Bayesia Engine API
Context
- You can access many of BayesiaLab’s functions outside the graphical user interface by using a range of Bayesia Engine APIs.
- With the Bayesia Engine APIs, you can access BayesiaLab’s functions programmatically and integrate them into your own applications.
- You can license Bayesia Engine APIs for use in two main contexts, i.e., for software development and production deployment.
For Production Deployment
- The principal objective of using the Bayesia Engine API is to automate certain BayesiaLab functions and deploy them at scale.
- For a much-simplified approach to deploying the Bayesia Engine API, we now offer a RESTful API as a stateless Web Service.
- This allows you to perform inference with a given Bayesian network model using straightforward HTTPS calls. Hence, no coding in Java is required.
- A typical application would be real-time scoring of large volumes of streaming data.
- To scale up the processing capacity of the Bayesia Engine API in production, you can launch any number of instances of the API in parallel, subject only to the maxima specified in your license agreement.
- The license fee for a production license is based on the number of API instances that can run simultaneously, regardless of the physical computers involved and their locations. This means that one instance each on five different servers costs the same as five instances on a single server. Please note one instance means that a network model is loaded by the API, ready for an operation to be performed.
- If speed is of no concern, tasks can also be performed by sequentially loading and unloading different models, which would require only a single instance at any given time.
API Overview
- The following table summarizes the functions and limitations of the available configurations of the Bayesia Engine APIs.
For Software Development | For Production Deployment | ||||
|---|---|---|---|---|---|
| Bayesia Engine Java API for Modeling and Inference | Bayesia Engine Java API for Net work Learning | Bayesia Java API for Inference | Bayesia Engine REST API for Inference | ||
Inference based on existing Bayesian network models | |||||
Modeling, i.e., building or modifying Bayesian networks using code | |||||
Parameter estimation of existing Bayesian network models | |||||
Learning, i.e., structural le arn ing of new Bayesian network models from data | |||||
Pricing | Fees for 12-Month License Rental (USD) | ||||
Location of Bayesian network models | .xbl files always reside on the customer's servers. | .xbl files reside on Bayesia servers in France | |||
(.xbl file) | |||||
Installation/usage | The Jar file must be installed on the customer's servers. | Web Service | |||
License management | When launching an instance, the Java APItransmits the credentials to the Bayesia License Server (BLS) for validation. Therefore, a permanent Internet connection is required. | Credentials are transmitted as part of each HTTPS call. | |||
License management | Completion of the Introductory BayesiaLab Course | ||||
Required software development expertise | Java programming skills | Experience with RESTfull APIs | |||