AI for Embedded Systems [EN]

We will show you how to assess and evaluate an embedded system to ensure its capability to run a specific model.

You will also be introduced to the most important tools and methods used when integrating models on embedded systems. You will learn what you need to pay particular attention to, especially in terms of software compatibility, development effort, performance, maintainability and robustness.

In order to achieve maximum performance and efficiency, we will teach you advanced techniques for optimizing machine and deep learning models. Finally, we will introduce you to setting up a consistent and high-quality development workflow (MLOps) that will guarantee the success of your projects.

Content

• Introduction to machine learning, deep learning and artificial intelligence
  • Overview and differences between the terms
  • Differentiation and areas of application
• State-of-the-art models and architectures
  • Overview of current models for time series and image data processing (time series processing and computer vision)
  • Practical examples for various fields of application
  • References to further courses in specific areas
• System evaluation and architecture for embedded AI
  • Criteria for the evaluation and selection of embedded systems for the use of AI
  • Important metrics for system evaluation and methods for determining them
  • Overview of performance classes and hardware architectures of embedded systems and their possible applications for AI
• Development and deployment of AI models on embedded systems
  • Available frameworks and tools for model integration
  • Best practices for development, maintenance, performance optimization and compatibility
• Optimization methods for AI models
  • Techniques for increasing the performance and efficiency of machine and deep learning models on embedded systems
• MLOps for embedded AI
  • Introduction to tools and processes for an end-to-end development workflow
  • Integration of the AI development process with the embedded system workflow
  • Designing a minimal and efficient MLOps workflow for embedded AI applications

Duration

2 days

Target audience

• Embedded Software Engineers
• Software Architects
• Head of Embedded Engineering
• CTOs/CIOs

Prerequisites

• Basic knowledge of the architecture and functionality of embedded systems
• Basic knowledge in the field of machine learning
• Basic knowledge in hard & software systems

This course is organised by Software Quality Lab Academy and being held by our partner Danube Dynamics.