Please note that Neuton no longer supports Internet Explorer.
We recommend upgrading to the latest Microsoft Edge, Google Chrome, or Firefox.
Close
A b o u t
N e u t o n
Neuton is a disruptive Neural Network Framework and Automated Machine Learning (AutoML) solution that is far more effective than any other framework, non-neural algorithm or AutoML product available on the market. It makes Artificial Intelligence (AI) available to everyone.

Very easy to use and no coding required to build your AI solution
No special background in IT, AI, and Machine Learning (ML) is needed
Resulting models are very compact, fast and accurate with fewer training samples required
Intelligence makes the world a better place
Take the Survey
After you try Neuton, please take the time to complete the survey. Your opinion is very important to us and will allow us to make the product better.
Neuton’s Divine
Journey
To make the world a better place by augmenting human ingenuity with wider usage and adoption of artificial intelligence while having transformative impact on economy, every industry, scientific breakthroughs and the quality of life.
Neuton's Story
Play video
Why Neuton
A lot of companies, especially those from the small and midsize segments, face many or all of the following significant challenges when attempting to implement AI:
Lack and high cost of data scientists
Rigorous tech and math skills necessary to build AI
Complexity of AI solutions
High cost of implementation and support
Unfortunately failing to address these challenges results in an inability to leverage AI to stay competitive and fuel growth leads to a very slow Go-To-Market.
Neuton solves all of the above problems via a streamlined workflow and by automating the entire process of AI model creation.
Working with Neuton is like having an unlimited number of data scientists at your disposal packaged within an affordable monthly subscription. As a result, Neuton makes the process of AI solution creation and support easy, fast and inexpensive.
10 times faster Go To Market
50% less cost of implementation
50% less cost of support
How is it Different?
Neural Network Framework
Unlike other solutions on the market, Neuton Neural Network Software is not based on any derivative or pre-existing frameworks or non-neural algorithms. It was inspired by ancient divine wisdom, invented and brought to the world by our team of scientists and engineers. Neuton successfully competes with and outperforms other frameworks and non-neural algorithms.
AutoML solution
Neuton AutoML is based on Neuton Neural Network Software and automates the whole process of machine learning with the state of the art technology. Its simplicity and ease of use makes artificial intelligence and machine learning models available to all user demographics irrespectively of their technical and scientific background. Unlike other AutoML solutions you do not need to be a “certified pilot” to operate Neuton.
SaaS Google Cloud AutoML solution
Very easy to use and no coding required
No special background in IT, AI and ML is needed
Predictions via WEB interface or REST API
Downloadable models for local use without Neuton
Far fewer cloud infrastructure requirements and therefore cost for the training and for the ready solution pool in comparison to competitors
Unique characteristics of the resulting model:
Self-growing and learning – there is no need to work on layers and neurons, just prepare a dataset and a neural network structure and a model will be grown automatically
Requires fewer training samples
Up to 100 times smaller than those built with existing frameworks and non-neural algorithms
Up to 100 times faster than those built by competitors
Up to 100 times less in neurons and coefficients than those currently on the market
So compact and fast that it can work on microcontrollers, smartphones etc.
Results in no overfitting
Far fewer absolute and relative errors in the validation samples than those built by competitors
The highest accuracy on a given dataset in comparison to the other solutions on the market
Creating neural networks in artificial intelligence is like driving a car. Many years ago, car owners knew their vehicles inside out, so that they could fix or modify them to improve their driving experience. Everybody was a car mechanic at the time.

Nowadays people drive cars and do not even realize what the car consists of. All they care about is smooth driving experience, design and safety. They just monitor speed, mileage, and a gas level. Maybe, if you build your own custom car for speed racing, you have to be a certified mechanic.

