Embedded Machine Learning


This seminar covers several topics, which are briefly presented here. In this seminar, the students discuss the latest research findings (publications) on the topics below. The findings are summarized in a seminar paper and presented to other participants in the seminar. Your own suggestions for topics are welcome, but not required. The seminar can be completed in German or English.

Machine learning on on-chip systems

Machine learning and on-chip systems form a symbiosis in which each research direction benefits from advances in the other. In this seminar, the students discuss the latest findings in both research areas.

Machine learning (ML) is finding its way more and more into all areas of information systems - from high-level algorithms such as image classification to hardware-related, intelligent CPU management. On-chip systems also benefit from advances in ML. Examples of this are adaptive resource management or the prediction of application behavior. Conversely, however, ML techniques also benefit from advances in on-chip systems. An example of this is the acceleration of training and inference of neural networks in current desktop graphics cards and even smartphone processors.

The students are able to independently research the state of research on a specific topic. This includes finding and analyzing, as well as comparing and evaluating publications. The students can prepare and present the state of research on a specific topic in writing.

Machine Learning for Optimization of Embedded Systems

Sophisticated resource management becomes a pressing need in modern embedded systems, where many connected devices collaborate towards achieving a specific goal and each of these devises may execute several  applications. The goal of resource management is to allocate resources  to applications while optimizing for system properties, e.g.,  performance, and satisfying its constraints, e.g., temperature. To  achieve full potential for optimization, state-of-the-art resource management has employed machine learning methods to learn relevant knowledge about the system with its two parts; hardware  and software, and exploited this knwoledge within its decision making process.

In this seminar, we will discuss the different machine learning approaches that are proposed to support resource management decisions. 


Please register in ILIAS to participate.