Speakers
Description
Area presentation: "Resource-Aware Machine Learning at Lamarr: A Guided Tour"
In this talk, we offer a guided overview of the resource-aware machine learning (RAML) research taking place at the Lamarr Institute. RAML aims to make machine learning systems not only accurate, but also efficient in terms of energy, latency, and computational resources. We highlight ongoing efforts within the institute and introduce some of the key researchers. Short spotlight presentations from three ongoing PhDs are presented to explore opportunities for potential collaboration. The session concludes with a brief outlook on upcoming challenges and how the Lamarr Institute aims to address them.
Break-through result: "Towards Anytime Models: A quick overview of recent results in Lamarr"
Anytime models offer flexible inference under resource constraints by producing usable predictions even when computation is interrupted. This talk outlines recent progress at the Lamarr Institute in developing such models across diverse applications. Starting from classical ideas like early exits, routing, and input-dependent computation, we show how these techniques can be extended to build full-fledged anytime systems. We highlight recent results in collaboration with the Area of Industry & Production and in the realm of image recognition, while closing the talk with a roadmap toward integrated energy-aware evaluation.
| Area Presenter | Jian-Jia Chen |
|---|---|
| Spotlight Presenter | Sebastian Buschjäger |