So why do you need to know and care about things like layers, epochs, iterations, one-hot-encoding, backpropogation or machine learning cloud services terms? Well, with Neuton you don’t. You just use the Intelligent Agent and enjoy your AI ride and we place all unnecessary technical details under the Neuton’s hood to make your experience easy and not dependent on your technical background. Rest assured Neuton has the best AI engine available on the market.
Benchmarks
To benchmark a model size correctly and allow for a clear comparison against both Neural and Non-neural solutions, Neuton has two measurements: level of accuracy relative to competitors and maximum accuracy achieved.
Maximum accuracy
Accuracy similar to other frameworks
Neurons
  • Neurons
  • Coefficients
  • Samples
  • Max Error
  • Metric
  • Metric value
  • Size, KB*
* - Size of models without metadata
   •   Similar results are produced with such frameworks as TensorFlow, Caffe2, CNTK, Torch, Theano
   •   All datasets were identically preprocessed WITHOUT FEATURE ENGINEERING performed
   •   Neuton has two benchmarks for “Neural”: with accuracy similar to competitors and with maximum accuracy achieved
MAE
  • MAE
  • MAE max
  • MAE min
  • RMSE
  • R2
  • RMSLE
  • Size, KB*
Accuracy
  • Accuracy
  • AUC
  • Gini
  • Prcision Weighted
  • Recall Weighted
  • F1 Weighted
  • Precision Macro
  • Recall Macro
  • F1 Macro
  • Size KB*
MAE
  • MAE
  • AE max
  • AE min
  • R2
  • RMSE
  • RMSLE
  • Size, KB*
Accuracy
  • Accuracy
  • Prcision Weighted
  • Recall Weighted
  • F1 Weighted
  • Precision Macro
  • Recall Macro
  • F1 Macro
  • Size KB*
* - Size of models without metadata
Similar results are produced with such frameworks as TensorFlow, Caffe2, CNTK, Torch, Theano
In order to benchmark algorithms against each other, all datasets were identically pre-processed.
Accuracy MAE to
Sample Size
Accuracy to Percentage of full training dataset
Mean Absolute Error to Percentage of full training dataset
Binary classification problem
Multiclass classification problem
Binary classification problem
Regression problem
Regression problem
Try Benchmark Models
Built with Neuton
Try our machine learning models on the following datasets. You can either download the archives and follow the instruction in the readme file to realize the benchmarks, or make predictions using the web interface.
The archive contains a model in a binary format, Python scripts for using and evaluating generated neural networks and a readme.txt file with instructions on how to use the model. The archive also contains information about the Neural Network model configuration, which Neuton was compared against.

Please click here to read detailed descriptions of each dataset.
TITANIC
Titanic.zip
Use Neuton based model to predict survival on the Titanic
SCHOOLS
Schools.zip
Use Neuton based model to predict school rating
hr
HR.zip
Use Neuton based model to predict attrition of your valuable employees
BOSTON HOUSING
Boston.zip
Use Neuton based model to predict median value of owner-occupied homes
UCI AIR QUALITY
Air.zip
Use Neuton based model to predict benzene concentration
Neuton Pricing
and Options
The online version of Neuton has far fewer infrastructure requirements, and is up to 50 times less costly than what you would normally pay for infrastructure with our competitors.

Rest assured that with Neuton you are receiving the best possible product and a great price.
Features
Free Trial (If eligible)
Predictions via REST API
Predictions via Web Interface
GPU Training and Model Creation
Downloadable Models
Gravity
$0
+ usage Monthly
Free Trial (If eligible)
Up to $500 credits or for 7 days
Predictions via REST API
5000 API Requests Per Month
Predictions via Web Interface
100 Rows in CSV file for web prediction
GPU Training and Model Creation
Unlimited
Downloadable Models
-
First Law
$2,699
month + usage Monthly
Free Trial (If eligible)
-
Predictions via REST API
Unlimited
Predictions via Web Interface
Unlimited
GPU Training and Model Creation
Unlimited
Downloadable Models
-
Second Law
$1,999
month + usage 1 year commitment
Free Trial (If eligible)
-
Predictions via REST API
Unlimited
Predictions via Web Interface
Unlimited
GPU Training and Model Creation
Unlimited
Downloadable Models
-
Third Law
$3,699
month + usage 1 year commitment
Free Trial (If eligible)
-
Predictions via REST API
Unlimited
Predictions via Web Interface
Unlimited
GPU Training and Model Creation
Unlimited
Downloadable Models
Unlimited
Google Infrastructure costs:
Cloud Storage, per GB per month* – $0.04
GPU Training, per hour - $0.70
Prediction, Hosted Model, per hour - $0.04
Estimated monthly infrastructure cost**:
$46.00
* Includes Network Egress, Backup, Monitoring

