Using ML.NET to Build Machine Learning Models/

...

/

Retraining and Saving an ML.NET Model

Retraining and Saving an ML.NET Model

Learn how to retrain and save an ML.NET model.

We'll cover the following...

With ML.NET, we can train and retrain the model from the code. Whether we're training a new model or retraining an existing one, the process is largely the same. We would execute the same methods in the core library. Then, once the model is trained or retrained, we can save it in a file for consumption.

Retraining and saving a model in ML.NET

Sometimes it’s necessary to retrain the model. For example, the original dataset might have been too small or the data inside it was inaccurate. Perhaps we also want to clear up the data further because certain features don’t make the model any more accurate, but their presence significantly increases the training time.

In this lesson, we'll go through the process of training and retraining the model by using ML.NET. We'll do so with the help of the following playground:

using Microsoft.ML;
using Microsoft.ML.Data;
using System;
using System.Linq;
using System.IO;
using System.Collections.Generic;
namespace ModelSavingExample.ConsoleApp
{
    public partial class ModelSavingExample
    {
        /// <summary>
        /// model input class for ModelSavingExample.
        /// </summary>
        #region model input class
        public class ModelInput
        {
            [LoadColumn(0)]
            [ColumnName(@"col0")]
            public string Col0 { get; set; }

            [LoadColumn(1)]
            [ColumnName(@"col1")]
            public float Col1 { get; set; }
        }

        #endregion

        /// <summary>
        /// model output class for ModelSavingExample.
        /// </summary>
        #region model output class
        public class ModelOutput
        {
            [ColumnName(@"col0")]
            public float[] Col0 { get; set; }

            [ColumnName(@"col1")]
            public uint Col1 { get; set; }

            [ColumnName(@"Features")]
            public float[] Features { get; set; }

            [ColumnName(@"PredictedLabel")]
            public float PredictedLabel { get; set; }

            [ColumnName(@"Score")]
            public float[] Score { get; set; }

        }

        #endregion

        private static string MLNetModelPath = Path.GetFullPath("ModelSavingExample.mlnet");

        public static readonly Lazy<PredictionEngine<ModelInput, ModelOutput>> PredictEngine = new Lazy<PredictionEngine<ModelInput, ModelOutput>>(() => CreatePredictEngine(), true);


        private static PredictionEngine<ModelInput, ModelOutput> CreatePredictEngine()
        {
            var mlContext = new MLContext();
            ITransformer mlModel = mlContext.Model.Load(MLNetModelPath, out var _);
            return mlContext.Model.CreatePredictionEngine<ModelInput, ModelOutput>(mlModel);
        }

        public static IOrderedEnumerable<KeyValuePair<string, float>> PredictAllLabels(ModelInput input)
        {
            var predEngine = PredictEngine.Value;
            var result = predEngine.Predict(input);
            return GetSortedScoresWithLabels(result);
        }

        public static IOrderedEnumerable<KeyValuePair<string, float>> GetSortedScoresWithLabels(ModelOutput result)
        {
            var unlabeledScores = result.Score;
            var labelNames = GetLabels(result);

            Dictionary<string, float> labledScores = new Dictionary<string, float>();
            for (int i = 0; i < labelNames.Count(); i++)
            {
                // Map the names to the predicted result score array
                var labelName = labelNames.ElementAt(i);
                labledScores.Add(labelName.ToString(), unlabeledScores[i]);
            }

            return labledScores.OrderByDescending(c => c.Value);
        }

        private static IEnumerable<string> GetLabels(ModelOutput result)
        {
            var schema = PredictEngine.Value.OutputSchema;

            var labelColumn = schema.GetColumnOrNull("col1");
            if (labelColumn == null)
            {
                throw new Exception("col1 column not found. Make sure the name searched for matches the name in the schema.");
            }

            // Key values contains an ordered array of the possible labels. This allows us to map the results to the correct label value.
            var keyNames = new VBuffer<float>();
            labelColumn.Value.GetKeyValues(ref keyNames);
            return keyNames.DenseValues().Select(x => x.ToString());
        }

        public static ModelOutput Predict(ModelInput input)
        {
            var predEngine = PredictEngine.Value;
            return predEngine.Predict(input);
        }

    }
}
An example of model retraining process

This playground contains the code that is generated by the binary classification ML.NET CLI task. The classification was based on sentiment analysis. However, we make some modifications to it. First, we remove the original model, so we're forced to retrain the model ...