Created release notes for v1.5.3

docs/release-notes/1.5.3/release-1.5.3.md

@@ -0,0 +1,45 @@
+# [ML.NET](http://dot.net/ml) 1.5.3
+
+## **New Features**
+- **New API for exporting models to Onnx**. ([#5544](https://github.com/dotnet/machinelearning/pull/5544)). A new API has been added to Onnx converter to specify the output columns you care about. This will export a smaller and more performant model in many cases.
+
+## **Enhancements**
+- Perf improvement for TopK Accuracy and return all topK in Classification Evaluator ([#5395](https://github.com/dotnet/machinelearning/pull/5395)) (**Thank you @jasallen**)
+- Update OnnxRuntime to 1.6 ([#5529](https://github.com/dotnet/machinelearning/pull/5529))
+- Updated tensorflow.net to 0.20.0 ([#5404](https://github.com/dotnet/machinelearning/pull/5404))
+- Added in DcgTruncationLevel to AutoML api and increased default level to 10 ([#5433](https://github.com/dotnet/machinelearning/pull/5433))
+
+## **Bug Fixes**
+- **AutoML.NET specific fixes**.
+  - Fixed AutoFitMaxExperimentTimeTest ([#5506](https://github.com/dotnet/machinelearning/pull/5506))
+  - Fixed code generator tests failure ([#5520](https://github.com/dotnet/machinelearning/pull/5520))
+  - Use Timer and ctx.CancelExecution() to fix AutoML max-time experiment bug ([#5445](https://github.com/dotnet/machinelearning/pull/5445))
+  - Handled exception during GetNextPipeline for AutoML ([#5455](https://github.com/dotnet/machinelearning/pull/5455))
+  - Fixed internationalization bug([#5162](https://github.com/dotnet/machinelearning/pull/5163)) in AutoML parameter sweeping caused by culture dependent float parsing. ([#5163](https://github.com/dotnet/machinelearning/pull/5163))
+  - Fixed MaxModels exit criteria for AutoML unit test ([#5471](https://github.com/dotnet/machinelearning/pull/5471))
+  - Fixed AutoML CrossValSummaryRunner for TopKAccuracyForAllK ([#5548](https://github.com/dotnet/machinelearning/pull/5548))
+- Fixed bug in Tensorflow Transforer with handling primitive types ([#5547](https://github.com/dotnet/machinelearning/pull/5547))
+- Fixed MLNet.CLI build error ([#5546](https://github.com/dotnet/machinelearning/pull/5546))
+- Fixed memory leaks from OnnxTransformer ([#5518](https://github.com/dotnet/machinelearning/pull/5518))
+- Fixed memory leak in object pool ([#5521](https://github.com/dotnet/machinelearning/pull/5521))
+- Fixed Onnx Export for ProduceWordBags ([#5435](https://github.com/dotnet/machinelearning/pull/5435))
+- Upgraded boundary calculation and expected value calculation in SrCnnEntireAnomalyDetector ([#5436](https://github.com/dotnet/machinelearning/pull/5436))
+- Fixed SR anomaly score calculation at beginning ([#5502](https://github.com/dotnet/machinelearning/pull/5502))
+- Improved error message in ColumnConcatenatingEstimator ([#5444](https://github.com/dotnet/machinelearning/pull/5444))
+- Fixed issue 5020, allow ML.NET to load tf model with primitive input and output column ([#5468](https://github.com/dotnet/machinelearning/pull/5468))
+- Fixed issue 4322, enable lda summary output ([#5260](https://github.com/dotnet/machinelearning/pull/5260))
+- Fixed perf regression in ShuffleRows ([#5417](https://github.com/dotnet/machinelearning/pull/5417))
+- Change the _maxCalibrationExamples default on CalibratorUtils ([#5415](https://github.com/dotnet/machinelearning/pull/5415))
+
+
+## **Build / Test updates**
+- Migrated to [Arcade](https://github.com/dotnet/arcade/) build system that is used my multiple dotnet projects. This will give increased build/CI efficiencies going forward. Updated build instructions can be found in the docs/building folder
+- Fixed MacOS builds ([#5467](https://github.com/dotnet/machinelearning/pull/5467) and [#5457](https://github.com/dotnet/machinelearning/pull/5457))
+
+## **Documentation Updates**
+- Fixed Spelling on stopwords ([#5524](https://github.com/dotnet/machinelearning/pull/5524))(**Thank you @LeoGaunt**)
+- Changed LoadRawImages Sample ([#5460](https://github.com/dotnet/machinelearning/pull/5460))
+
+
+## **Breaking Changes**
+- None