Predict the future

World wide AI platform, blockchain for deep learning.

SCROLL

PROJECT

Problem

Natural disaster, traffic accidents, financial crisis- Our world brims with problems difficult to predict. We treat these uncertainties as risks, constantly hindering the improvement of economic activity and industrial efficiency.

Important components of AI

  • 1
    Data

    Data

    Data: Different types of data sets that exist on the web, such as web service APIs and IoT sensor data.

  • 2
    Model

    Model

    Model: The “base” part of the artificial intelligence (AI) system that describes how the data given to the system will be processed (mathematically). There are many different types (such as neural networks and decision trees) used for different purposes. Machine learning is a type of AI system where the model can continuously self-improve for better accuracy when exposed to more data.

  • 3
    Machine

    Machine

    Machine: Machine powered resources (processing power such as GPU and memory) required for processing data as per described in the model.

Environment

Data
Acrion
(Prediction)
Agent
(AI)

Model

Machine

The Status Quo

At this moment in time, artificial intelligence (AI) systems have not yet reached the level of performance or prediction accuracy necessary for widespread commercial use. Current AI implementations are often developed in centralized environments, where its three main components (defined above as data, model, and machine) are provided by a single entity or vendor. This presents a quandary: the accuracy of the prediction suffers significantly unless all three main components are present without lacking, yet most singular entities find satisfying this requirement in full to be a daunting task.

The Future

Daisy is a platform where all AI development is possible in a decentralized environment. The platform eliminates obstacles present in current developmental practices, such as defects in any data or model, or the machine not having the processing power to compute large data sets. It enables the production of AI implementations to predict with greater accuracy, all while combatting a multitude of other present issues.

Environment

Agent(AI)

Environment

Data

Model

Machine

Prediction

How

1Block chain

All data, model, and machine resources contributed to the platform will be deployed on the blockchain and joined in a network. Development and resources previously realized by a single entity are now are dispersed amongst peers on a decentralized by blockchains.
*patent pending

2Collaborative Intelligence

Daisy welcomes everybody from global individual developers to large corporations to participate and contribute to our platform. With each participant submitting any and all combinations of data, model, and machine, AI implementations developed on the Daisy platform are made with the collaborative intelligence of our users.

3Prioritize

Daisy configures itself to maximize performance by prioritizing the user submitted contributions to the platform. In order to achieve this, Daisy repeatedly utilizes elements that effect performance positively and phase out those that do not, naturally creating order amongst the contributed elements.

4Reward system

Users that have contributed elements proven useful to an AI solution on the Daisy platform are compensated per contribution via ”ℓ” (liter), a token issued by and used as the official currency on the platform. Tokens are distributed impartially through the utilization of proprietary technology, “Neuron as an Agent” *. With this technology, an algorithm is used to calculate the weight of each working contribution towards an AI solution, tying its effectiveness to the user’s compensation.
* Presented at the international conference ICLR18 Workshop, patent pending.

What is ℓ ?

ℓ is a cryptocurrency named after the metric unit of water, litre. The fluidity of the name is a figurative parallelization, as the ℓtoken represents the necessity of the “blood” flowing through the “brains” of the Daisy platform.

5Security

The Daisy platform ensures the complete protection and confidentiality of any and all information submitted to the platform by providing confidential computation through homomorphic encryption, enabling information contributed for usage in a model to be computed without disclosure.

6Deep Learning

The Daisy platform is architected to naturally utilize deep learning, multiple processing layers of the model to hierarchically expand. Daisy implements a multi-layer neural network connected in multiple stages by multiple contributors which utilize backpropagation and end-to-end training in order to make powerful predictions.

Environment

Agent(AI)

Environment

Data

Model

Machine

Prediction

Block chain Platform
General user
Going "Deep"

TEAM

MANAGER

CEO Founder

Shohei Ohsawa

WHY ME?

Assistant professor in the University of Tokyo, experts at deep learning and blockchian.
He got a doctoral degree at Matsuo Lab. in 2015.
In the former job, IBM Research, he worked with finantial institutions as a researcher of Hyperledger, and received the president award. His paper with regard to semantic web was received the Best Paper Award.

COO Co-founder

Seiya Takenaka

Entrepreneur. At his previous job, he released a social media network application and was contracted for system developing as well as consulting work with various companies. He participated in TEDx Talk ’15 and received an award in the MIT Business Contest. He joined Daisy as COO and is currently in charge of management and business strategy.

TEAM

PARTNER

Prediction