Machine Learning Model APIs
Use these APIs to deploy machine learning (ML) models and generate predictions. Fusion’s Machine Learning (ML) Model service stores deployed models and runs prediction jobs.
-
Use this API to deploy ML models and generate predictions.
-
The Serialized Model API is for fetching, updating, or deleting deployed models.
For more information, view the API specification.
Fusion also provides convenient jobs you can configure to deploy your models via the ML Model service, plus pipelines and pipeline stages for querying them. See these topics:
-
Develop and Deploy a Machine Learning Model provides general instructions for using ML models with Fusion, including pre-trained models and your own custom models. Lucidworks provides pre-trained models for sentiment analysis and prediction and Smart Answers. Then you can query your model using the Machine Learning query pipeline stage.
-
Train a Smart Answers Supervised Model explains how to configure the Smart Answers Supervised Training job to train a model on an existing body of question/answer data and deploy it for use with your Smart Answers pipelines.
-
If your existing question/answer data is sparse or your data is in another format, use the Train a Smart Answers Cold Start Model to train and deploy a model you can use to get started. Lucidworks also provides pre-trained cold start models you can use.