WEDNESDAY October 18, 5:30pm - 6:30pm | Crystal
EVENT TYPE: PANELMachine Learning for Embedded Systems: Hype or Lasting Impact?
X. Sharon Hu - Univ. of Notre Dame
Machine learning (ML), especially deep learning and neural networks, is attracting a lot of attention in a number of application domains including embedded systems. ML can offer alternative ways to exploit data and usage patterns for embedded system design. However, one may argue that the general ML ideas have been used in embedded system design for a long time. Furthermore, besides challenged by limited resources such as time, energy and memory, many embedded systems are also required to provide guaranteed services such as timing, security, and reliability. These requirements are intrinsically in conflict with the use of ML. It then begs the question whether ML can really lead to fundamental advances for embedded systems or will remain limited to a few clearly defined application scopes. In this panel, leading embedded system experts will “declare” and support their respective positions and challenge others’ opposing positions. Audience is invited to join this active debate, which should lead to insights on the true potentials and limitations of exploiting ML for embedded systems.