Neuton
TinyML^®

Make Edge Devices Intelligent

Automatically build extremely tiny models without coding and embed them into any microcontrollers and sensors.

Free Unlimited Plan

Explore a Novel Approach to Bringing Intelligence and Additional Functionality to Edge Devices

Exceptionally Tiny Models without Loss of Accuracy

Less than

5 KB

is an average model size achieved by our neural framework

Even

8-bit

microcontrollers can be AI-driven and run robust neural networks

Up to

10x

smaller models compared to other existing algorithms

Only

1 iteration

is enough to build an accurate, fast and incredibly tiny model

No-Code Automated Tiny ML Platform

Neuton is a no-code Tiny AutoML platform empowered by a patented Neural Network Framework under the hood. Offering a highly automated and transparent pipeline, Neuton doesn't require a lot of actions from the user's side. It helps to automatically build extremely compact and accurate models without additional compression and natively embed them into 8, 16, and 32-bit MCUs!

Free for Developers

Neuton in Action

Smart Ring as a Remote Control

With this solution, you can use a smart ring to control your TV through various gestures.

Classes: 7
Accuracy: >96%
Inference Time: <1.5 ms
Total footprint: ~11 Kb

Hand Washing Tracking Solution

Our solution can detect and log how many times a user has washed their hands.

Classes: 2
Accuracy: >98%
Inference Time: <1.5 ms
Total footprint: ~3 Kb

Touch-free Interaction for Smartwatches

This ready-made solution allows for various gestures to interact with smartwatches and other wearables.

Classes: 5
Accuracy: >97%
Inference Time: <1.5 ms
Total footprint: ~4 Kb

Logistics: On-device Package Tracking

This solution can recognize seven different events that may occur during the package delivery process.

Classes: 7
Accuracy: >98%
Inference Time: <1 ms
Total footprint: ~4 Kb

Human Activity Recognition on the STMicro ISPU

The solution is capable to recognize daily human activities directly on the ultra-low power ISPU sensor.

Classes: 7
Accuracy: >94%
Inference Time: <3.2 ms
Total footprint: ~11 Kb

Tiny ML Teeth-brushing Tracking Solution

The solution can identify which specific area of your oral cavity needs cleaning and how effectively it is being cleaned.

Classes: 11
Accuracy: >97%
Inference Time: <2 ms
Total footprint: ~4.5 Kb

Remote Control Application

This solution of <4 kb in total footprint can remotely control your media system or presentation slides through gestures.

Classes: 8
Accuracy: >98.8%
Inference Time: <2.5 ms
Total footprint: ~4 Kb

Reproduce it
on your device

Daily Activity Recognition on the Nicla Sense ME

The solution is highly efficient and has a small footprint. It can recognize complex human activity and can be easily replicated on your device.

Classes: 5
Accuracy: >98.5%
Inference Time: <2.5 ms
Total footprint: ~5 Kb

Build Extremely Tiny Models. Run Them on the Smallest IoT Devices.

Neuton – the First Neural Network Framework Designed to Create Models with Minimal Size and without Loss of Accuracy

Neuton is based on a unique patented machine learning algorithm that forgoes error backpropagation and stochastic gradient descent, growing the network structure neuron by neuron. This helps to build neural networks without compromising between accuraсy and size:

with the excellent generalizing capability
with minimal size, often less than 1 Kb
without loss of accuracy
without compression techniques

Read Article

5 Unique Features of Neuton Network Framework

Selective connections

Unique algorithm

Automatic structure growth

No manual search

Constant cross-validation

Selective approach to the connected features

A selective approach to the connected features helps to establish only necessary neuron connections while creating a model, not allowing the structure to grow randomly.

Unique patented global optimization algorithm

Neuton's algorithm enables the platform not to base its computation on error backpropagation and stochastic gradient descent. It provides the opportunity to successfully create compact models, minimizing the likelihood of hitting local minima.

Automatic neuron-by-neuron structure growth

The principle of neuron-by-neuron structure growth helps Neuton to learn from the general features to the most specific ones, and select the optimal model size and high accuracy, required to solve the task.

No manual search for neural network parameters

Neuton eliminates the time-consuming random manual search for neural network parameters (the number of layers and neurons in a layer, type of activation function, batch size, learning rate, etc.), and allows you to quickly and efficiently find the optimal structure.

Constant cross-validation

The step-by-step growth of the neural network makes it possible to cross-validate with the addition of each neuron and to increase the generalizing capabilities of the model.

