When
2 buzz-words meet, it generates an ambitious session.
Put
artificial intelligence in an IoT project is an undeniable added value. And
bringing artificial intelligence (understand the execution of a predictive
algorithm) closer to usages is a major issue when developing an AI project in
an IoT context. This session explores the issues to be overcome, starting from
the principles of Edge Computing and the constraints it adds to AI. The session
will deal with hardware concerns (GPU, TPU, etc.), Deep Learning and Data
Science frameworks (TF, ONNX, quantization, etc.) and containerized
architecture. It will focus espcially on Azure IoT Edge. This session will show
some demos and give some feedbacks from an internal project: HappyFace.