Neuton (network)

From HandWiki

Neuton is a neural network framework and automated machine learning (AutoML) solution developed by Bell Integrator. It is based on its own proprietary algorithm protected by a patent issued in 2019.[1]

Overview

Neuton allows users to build, train and deploy neural network models to solve problems of multivariate regression and classification. Neuton is available as an enterprise solution to run on-premises or an online service on Google Cloud Platform. Predictions are made via a REST API or Web interface. The resulting model can be used as a service deployed in the cloud or downloaded for local use.[2] Benchmark tests verify that Neuton outperforms competitors by several orders of magnitude.[3]

Features

Neural Network Framework

  • Automated Neuton model architecture creation[4]
  • Automated hyper parameters configuration[5]
  • Regression and Classification problem solving
  • No limitations on dataset size for training
  • No overfitting
  • Automated training completion upon reaching optimal model and predictive power while avoiding overfitting[6]

Auto ML

  • Automatic and manual problem type definition (regression or classification) based on dataset analysis
  • Automated dataset preparation (preprocessing)
  • Automated feature engineering
  • Automated training
  • Automatic training completion upon user input thresholds reach[7]
  • Stop and resume training capability[8]
  • CSV datasets support

Preloaded datasets

  • Models quality metrics visualization[3]

Predictions

  • Capability to download trained models in HDF5 format with Python API for local usage along with script examples on how to use Neuton models offline
  • Web Interface for predictions on trained models[8]
  • Predictions via REST API
  • No limitations on input data size for predictions[9]

Infrastructure

  • Cloud Based solution
  • SaaS solution
  • GPU Support for training[10]
  • Automatic and seamless provisioning and release of infrastructure necessary to perform training and web hosting for predictions[11]

References

  1. Munford, Monty. "Bell Integrator's Neuton Is A New Neural Network Framework For AI And Machine Learning". https://www.forbes.com/sites/montymunford/2018/10/04/bell-integrators-neuton-is-a-new-neural-network-framework-for-ai-and-machine-learning/. 
  2. "Neuton: another, problematic neural system structure". October 5, 2018. https://www.inventiva.co.in/trends/neuton-another-problematic-neural-system-structure/. 
  3. 3.0 3.1 Integrator, Bell (October 4, 2018). "Bell Integrator Launches Neural Network Framework to democratize AI". http://www.globenewswire.com/news-release/2018/10/04/1601624/0/en/Bell-Integrator-Launches-Neural-Network-Framework-to-democratize-AI.html. 
  4. "Bell Integrator запускает прорывную нейронную сеть!". October 10, 2018. https://www.saratovit.ru/2018/10/10/bell-integrator-zapuskaet-proryvnuyu-nejronnuyu-set/. 
  5. Cheung, K. C. (October 9, 2018). "Neuton: A New Disruptive Neural Network Framework Far More Effective Than Any Other Framework". https://algorithmxlab.com/blog/neuton-a-new-disruptive-neural-network-framework-far-more-effective-than-any-other-framework/. 
  6. "Bell Integrator Launches Disruptive Neural Network Framework". http://bell.one/Neuton.ai. 
  7. "NEUTON Trademark of Bell Integrator, Inc. Serial Number: 87896515 :: Trademarkia Trademarks". https://trademark.trademarkia.com/neuton-87896515.html. 
  8. 8.0 8.1 "AI for everyone is just a click away". December 10, 2018. https://www.reuters.com/sponsored/article/AI-for-everyone-is-just-a-click-away. 
  9. "Фреймворк Neuton сделает искусственный интеллект доступным непрофессионалам". https://www.itweek.ru/ai/article/detail.php?ID=203816. 
  10. Anadiotis, George. "Neuton: A new, disruptive neural network framework for AI applications". https://www.zdnet.com/article/neuton-a-new-disruptive-neural-network-framework-for-ai-applications/. 
  11. "Auto ML solution makes AI 'available to everyone'". October 11, 2018. https://www.smart2zero.com/news/auto-ml-solution-makes-ai-available-everyone. 

Further reading

External links