ML Kit Example App

Content Type: Sample
Categories: Artificial Intelligence

Overview

Studio Pro supports a unique approach to machine learning (ML) model integration in Mendix apps. Collectively called the Machine Learning (ML) Kit, this functionality allows Mendix developers to deploy a machine learning model built using common ML framework and language into the Mendix Runtime.


In this sample app you can find the resources for start creating Smart Apps with MLKit

Documentation

Typical usage scenario

This sample app can be used to create and understand ML Kit in Studio Pro.


Features and limitations


Shallow Models

Clasification : Iris Species

Here you can find examples for solving the standard Iris species classification problem. We provide a decision tree example, a random forest, a logistic regressor and an Ensemble demo for combining all of these should you want to reuse any of the ML design patterns.


Clasification : Content filtering

There we implemented a Spam filter using a TF/IDF Vectorizer


Clasification : Titanic survivors

In this module you can find an example of solving the titanic challenge dataset with a SKLearn pipeline that handles missing features and applies normalization, along a Xgboost classifier.


Deep learning Models

Of course you can run neural networks in Mendix! Please see our examples below, both for implementation and especially, for pre/post processors.


Computer Vision


Image Classification : ResNet50

Find here a standard ResNet50 example fully implemented. Please refer to the ONNX Model Zoo for details on model and training


Style Transfer: Fast Neural Style

Yes, you can make an app that turns any picture into a Vermeer with Mendix. Here we show you how can you apply a Mosaic style transfer from this model in ONNX Zoo, along with pre and post processors.


Language Models with Generative Interfaces

In the BERT example you can see how we implemented BERT in Mendix, complete with pre/post processors and tokenizers.

Please make sure that the bertsquad-12-int8.onnx file is present in the /path/to/your/app/mlsource/bert folder. You can download it from here if needed.


Known bugs

If you are having errors launching the app. Update MendixSSO and CommunityCommons packages from marketplace


Releases

Version: 1.0.0
Framework Version: 9.24.4
Release Notes: Updated marketplace modules. Snyk vulnerabilities are fixed.