EDGE AI FOUNDATION
STMicroelectronics has introduced a new bio-sensing chip for the next generations of healthcare wearables like smart watches, sports bands, connected rings, or smart glasses. The ST1VAFE3BX chip combines a high-accuracy biopotential input with ST’s proven inertial sensing and AI core, which performs activity detection in the chip to ensure faster performance with lower power consumption.
Demonstrates First-of-Its-Kind Ultra-Efficient Small Language Models (SLMs) and AI Agents
Aizip, in close collaboration with Renesas, announced today the demonstration of ultra-efficient small language models (SLMs) and compact AI agents on Arm-based micro-processor units (MPUs) for a wide range of applications in edge markets. This advancement paves the way for efficient and effective human-AI interactions in home appliances, enterprise kiosks, and many other edge devices.
Large language models (LLMs) have revolutionized the AI landscape, endowing AI systems with logic and reasoning capabilities beyond simple sensing and perception. By leveraging LLMs and recent advancements in multi-modal representation, AI agents can now interact with their environments to utilize tools or perform tasks based on complex and often ambiguous human commands. However, these advanced AI systems typically require substantial effort to train and significant resources to deploy.
Aizip’s mission is to enable pervasive intelligence by building ultra-efficient, robust, and scalable AI models that can be deployed anywhere, anytime. Aizip pushes the boundaries of efficient AI, uncovering key insights such as data-centric efficiency and AI-design automation. Leveraging its expertise in developing efficient and robust edge models, Aizip has now created a series of ultra-efficient small language models (SLMs) and AI agents, named Gizmo, ranging in size from 300 million to 2 billion parameters. These models support diverse platforms, including MPUs and application processors for a broad range of applications.
The open-source ML Systems book serves as an educational resource that aims to make the principles and applications of ML systems accessible to a wide range of individuals. It focuses on TinyML, which aligns perfectly with the mission of the TinyML Foundation. The book includes numerous use cases and examples in the labs, all based on TinyML.
While much of the tech world remains fixated on the latest large language models (LLMs) powered by Nvidia GPUs, a quieter revolution is brewing in AI hardware. As the limitations and energy demands of traditional deep learning architectures become increasingly apparent, a new paradigm called neuromorphic computing is emerging – one that promises to slash the computational and power requirements of AI by orders of magnitude.
The International Center for Theoretical Physics will be holding a workshop in Sao Paulo, Brazil from July 22 – July 26.
This hands-on workshop focuses on TinyML applications relevant to Latin American researchers, providing training on commercially available hardware optimized for embedded ML deployment. By making TinyML more accessible, especially in the Global South, this workshop will empower researchers to develop localized solutions that benefit their communities.
For more information please see – https://indico.ictp.it/event/10499
SILICON VALLEY, Calif. , June 24, 2024 /PRNewswire/ — From the TinyML Summit in Milan Italy, embedUR systems Inc., a leader in embedded systems and Edge AI, and a tinyML Foundation strategic partner, is thrilled to announce the launch of ModelNova, a groundbreaking software hub catering for Edge AI solutions.
ModelNova is a model zoo for pre-trained AI models, optimized for different software frameworks and a variety of low-power hardware platforms with and without native AI acceleration. This innovative platform streamlines Edge AI product development, enabling rapid prototyping and deployment of intelligent applications on edge devices, in a fraction of the time it used to take to develop Edge AI solutions from scratch.
ModelNova addresses a significant challenge faced by developers: the complexity of selecting, creating, training, porting, and optimizing AI models for different hardware platforms, especially low-power IoT devices.
Edge AI, empowered by the recent advancements in Artificial Intelligence, is driving significant shifts in today’s technology landscape. By enabling computation near the data source, Edge AI enhances responsiveness, boosts security and privacy, promotes scalability, enables distributed computing, and improves cost efficiency.
The tinyML Foundation has partnered with Wevolver to create a detailed report on the current state of Edge AI. This document covers its technical aspects, applications, challenges, and future trends. It merges practical and technical insights from industry professionals, helping readers understand and navigate the evolving Edge AI landscape.
Academia is the AI leadership of tomorrow and they are not only driving AI research, applications and a talent pipeline but also present a great opportunity for jobs, internships and mentorships. They have always been a strong pillar of the tinyML Foundation!
As our next phase of strengthening our bridge with academia, we are joining forces and building out a Discord server for a worldwide conversation on tinyML and all things AI in resource constrained environments… jump in!
Thanks to the group effort of the Datasets and Benchmarking Working group, The tinyML Foundation is announcing the development of new datasets designed for resource constrained AI and ML applications, which will be freely available to use and contribute to by the community worldwide.
The purpose of this effort is to drive higher quality training data and speed the development and deployment of high quality solutions and provide more comparable benchmarks for developers and integrators of this technology.
The first datasets to be made available will be Visual Wake Words, via GitHub shortly. This will be followed by other key datasets that the tinyML Foundation will curate and develop with our community.
Are you interested? Join the conversation on our discord channel #datasets or email us at datasets@tinyml.org.
In a bid to address the rising concern of traffic-related fatalities and injuries, the Global tinyML Traffic Hackathon took place in partnership with the City of San José’s Vision Zero program. With pedestrian fatalities constituting a significant portion of traffic-related deaths, the hackathon aims to leverage the power of energy efficient Machine Learning (tinyML) to detect pedestrians and create innovative solutions for enhancing traffic safety.[1] This article delves into the key details of that hackathon, the technology engineer’s utilized and its potential impact on traffic safety.