Close

Presentation

A Novel Technique to Support Deep Learning Applications in a Model-Based Embedded Software Design Methodology
DescriptionAs deep learning(DL) applications are getting popular in embedded systems, how to support DL applications in the model-based design methodology becomes a challenging problem.
A popular solution is to specify each DL application with a model. In this work, we propose a novel methodology to leverage a deep learning SDK that is widely used for DL application development. After obtaining candidate mappings of DL networks with the SDK, we perform mapping of dataflow tasks and selection of mapping candidates through a meta-heuristic. The viability and efficiency of the proposed scheme are verified with a real-life example and randomly generated graphs.
Event Type
Work-in-Progress Poster
TimeTuesday, July 11th6:00pm - 7:00pm PDT
LocationLevel 2 Lobby
Topics
AI
Autonomous Systems
Cloud
Design
EDA
Embedded Systems
RISC-V
Security