NEWS

2018/07/19

alt Inc. Announces "al+ emeth" Distributed Neural Network Technology

alt Inc. has announced "al+ emeth" Distributed Neural Network technology.

The al+ emeth uses a new decentralized architecture that combines distributed computing technology with neural network technology. Creating a network from the under-utilized computing power of personal computers and servers across the globe, via the Internet, this platform enables the securing of the computer processing power needed for machine learning of artificial intelligence.

This mechanism performs parallel processing by assigning processes to distributed nodes at the neuron or layer level for the neural networks. alt Inc. has named this "al+ emeth Super Neural Networks" and believes it will vastly improve the speed and depth of machine learning.

It is the distributed neural network technology that alt Inc. will be providing, while the neural networks themselves comprise the collection of participant nodes in the network. Participants provide the computer processing power via the network and can receive compensation for their contribution. This assures an economic rationale that will support the autonomous growth of the network and further participation in that network as nodes.

This mechanism also enables maintained confidentiality for data when our al+ stack (see note) technology is used to create personalized learning models from the accumulated personal data in the ways which allow people to take advantage of the al+ emeth computing power while protecting the privacy of learning data. This is made possible by vectorizing the learning data, blocking any back calculation or reverse engineering of existing personal data, and then fragmenting the data before requesting node processing.

alt Inc. will use the al+ emeth Super Neural Networks to both accelerate the pace of personal artificial intelligence development and to expand the scope of that development.

alt emeth基盤の発表

Yoshikazu Nishimura, CTO, alt Inc. presented the alt emeth at the P.A.I. Conference 2018, 12th July, Tokyo