Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Automated Configuration Synthesis for Machine Learning Models: A git-Based Requirement and Architecture Management System

Abdullatif AlShriaf, Hans-Martin Heyn, Eric Knauss

TL;DR

GRAMS addresses the challenge of maintaining coherent runtime configurations for AI-enabled software by automatically synthesizing configurations from git-stored textual requirements and a compositional architectural framework. The method extends T-Reqs with a TTIM-based traceability layer and a machine-readable YAML intermediary that maps runtime scenarios to OptimizerInput specifications, enabling traceable, dynamic property adjustment for deployment targets. The approach yields two outputs: a machine-readable configuration schema and a set of optimizer inputs, with validation through consistency and semantic equivalence checks. The work demonstrates end-to-end flow from requirements to deployment-ready configurations, supporting runtime optimization and iterative, safe changes in AI-enabled systems.

Abstract

This work introduces a tool for generating runtime configurations automatically from textual requirements stored as artifacts in git repositories (a.k.a. T-Reqs) alongside the software code. The tool leverages T-Reqs-modelled architectural description to identify relevant configuration properties for the deployment of artificial intelligence (AI)-enabled software systems. This enables traceable configuration generation, taking into account both functional and non-functional requirements. The resulting configuration specification also includes the dynamic properties that need to be adjusted and the rationale behind their adjustment. We show that this intermediary format can be directly used by the system or adapted for specific targets, for example in order to achieve runtime optimisations in term of ML model size before deployment.

Automated Configuration Synthesis for Machine Learning Models: A git-Based Requirement and Architecture Management System

TL;DR

GRAMS addresses the challenge of maintaining coherent runtime configurations for AI-enabled software by automatically synthesizing configurations from git-stored textual requirements and a compositional architectural framework. The method extends T-Reqs with a TTIM-based traceability layer and a machine-readable YAML intermediary that maps runtime scenarios to OptimizerInput specifications, enabling traceable, dynamic property adjustment for deployment targets. The approach yields two outputs: a machine-readable configuration schema and a set of optimizer inputs, with validation through consistency and semantic equivalence checks. The work demonstrates end-to-end flow from requirements to deployment-ready configurations, supporting runtime optimization and iterative, safe changes in AI-enabled systems.

Abstract

This work introduces a tool for generating runtime configurations automatically from textual requirements stored as artifacts in git repositories (a.k.a. T-Reqs) alongside the software code. The tool leverages T-Reqs-modelled architectural description to identify relevant configuration properties for the deployment of artificial intelligence (AI)-enabled software systems. This enables traceable configuration generation, taking into account both functional and non-functional requirements. The resulting configuration specification also includes the dynamic properties that need to be adjusted and the rationale behind their adjustment. We show that this intermediary format can be directly used by the system or adapted for specific targets, for example in order to achieve runtime optimisations in term of ML model size before deployment.
Paper Structure (12 sections, 6 figures)

This paper contains 12 sections, 6 figures.

Table of Contents

  1. Background
  2. Introduction
  3. Context: Compositional Architectural Framework
  4. T-Reqs: Text-based Requirements Management System
  5. Applying GRAMS
  6. Quality Assurance Checks in GRAMS
  7. Discussion & Conclusion
  8. Instantiating the Extended Architectural Framework
  9. Target System Configuration Schema
  10. Intermediary Output Format
  11. GRAMS Output
  12. GRAMS Checks

Figures (6)

  • Figure 1: The VEDLIoT Architectural Framework Heyn2023.
  • Figure 2: Extended Architectural Framework for GRAMS: New types added for GRAMS are highlighted in blue
  • Figure 3: Framework traversal algorithm for GRAMS in pseudo-code.
  • Figure 4: A schema-type element that includes the JSON schema that describes an OptimizerInput property related to ethernet latency.
  • Figure 5: Input to GRAMS: A sample JSON schema describing the configuration of a target system that has two properties: ethernet latency and model latency.
  • ...and 1 more figures