Artificial Intelligence (AI) plays a significant role in various aspects of SAP’s software offerings, enhancing functionality, efficiency, and decision-making capabilities across different business processes. With SAP’s overarching goal to empower organizations to leverage all the technology that they are offering for increased intelligence, SAP has introduced Intelligent Scenario Lifecycle Management (ISLM).
In this blog series, I will give you the overview to ISLM tool and brief guide on how to get the maximum value out of the pre-built scenarios provided by SAP.
This blog is built as a series of several parts and I’ll be updating the links to other parts once they’re published.
Intelligent Scenario Lifecycle Management (ISLM) is a framework facilitating lifecycle management tasks for machine learning scenarios. It serves as a self-service tool, empowering users to address diverse operational needs inherent in machine learning scenarios efficiently. With ISLM, you can train intelligent scenarios and utilize the resulting trained models to obtain precise inference results.
In the above diagram, we have three boxes at the bottom. The first one being the business application, this is where the business user would consume any kind of intelligence. Then at the extreme right, we have the ML runtimes or AI Services which could be either embedded within the same stack in case of HANA ML or it could be, for example SAP BTP. Then right at the center at the core of it, we have ISLM, which i Bridges the gap between the SAP business application and AI services. ISLM is a key enabler for our customers, our partners and our internal teams to create these AI use cases, and not just create but also to operate the entire life cycle of it.
The categorization of scenarios within a business application depends on the specific machine learning scenario involved:
- Embedded: In this methodology, a business application such as SAP S/4HANA operates within the same stack as its machine learning counterpart, SAP HANA Machine Learning, utilizing either the SAP HANA Automated Predictive Library (APL) or the SAP HANA Predictive Analysis Library (PAL) for analytics tasks. APL provides business analysts with data mining capabilities through an Automated Analytics engine, facilitating the development of predictive modelling processes. On the other hand, PAL offers advanced analytics algorithms tailored for data scientists. These tools can be effectively applied to address various use cases such as forecasting, trend analysis, and more
- Side by Side: In this setup, a business application like SAP S/4HANA operates on a distinct stack from its machine learning provider, such as SAP Data Intelligence. Remote machine learning is employed for advanced use cases, such as image recognition, sentiment analysis, and deep learning for natural language processing, leveraging neural networks
ISLM framework consists of two SAP Fiori applications:
Intelligent Scenarios and Intelligent Scenario Management. These applications allow you to create and manage the lifecycle of intelligent scenarios.
ISLM comprises of the following functionalities:
- Display: You can view the details of the intelligent scenarios that are displayed, such as package, scenario type, inputs and outputs, API details, and so on.
- Create: You can create an intelligent scenario of the type Side-by-Side and Embedded for your specific business need.
- Train: You can train the scenario to generate a trained machine learning model or artifact for the business consumption.
- Deployment: You can deploy the trained machine learning models for inference consumption. This functionality applies to Side-by-Side scenarios only.
- Activation: You can activate the trained machine learning model for which you want to get the inference results.
- Inference Results – You can get the predictions that is generated from the trained models.
Here are some of the advantages offered by ISLM:
- Serves as a unified entry point for Citizen Data Scientists to conduct machine learning lifecycle management operations on both embedded and remote machine learning models.
- Offers support for both On-Premise and SAP Cloud solutions in conjunction with the SAP solution it integrates with, such as SAP S/4HANA.
- Delivers comparable functionality to the previous SAP solution, SAP Predictive Analytics Integrator (PAi).
- Establishes a standardized approach for managing training, deployment, and activation across all intelligent scenarios.
- Supports both Embedded and Side-by-Side scenario configurations.
- Provides the flexibility to create custom intelligent scenarios and oversee the model training, deployment, and activation directly within ISLM.
- Coordinates machine learning operations in the embedded approach utilizing SAP HANA predictive libraries like APL and PAL, as well as in the Side-by-Side approach with remote providers such as SAP Data Intelligence.
Intelligent Scenarios App
- Here you can check the standard pre-built scenarios provided by SAP. Custom scenarios can also be created from this app.
- You need to add a Machine Learning Model here
- In the next screen you need to populate the Technical fields – Training Dataset, Apply Dataset, Target
- Once done, you have to review the model and publish it. The technical details can be found here:
Intelligent Scenario Management App
- All the standard Intelligent scenarios provided by SAP can be seen here. You can select any scenario to use it to train the model
- One or multiple filters are available for your selection. Once filters are selected, you need to train the model
- After the model is trained, you need to activate it. Now the Intelligent scenario is active and ready for use
Conclusion
This blog post should help you to understand in detail about the ISLM tool and it’s offerings. Also, you will get brief idea on SAP’s strategy in regards to Intelligent solutions to business processes.
