TOKYO, May 16, 2017 – (JCN Newswire) – Fujitsu today announced
development of its Fujitsu AI Solution Zinrai Deep Learning System, comprised
of a dedicated deep learning server incorporating the latest GPU(1) from
U.S.-based NVIDIA Corporation, along with operationally tested storage and
software. Sales of the system begin today in Japan.

Using this system, customers looking to build a system using
deep learning in an on-premises environment can quickly set up a deep learning
platform offering world-class speed. In addition, depending on needs, the
customer can also easily build a hybrid cloud environment incorporating the
Fujitsu Cloud Service K5 Zinrai Platform Service Zinrai Deep Learning, which
provides a deep learning platform in the cloud. Given this, it becomes possible
to use according to objective, such as by performing learning processing with
low usage frequency in the cloud, and learning processing with a high usage
frequency in an on premises environment.

Going forward, Fujitsu aims to extend the Zinrai Deep Learning
System to regions outside Japan, and to further strengthen its solutions to
flexibly meet customers’ various AI-related needs.

Background

In recent years, the use of deep learning, a type of
artificial intelligence (AI) machine learning technology, has been expanding,
which has resulted in dramatic improvements in AI recognition accuracy,
accelerating the development and spread of services using AI. In order to
achieve high recognition accuracy using deep learning, however, large volumes
of training data need to be processed over and over, creating issues with the
significant time and computational resources required.

In order to respond to this type of issue, in April 2017,
Fujitsu launched Zinrai Deep Learning, which offers a deep learning platform with
world-class speed as a cloud service.

In addition to that service, Fujitsu has now begun sales of
the Zinrai Deep Learning System, which offers a dedicated deep learning server
incorporating the latest NVIDIA Tesla P100 GPU, along with operationally tested
storage and software as a complete system for customers that, for such reasons
as preferring to not take data off site, want to build a deep learning platform
in an on-premises environment. 

System Features

1. Offers a deep learning platform with world-class speed,
incorporating the latest GPU

This system provides a deep learning platform with
world-class speed through a dedicated deep learning server employing the latest
NVIDIA Tesla P100 GPU. Because up to eight GPUs can be incorporated into a
single server, customers have the flexibility to select a system structure
based on application and in consideration of Fujitsu’s recommended
computational capabilities.

2. Offered as a complete system including operationally
tested storage and software

This system is offered as a complete package, combining a
dedicated deep learning server that includes over 100 types of necessary
software in a ready-to-use state with operationally tested storage. Customers
can therefore quickly build a stable deep learning platform in an on-premises
environment.

3. Easy creation of hybrid cloud environment incorporating
Zinrai Deep Learning cloud service

Customers can easily build a hybrid cloud environment by
combining the on-premises deep learning platform built with the Zinrai Deep
Learning System with the Zinrai Deep Learning cloud service, so that tasks can
be flexibly divided based on application.

Pricing and Availability

Sales Target

A total of 1,900 systems by fiscal 2020

(1) GPU
Graphics processing unit. A processor specialized for
processing images. In recent years, it has been applied to problems outside
image processing due to its high processing capability.