Schedule
Timezone: PDT
Running TF Lite on Microcontrollers without hardware in Renode
Michael GIELDA, VP Business Development
Antmicro
The incorporation of Machine Learning into increasingly smaller, low-power devices is opening up new use cases, and with TensorFlow Lite Micro, ML models for actions such as keyphrase detection or gesture recognition can be deployed on tiny embedded and IoT devices.
However, developing and testing software in embedded systems could be challenging due to difficulty in setting up and configuration of complex environments, deterministic and repeatable testing with various input data.
The open source Renode 3 simulation framework from Antmicro allows TinyML developers to overcome those challenges, enabling the simulation of physical hardware systems, including the CPU, peripherals, sensors, environment and – in case of multi-node systems – wired or wireless medium between nodes. Using Renode, Continuous Integration of your TinyML application can be performed to make sure it continues to work as development progresses; performance metrics have also been added recently to allow comparative analyses. Renode was used to port TF Lite Micro to several RISC-V and Arm platforms and add support for the Zephyr RTOS (recently described in a TensorFlow Lite blog note 5). In his talk, Michael Gielda will present the advantages of using simulation and Continuous Integration for TinyML development as well as explain how to run TensorFlow in Renode and how to build your own application.
Michael GIELDA, VP Business Development
Antmicro
Michael Gielda is VP Business Development and co-founder at Antmicro, a software-driven tech company developing modern edge AI systems for various branches of industry including aerospace, medical systems and robotics. He is also Chair of Outreach at CHIPS Alliance and Vice Chair of Marketing at RISC-V International. Michael holds a B.Sc. in Computer Science from Poznan University of Technology, where he remained in an undergraduate research position, working with drones, intelligent transportation and wireless sensor networks, before going on to found Antmicro.
Timezone: PDT
Building Products using Edge AI / TinyML on MCUs
Stuart FEFFER, Co-founder and CEO
Reality AI
Reality AI is the leading product development environment for Edge AI / Tiny ML on MCUs. Tools that generate TinyML models without code have become commonplace, but there is much more you can use machine learning to do. We’ll show you how to use AI to drive sensor selection and placement, how to use ML to determine minimum component specifications, how to minimize the cost of data collection, and also how to generate sophisticated, explainable ML models based on sensor data – automatically. We will use case studies to explore the difference between doing projects and building products, showing examples from Reality AI Tools 4.0.
- How to generate sophisticated Edge AI/TinyML models automatically
- How to use AI to optimize sensor selection and placement
- How to use AI to set minimum component specifications
- How to minimize the cost of data collection
Stuart FEFFER, Co-founder and CEO
Reality AI
Stuart Feffer is co-founder and ceo of Reality AI. He’s a multiple-time founder, with his last startup acquired by Wells Fargo. Stuart holds a PhD from UC Berkeley and a BA from The University of Chicago.
Schedule subject to change without notice.