PORTLAND, OREGON, May 14, 2024 — SensiML Co., Ltd., a leader in AI/ML software for the IoT and a subsidiary of QuickLogic, today announced that it is disrupting the TinyML market by offering the first complete open source AutoML solution for edge AI/ML application development. Popular Analytics Studio application. While open source models are already prevalent in widely adopted AI libraries such as TensorFlow and PyTorch, comprehensive AutoML development tools targeting IoT edge devices have so far eluded us.
AutoML (automated machine learning) simplifies and significantly speeds up the process of creating machine learning models. This makes machine learning more accessible to developers without data science expertise. Building ML models for IoT microcontrollers and edge SoCs is particularly complex because it requires a combination of data science and embedded code optimization for devices with limited memory and compute power. AutoML can help you overcome these challenges.
SensiML's pioneering open source product promises to bring enhanced creativity, innovation, and AI code transparency to the global community of IoT device developers and is a fast-growing market predicted by ABI Research. By 2027, SensiML's Analytics Studio brings intelligent sensing capabilities to a wide range of IoT edge devices, such as the following real-world application examples.
- Wearable devices and clothing that analyze and guide proper human movement and ergonomics in real time
- Predictive maintenance and anomaly detection sensors that locally recognize and react to failures in factory/plant machinery, pumps, and valves.
- Building automation and security endpoints with acoustic event detection, keyword recognition, and speaker identification
Until now, IoT device developers working on often first-time AI/ML projects have had to navigate a fragmented market of proprietary tools with varying capabilities and unclear roadmaps. The open source release of SensiML's Analytics Studio marks a significant milestone for the IoT Edge AI software tools industry, delivering:
- Generating a platform-independent model: SensiML's plug-in-style open source architecture supports a wide range of MCUs, AI/ML acceleration SoCs, and AI engines, with flexible tools that are not tied to any specific vendor, chipset, or inference engine. Empower developers to build ML datasets with confidence.
- Time series sensor input: Supports all possible time series sensors including microphones, accelerometers, gyros, IMUs, load cells, strain gauges, and PIR sensors. Sensor fusion algorithms can be used to mix inputs to create more complex models.
- Rapid innovation: The rapid evolution of AI/ML requires an open source approach to leverage the expertise of the broader developer community and accelerate key innovations such as advances in generative AI, synthetic data, and edge learning.
- Flexibility: Analytics Studio supports multiple model development mechanisms, from point-and-click AutoML-powered model generation, to code-free GUI-based modeling with full pipeline control, to fully programmatic Python SDK model creation. Masu.
- Scalability: Analytics Studio provides model generation for basic feature-based models, regression models, classic ML, and deep learning neural networks. The rich library of over 80 feature generators also includes the ability to easily add custom transformations, filters, features, and classifiers, making it easy for community developers to enhance functionality.
By moving to a dual licensing model with an open source option, SensiML offers its IoT Edge AutoML solution as a foundational code base built over seven years, allowing the broader developer community to collaborate on improvements and contributions. That's what I'm trying to do. With community support, SensiML aims to extend Analytics Studio to include:
- Developing and tuning generative AI models
- Augmenting synthetic datasets
- Local LLM support
- Object recognition from image and video data streams
- Tuning and learning enhanced edge models
- Further integration/optimization of MCU, MPU, NPU, GPU
- Add pre-trained model templates for real-world use cases
New and existing users have the flexibility to choose between SensiML's open source version of Analytics Studio or a fully managed and supported SaaS cloud service implementation based on the same core technology.
Chris Rogers, CEO of SensiML, said: “Four years ago, our parent company, QuickLogic, introduced the first open source eFPGA solution. “We are leveraging this success to democratize edge AI/ML development with robust tools. This open source initiative will accelerate edge AI/ML adoption, benefit end-user flexibility, and increase the value of SensiML's SaaS growth and private label tools for a growing number of industry partners. Sho.”
availability
SensiML will publish a public GitHub repository and AutoML engine documentation early this summer. Developers interested in receiving updates and contributing to this pioneering technology can sign up at the following URL: https://sensiml.com/blog/opensource.
About SensiML
SensiML, a subsidiary of QuickLogic (NASDAQ: QUIK), provides cutting-edge software that enables ultra-low-power IoT endpoints with AI and transforms raw sensor data into meaningful insights on the device itself. . The company's flagship solution, SensiML Analytics Toolkit, provides an end-to-end development platform spanning data collection, labeling, automated algorithm and firmware generation, and testing. The SensiML Toolkit supports a growing number of hardware applications, including Microchip's 8/16/32-bit MCUs, Arm Cortex-M class and above microcontroller cores, Intel x86 instruction set processors, and AI/ML-optimized heterogeneous core SoCs. support wear. For more information, please visit: https://sensiml.com.
Source: SensiML