The models designed by the Neuton.AI platform are optimized to be compact and precise, enabling them to be seamlessly integrated into smart sensors.

All ML analysis is performed within the low-power sensor
Optimized for enhanced battery efficiency, minimizing power consumption
Seamless integration into ultra-low power smart sensors

Community Projects

Activity Recognition

Controlling home electronics using head gestures

CATO, developed by Autonomous Living Technologies, utilizes Neuton.AI's neural networks and ARM's Cortex M4 for efficient gesture recognition. This enables individuals with disabilities to control computers, smartphones, tablets, TVs, motor and track systems, and smart homes seamlessly.

Predictive Maintenance

Predictive Maintenance of Compressor Water Pumps

Ever thought about how water enters the central heating pipes? The secret is the proper operation of water pumps. Learn how to use RSL10 sensors and Neuton to run models for timely pump maintenance.

Smart Human Interfaces

Handmade Drawing Recognition Interface as from my Smartphone

How to quickly develop TinyML models to recognize custom, user-defined drawn gestures on touch interfaces? Check out the project inspired by a gesture recognition feature on a simple smartphone.

Condition Monitoring

Detecting Unstable Electrical Grid with TinyML

Electricity permeates the entire infrastructure of modern cities, so it's really important to monitor and prevent overloads. Explore how to predict electrical grid stability with Neuton and Particle IoT.

Explore All Projects

Evaluate the Uniqueness of Neuton’s Approach by Comparing Benchmarks

Well-Known Magic Wand Case is 33 times Faster with Neuton

Total Footprint (kb)		TFLM	Neuton	Neuton Advantages
Flash	TinyML framework (model + inference engine + DSP)	79.96	5.42	14 times smaller
Flash	Device drivers and business logic	93.47	93.47	14 times smaller
SRAM	TinyML framework (model + inference engine + DSP)	18.2	1.72	10 times smaller
SRAM	Device drivers and business logic	45.69	45.69	10 times smaller
Inference time (us)		55 262	1 640	33 times faster
Holdout validation Accuracy		0.93	0.94	0,7% higher accuracy

Total Footprint (kb)	TFLM	Neuton
Flash
TinyML framework (model + inference engine + DSP)	79.96	5.42
Device drivers and business logic	93.47	93.47
Neuton Advantages	14 times smaller
SRAM
TinyML framework (model + inference engine + DSP)	18.2	1.72
Device drivers and business logic	45.69	45.69
Neuton Advantages	10 times smaller
Inference time (ms)	TFLM	Neuton
Inference time (ms)	55 262	1 640
Neuton Advantages	33 times faster
Holdout validation Accuracy	TFLM	Neuton
Holdout validation Accuracy	0.93	0.94
Neuton Advantages	0,7% higher occuracy

Explore Case More Benchmarks

We compare two models:

The trained TFLM Magic Wand model built into the Arduino SDK

The Neuton model trained on TF dataset from the Google repository

Both models are validated on the same holdout dataset and were tested on the same MCU (Arduino Nano 33 BLE Sense).

Neuton vs. Non-neural network algorithms

This comparison illustrates the ability to achieve compact sizes while maintaining fast inference with the help of neural networks. This opens up wide opportunities for working with more complex tasks where non-neural network algorithms show worse performance.

Case	Holdout Accuracy			Inference Time, us			SRAM, kB			FLASH, kB
Case	Neuton	PME*	More accurate	Neuton	PME*	Times faster	Neuton	PME*	Times smaller	Neuton	PME*	Times smaller
Gearbox Fault Diagnosis	0.84	0.74	12%	56 776	84 080	1.5	3.54	4.95	1.3	9.64	11.15	1.1
Air-Writing Digits Recognition	0.93	0.82	12%	18 172	52 000	2.9	2.57	4.92	1.9	8.91	9.79	1.1
Arrhythmia Diagnostic binary	0.93	0.72	23%	5 212	106 220	20	0.98	1.9	1.9	2.59	8.61	3.3
Arrhythmia Diagnostic multi	0.96	0.77	20%	14 232	N/A	N/A	2.58	9.31	3.6	6.24	16.95	2.7

Realized on 8-bit MCU with AVR GNU Toolchain (7.3.0-atmel3.6.1-arduino7) & 16 MHz CPU frequency
* In this comparison, the PME algorithm was used (pattern matching engine is a distance based classifier), which was automatically built by the SensiML service

More Benchmarks

Gearbox Fault Diagnosis

The goal is to detect broken tooth conditions in the gearbox. The dataset includes the vibration data recorded with 4 vibration sensors that were placed in 4 different directions, and under different load variations.

