Distributed Processing for Machine Learning

0

BTC

0

LTC

0

ETH

0

ETC

0

WAVES

0

WEUR

0

USD

ABOUT

Anryze Computing Network is a globally distributed computer power for Data Processing. Decentralized architecture provides more power, higher speed of processing and affordable costs, 50-60% less than cloud computing

HOW DOES ANRYZE DDPN WORK

Decentralized system for data processing is a network where complex and volumetric tasks can be distributed to independent instances and run simultaneously. Anryze relies on the own developed Desktop Grid computing solution – Valex.

All tasks in the system have their own type. Each type has Tools to solve the task. Tools have a list of dependencies and a list of minimum requirements for the miner’s hardware. Tool is a file or several files that are transferred to the miner in encrypted form and unpacked in a secure storage. A json file with a brief description should be attached to the tools (task entry point, return value, possible errors …)

Each participant of the Network is able to install the software and receive revenue from the mining (data processing). Miners are able to set working conditions to optimize their income and work automatically, getting paid directly.

Distributed computing is more affordable than private monopolized cloud services due to the absence of administrative costs (data center buildings, personnel, infrastructures) and monopolistically high tariffs for the use of centralized providers. Anryze Network offers to the customers ready-made integration solutions.

DATA PROTECTION

The client have different security levels and use TLS with internal encryption protocol (VALEX protocol) to communicate with server. All files that are used on the client (models of neural networks, user files, etc.) or executed by the client (scripts, applications, etc.) are encrypted on the server and the key is transferred from the server to the client. Files can be stored on the client’s hard disk (if large), or in arrays of the file system object representation. Encryption uses the algorithm AES 256 bit.

SOLUTIONS

Machine learning

Neural network training tasks require a big amount of data to be downloaded to remote PCs. The size of training data varies from several gigabytes to hundreds of terabytes. Anryze resources are distributed across the world, which helps to speed up the process of machine learning.

Anryze computing platform consists of hardware that is optimized to speed up and scale up machine learning workloads for training and inference programmed with: TensorFlow, Theano, Caffe, Apache Spark MLib

Rendering

Use the unlimited resources of the Anryze platform for CGI rendering.

Using Anryze you can render faster with hundreds of instances involved for low, competitive prices and with strict security for your data (all data transfers are encrypted).

Here is our team

Anton

Gera

Founder, serial entrepreneur, CEO at Anryze

Oleg

Zaychuk

Dev-ops/system architect

Max

Kudymets

Project Manager

Alex

Pogrebenniy

Back-end developer

Mike

Ezhov

Founder, entrepreneur, CMO at Anryze

Valentine

Chernushevich

Blockchain developer

Angelina

Shynkarenko

PR & Community

Ksenia

Leschenko

front-end developer

Alexey

Kuznetsov

efficiency manager

Advisers

Serge

Milman

Adviser, Dellwood Capital Hedge fund

0

BTC

0

LTC

0

ETH

0

ETC

0