Israeli chipmaker Polyn's first neuromorphic chip is born

Israel’s Polyn’s neuromorphic chip has been successfully packaged and evaluated, bringing TinyML, or tiny machine learning technology, another step toward becoming more mainstream.

TinyML, or optimizing machine learning (ML) models to run on resource-constrained devices, is one of the fastest-emerging subfields of ML. To achieve the kind of ultra-low-power, high-performance computing required for TinyML (or TinyAI, as it’s sometimes called), engineers are exploring many exciting new technologies.

An overview of TinyML's place in edge computing. Image courtesy of Signoretti

Recently, Israeli company Polyn announced that its latest neuromorphic analog signal processor TinyML/TinyAI processor has been successfully packaged and evaluated.

In this article, we’ll look at the technology provided by Polyn to understand the impact it may have on TinyML as a whole.

Neuromorphic Computing for Artificial Intelligence

In the pursuit of lower power, higher performance AI computing hardware, one of the exciting emerging technologies is neuromorphic computing.

The concept of neuromorphic computing is that the human brain is the most energy-efficient computing device known to man. When trying to run AI applications, creating computing hardware that mimics the biological processes of the brain as closely as possible would be the most efficient. While this sounds like a daunting task, engineers can attempt this approach through a combination of hardware and software.

Implementation of a neuromorphic solution. Image courtesy of Balaji

From a hardware perspective, neuromorphic chips attempt to mimic the brain by acting as circuit elements of neurons, axons, and the weighted connections between them.

To further simulate the brain, this hardware is often implemented with analog circuits, which also helps improve performance and power efficiency. Neuromorphic computing then relies on specialized neural networks, such as spiking neural networks and electrical signal modulation to simulate changes in brain signals.

With this basic understanding, let's take a look at Polyn's new technology.

Polyn’s NeuroSense and NASP technologies

Polyn announced that its proprietary neuromorphic computing chip called NeuroSense has been packaged and evaluated for the first time. NASP technology, short for neuromorphic analog signal processor (NASP) technology, is designed to be a real-time edge sensor signal processor.

NASP demonstration chip. Image courtesy of Polyn

According to Polyn, the technology leverages a unique platform that takes a trained neural network as input and uses mathematical modeling to synthesize the neural network into a true neuromorphic chip. The NASP chip uses analog circuits, where neurons are implemented using operational amplifiers and axons are implemented with thin-film resistors.

They claim that their platform-related needs are fully prepared.

NASP design process. Image courtesy of Polyn

The newly packaged and evaluated NeuroSense chip is implemented in 55nm CMOS technology. In addition, when it acts as an edge signal sensor, it is able to process raw sensor data using neuromorphic computing without any digitization of the analog signal.

For this reason, the company calls it the first neuromorphic analog TinyML chip that can be used directly next to the sensor without the need for an analog-to-digital converter (ADC).

While many technical specifications are still unknown, Polyn’s NASP is said to offer 100uW of power consumption and “twice the accuracy” of traditional algorithms for always-on applications.

Bringing TinyML chips into the future

For now, Polyn is encouraged by its developments, saying the successful packaging and evaluation of its chip validates its technology and the entire NASP system. In the future, Polyn said it hopes to offer the chip to customers in the first quarter of 2023 as a wearable device that integrates photoplethysmography (PPG) and inertial measurement unit (IMU) sensors.