Air-Writing Digits Recognition

The goal is to correctly identify the digit that was written by a user in the air. The dataset contains 200 samples for each handwritten digit (from 0 to 9). Each sample was recorded for 2 seconds with a frequency of 100 Hz.

Arrhythmia Diagnostic

Binary Classification and Multiclass

The goal is to predict the possibility of cardiac arrhythmias and identify the type of arrhythmia based on the samples of ambulatory ECG recordings.

Embed into Really Tiny Edge Devices

Neuton's silicon agnostic models can be deployed on microcontrollers and small computing devices with limits as challenging as the following characteristics:

Total on-device memory < 100 kB

Energy – ɥW scale

Processor < 100 MHz

8-,16- and 32-bit capacity

Learn How to Embed TinyML

Create Tiny Models without Сompression

Neuton models maintain all of their original characteristics, without any reduction of accuracy. Neuton does not reduce the model size after its creation.

Neuton does not use quantization, pruning, clustering, nor distillation.

With Neuton.ai, you only need to press the button to turn data into a model. Tools like these enable almost anyone to get results.

Massimo Banzi

Co-founder at Arduino

Explore More

Neuton created ultra-tiny deployable machine learning models with footprints of a few kilobytes. While these results fit comfortably MCUs such as STM32, they are essential to run in-sensor neural computing workloads.

Danilo Pietro Pau

Technical Director, IEEE and ST Fellow; System Research and Applications, STMicroelectronics

Explore More

As one of the key players in Arm's partner ecosystem, Neuton empowers developers to successfully go through the entire lifecycle of bringing machine learning to the edge without any deep technical knowledge.

George Anadiotis

Contributing Writer at ZDNet

Explore More

You need to find ways to save time and make model creation run faster, and AutoML tools like Neuton are incredible, as you spend a lot less time on feature engineering, tuning the hyperparameters, evaluating the performance, and other iterations.

Fabio Correa

Director of Business Analytics at Microsoft

Explore More

A strong, thriving ecosystem is critical to delivering new AI solutions for the next decade of compute. The work Neuton is doing will enable even more developers to unlock new opportunities through ML at the edge.

Mary Bennion

Director of Ecosystem at Arm

Explore More

Neuton Pricing and Options

Zero Gravity

Self-service free unlimited plan.

Enterprise Plan

Get assisted and supported

Neuton Services are absolutely free of charge. Get up to $500 in credits for infrastructure costs on Google Cloud Platform.

Training

Unlimited number of models

Digital Signal Processing:

Windowing & Feature Extraction

Feature Selection

Results

Neuton Inference Runner Executable

Downloadable C library for Cortex M0, Cortex M4, Cortex M33, and STMicro ISPU

Analytics Tools

Data Analysis

Model Growth Chart

The total footprint calculation

Model Quality Diagram

Feature Importance Matrix

Confusion Matrix

Infrastructure

Only pay for what you use. Google Cloud Platform Costs

Minimum 100 hours of training included

$0/mo

Use for Free

Build AI-driven IoT devices using the smallest neural networks and taking advantage of professional services from Neuton’s team

To launch large-scale IoT projects, Neuton offers a special Enterprise Plan with individual approach and a full cycle of end-to-end data science services.

Send a Request

Support Library

Video tutorials

Neuton's brief demos and detailed overview

Quick Start

A 5-minute read on how to easily create an AI solution

User guide

Introduction to subscription and Neuton's pipeline

News

11.29.2024

Neuton.AI at Electronica 2024: Presenting Ultra-Compact Neural Networks

At Electronica 2024, Neuton.AI showcased its ultra-compact neural networks, engineered for seamless integration with low-power MCUs and smart sensors in wearables.

11.5.2024

Neuton.AI at TechCrunch Disrupt 2024: Revolutionizing Wearables

At TechCrunch Disrupt 2024, Neuton.AI presented its tiny neural networks, transforming wearables and consumer electronics with low-power MCUs and sensors.

10.28.2024

Neuton.AI at Gitex Global 2024: Showcasing Ultra-Tiny Neural Networks

At Gitex Global 2024, Neuton.AI showcased its ultra-tiny neural networks, enhancing devices from health tech to drones, alongside industry innovations.

Neuton's Divine Journey

To make the world a better place by augmenting human ingenuity through wider adoption and usage of artificial intelligence, while having a transformative impact on the economy, all industries, their associated scientific breakthroughs and overall quality of life.

Compatible with products of major semiconductor vendors