Edge Impulse combines AutoML and TinyML to make AI ubiquitous
Edge Impulse, a San Jose-based startup, aims to bring AI capabilities to microcontrollers that power consumer electronics devices. The company has built a cloud-based platform for developers to build AI models targeting microcontrollers.
Edge pulse combines two popular techniques for bringing AI to microcontrollers – AutoML and TinyML.
AutoML focuses on two essential aspects of machine learning: data acquisition and prediction. The AutoML platform will summarize all the steps that take place between these two phases. Essentially, developers bring their own set of data, identify the labels, and press a button to generate a fully trained model that’s ready to predict.
TinyML is a technique that optimizes machine learning models for embedded devices with limited resources. These on-board devices operate on batteries with limited processing power and memory availability. Traditional ML models cannot be deployed on these devices. With TinyML, models are converted and optimized to run on the smallest unit of an electronic device – the microcontroller.
Targeting microcontrollers for AI execution is a complex and time consuming process. Developers need to acquire data from various sensors connected to the microcontroller. Then, they must preprocess and normalize the dataset before passing it on to a complex neural network. Artificial neural networks need massive computing power provided by high-end processors and GPUs to train models. Once the model is trained and evaluated for accuracy and precision, it is converted and compacted for operation in microcontroller-based devices.
Edge Impulse handled the entire pipeline, from data acquisition to model deployment. It has tools that can be run on Windows or macOS to collect sensor data and ingest it into the cloud-based platform. Depending on the data format, such as time series, audio, and video, developers can choose the appropriate neural network architectures and optimization techniques to train the model. Finally, Edge Impulse converts the model, optimizes it, and generates an artifact that can be directly deployed in the device. All of this is done without writing a single line of code.
Behind the scenes, Edge Impulse builds on the infrastructure, frameworks, and tools commonly used by expert data scientists and ML engineers. It takes advantage of parallelization for data processing and sampling of datasets. TensorFlow is used to build and train neural networks supported by powerful AI accelerators such as GPUs.
TensorFlow Lite, a device-optimized version of TensorFlow, is used to convert and compact the model. Edge Impulse can also generate a web assembly or TensorRT model optimized for NVIDIA Jetson-based Edge devices.
The main benefit of running AI in embedded systems is to make predictions without relying on the cloud or external services. Models become part of the firmware flashed in devices.
Edge Impulse has partnered with silicon companies and DIY electronics kit developers to accelerate the creation of AI models for microcontrollers. It is gaining ground within the community of hobbyist developers and IoT companies to integrate AI into the ecosystem of on-board devices.
Last month, Edge Impulse raised a $ 15 million Series A investment round led by Canaan Partners with Acrew Capital, Fika Ventures, Momenta Ventures and Knollwood Investment Advisory. This funding will contribute to the mission of democratizing ML for millions of developers and engineers deploying to billions of edge devices.