How to use a generated H2O Autoencoder / Anomaly Detection Model for Inference in Java API?

Question

I want to use H2O Autoencoder (Anomaly Detection) for Inference / Prediction in a Java class.

I built the autoencoder example "ECG Hearbeats" from H2O DeepLearningBooklet with R and saved it. I can succesfully import the generated Java class and its related h2o-genmodel.jar into my Java project.

Unfortunately, I cannot find an example or documentation how to use it there.

Here is my first try with some code and some guesses from my experience with other H2O models used for inference in Java code:

private static String modelClassName = "machinelearning.DeepLearning_model_R_1509973865970_1";

public static void main(String[] args) throws Exception {

    hex.genmodel.GenModel rawModel;
    rawModel = (hex.genmodel.GenModel) Class.forName(modelClassName).newInstance();
    EasyPredictModelWrapper model = new EasyPredictModelWrapper(rawModel);

    RowData row = new RowData();
    // row.put(key, value); // TODO Add new line of input data, e.g.:
    // 2.10,2.13,2.19,2.28,2.44,2.62,2.80,3.04,3.36,3.69,3.97,4.24,4.53,4.80,5.02,5.21,5.40,5.57,5.71,5.79,5.86,5.92,5.98,6.02,6.06,6.08,6.14,6.18,6.22,6.27,6.32,6.35,6.38,6.45,6.49,6.53,6.57,6.64,6.70,6.73,6.78,6.83,6.88,6.92,6.94,6.98,7.01,7.03,7.05,7.06,7.07,7.08,7.06,7.04,7.03,6.99,6.94,6.88,6.83,6.77,6.69,6.60,6.53,6.45,6.36,6.27,6.19,6.11,6.03,5.94,5.88,5.81,5.75,5.68,5.62,5.61,5.54,5.49,5.45,5.42,5.38,5.34,5.31,5.30,5.29,5.26,5.23,5.23,5.22,5.20,5.19,5.18,5.19,5.17,5.15,5.14,5.17,5.16,5.15,5.15,5.15,5.14,5.14,5.14,5.15,5.14,5.14,5.13,5.15,5.15,5.15,5.14,5.16,5.15,5.15,5.14,5.14,5.15,5.15,5.14,5.13,5.14,5.14,5.11,5.12,5.12,5.12,5.09,5.09,5.09,5.10,5.08,5.08,5.08,5.08,5.06,5.05,5.06,5.07,5.05,5.03,5.03,5.04,5.03,5.01,5.01,5.02,5.01,5.01,5.00,5.00,5.02,5.01,4.98,5.00,5.00,5.00,4.99,5.00,5.01,5.02,5.01,5.03,5.03,5.02,5.02,5.04,5.04,5.04,5.02,5.02,5.01,4.99,4.98,4.96,4.96,4.96,4.94,4.93,4.93,4.93,4.93,4.93,5.02,5.27,5.80,5.94,5.58,5.39,5.32,5.25,5.21,5.13,4.97,4.71,4.39,4.05,3.69,3.32,3.05,2.99,2.74,2.61,2.47,2.35,2.26,2.20,2.15,2.10,2.08

    AutoEncoderModelPrediction p = model.predictAutoEncoder(row);

    System.out.println(p.reconstructedRowData);
    System.out.println(p.reconstructed[0]);
    // TODO How to do get the MSE from object 'p'? 

This code actually compiles and runs. However, I do not really understand how to

1. configure EasyPredictModelWrapper correctly (or do I just need to instantiate it?)
2. add a new single event for prediction as I do not have keys, but only values for the ECG Heartbeats. (I assume the method row.putAll is the best to the various features of the ECG data set ?!)
3. get the MSE out of the prediction (similar to what we see in the R / Python example with 'recon_error <- as.data.frame(recon_error))' where you can see the reconstruction error for all 23 ECG lines? (then I can define a threshold, and build a real time application which creates an alert if a threshold is reached)

I assume the answer is simple, but without documentation not easy to find :-)

Thanks for help.

Answer 1

(code example for main.java at the end)

1. you have it correctly and it is instantiated in this line rawModel = (hex.genmodel.GenModel) Class.forName(modelClassName).newInstance();

2. the key is the column header and value is the actual value, if the H2Oframe doesn’t have column headers then H2O will automatically assign them C1, C2, etc. You can manually write this or use a loop using System.out.println(java.util.Arrays.toString(rawModel.getNames())); (see code snippet for example of this)

3. there is not a method for this currently, but you can get the original values and reconstructed values and then calculate the MSE from that (see code snippet below, the last few lines calculate the MSE using the original and reconstructed arrays)

When I created my model I called it anomaly_model (see the code directly below, model_id is one of the parameters) and you will see that used in the last code snippet below, so if you use a different name you will need to update that part.

anomaly_model <- h2o.deeplearning(x = names(train_ecg), training_frame = train_ecg, activation = "Tanh",
                              autoencoder = TRUE,hidden = c(50,20,50),sparse = TRUE,l1 = 1e-4,epochs = 100, model_id = 'anomaly_model')

Here is example code for how to create the main.java file, pass in column names for your keys, and calculate MSE with built in method results.

(Note: I generated random values for the row.put(key, values) you can put whatever you want there instead)

import java.io.*;
import hex.genmodel.easy.RowData;
import hex.genmodel.easy.EasyPredictModelWrapper;
import hex.genmodel.easy.prediction.*;

public class main {
  private static String modelClassName = "anomaly_model";

  public static void main(String[] args) throws Exception {
    hex.genmodel.GenModel rawModel;
    rawModel = (hex.genmodel.GenModel) Class.forName(modelClassName).newInstance();
    EasyPredictModelWrapper model = new EasyPredictModelWrapper(rawModel);

    java.util.Random rng = new java.util.Random();
    RowData row = new RowData();
    for (String colName : rawModel.getNames()) {
      row.put(colName,rng.nextDouble());
    }

    AutoEncoderModelPrediction p = model.predictAutoEncoder(row);
    System.out.println("original: " + java.util.Arrays.toString(p.original));
    System.out.println("reconstructedrowData: " + p.reconstructedRowData);
    System.out.println("reconstructed: " + java.util.Arrays.toString(p.reconstructed));

    double sum = 0;
    for (int i = 0; i<p.original.length; i++) {
      sum += (p.original[i] - p.reconstructed[i])*(p.original[i] - p.reconstructed[i]);
    }
    double mse = sum/p.original.length;
    System.out.println("MSE: " + mse);
  }
}

Hope this helps!