Getting started with Real-time Machine Learning (Alpha)
Real-time Machine Learning is not available to all users yet. This feature is in alpha and still being tested. This document is subject to change.
In order to utilize Real-time Machine Learning, you need to have access to an organization provisioned with Adobe Experience Platform and Data Science Workspace. Additionally, you need to have a complete dataset for use in training and scoring.
The guides for Real-time Machine Learning require a working understanding of Python 3, Jupyter notebooks , data science, and machine learning.
- DSL: Domain Specific Language.
- Edge: Real-time Machine Learning scoring service can be run on Edge clusters closer to your activations and applications.
- Hub: The current alpha is running the Real-time Machine Learning scoring service on the Adobe Experience Platform Hub while the Experience Edge Network is in development.
- Node: A Node is the fundamental unit of which graphs are formed. Each node performs a specific task and they can be chained together using links to form a graph that represents an ML pipeline. The task performed by a node represents an operation on input data such as a transformation of data or schema, or a machine learning inference. The node outputs the transformed or inferred value to the next node(s).
Datasets in Adobe Experience Platform
To start using Real-time Machine Learning, you must have access to a dataset. You have the option to use an external dataset and upload it to your JupyterLab environment or create a new dataset within Platform if you have not done so already.
If you already have a dataset you wish to use, you can skip to Next steps .
Use an external dataset
To learn more about using an external dataset such as uploading data to your JupyterLab environment, visit the tutorial on analyzing your data using notebooks .
Create a new dataset
To create a new dataset for use in Real-time Machine Learning, you need a data-schema for your dataset. Next, you need to ingest data using the schema you created. Use the following tutorials to create and populate a dataset for Platform:
Once you have prepared your data for Real-time Machine Learning, start by following the Real-time Machine Learning notebook user guide to learn how to create and upload an ONNX model to the Real-time Machine Learning model store.