** Estimated monthly infrastructure costs are based on 24 hours of GPU usage for training, 30 Days of hosting for one model, uploaded data size (training datasets) ~1 GB. Pricing for Neuton’s First Law, Second Law and Neuton’s Third law excludes infrastructure costs.
Included Support – Email support with 2 Business Days resolution
More options
ENTERPRISE
Run Neuton on-premises
Upon Request
Embedded
Augment your device with AI capabilities
Upon Request
Free Trial
Try Neuton Gravity for free
New customers to Google Cloud Platform (GCP) and Neuton are eligible to up to $500 in credits to try Neuton for free. In order to be qualified to redeem the maximum amount of available edits, customers will need to adhere to the following criteria:
You must register with a corporate or enterprise email domain. No personal email accounts allowed (e.g. Gmail, Yahoo)
You must have an active billing profile within the Google Marketplace.
You may only redeem the credits for Neuton once. If you have already redeemed credits with other Google Cloud Partners, you are not eligible to redeem them again
This trial lets you try out this solution without paying for infrastructure up to $500.00, or for a maximum of 7 days, whichever comes earlier. During the trial, you will be billed for infrastructure fees and then credited at the same time for those infrastructure fees up to $500.00 based on the available credits.
When your trial use exceeds $500.00 in infrastructure fees or 14 days, whichever comes earlier, you will be charged infrastructure fees unless you stop or terminate the subscription. You may cancel the trial at any time by cancelling your plan prior to the end of the trial.
Neuton Currently Features
Automated creation of the ML model structure to utilize highest predictive power possible
Automated training and growing of the model with the best characteristics
Automated dataset preparation (preprocessing)
Automated feature engineering
Predictions via Web interface
Predictions via REST API
Downloadable models for local use without Neuton
Embedded models capability
Python and other API for local usage
Feature importance capability
GPU Support for training
Online version
Problems Neuton Currently Solves
Regression. Prediction of continuous numeric value (Ex.: temperature or real estate prices)
Binomial classification. Prediction of one of the two outcomes (Ex.: ‘Yes’/’No’, ‘White’/’Black’, 1/0)
Multinomial classification. Prediction of a value that may have 3 or more categorical outcomes (Ex.: next best offer prediction from 10 categories of products)
Positioned
for the Future
Support of Image/Audio/Video
Multi-GPU support
Model interpretation
Data and model automatic visualization
Multi host training
Video tutorials
How can I create an AI solution with Neuton?
watch video
How can I sign up and register in Google Cloud Platform?
watch video
FAQ
Got questions? Find answers here.
See if we have already answered them. If not, please don't hesitate to drop us a line and we will get back to you shortly!
Release notes
Version 2.2.0
Release Date: December 1, 2019
Use Cases
Empowered by unique differentiators, Neuton opens exciting new opportunities and offers a number of practical use cases.
EXPLORE
Media About Neuton
About Us
Bell Integrator is a global consulting and technology services provider driving innovation and enabling digital transformation.
Our team of scientists accumulates more than 700 years combined experience as scientific researchers while successfully solving complex algorithmic problems in the following areas: neural networks in artificial intelligence, machine learning cloud services, augmented reality, video analytics, internet of things, blockchain.
10
delivery centers
in 6 countries
2500+
company
employees
300+
high profile
customers
16
years
in IT business
Contact Us
Get in touch to learn more
Contact Us