loss = double.NaN,
+                };
+                tuner.Update(trialResult);
+            }
+        }
+
         [Fact]
         public void LSE_maximize_test()
         {
@@ -346,5 +380,16 @@ private class LSE3DSearchSpace
             [Range(-10.0, 10.0, 0.0, false)]
             public double Z { get; set; }
         }
+
+        private SweepablePipeline CreateDummySweepablePipeline(MLContext context)
+        {
+            var mapKeyToValue = SweepableEstimatorFactory.CreateMapKeyToValue(new MapKeyToValueOption
+            {
+                InputColumnName = "input",
+                OutputColumnName = "output",
+            });
+
+            return mapKeyToValue.Append(context.Auto().BinaryClassification());
+        }
     }
 }