Introducing DML — Decentralized Machine Learning Protocol
Decentralized Machine Learning unleashes untapped private data, idle processing power and crowdsourced algorithms development by on-device machine learning, blockchain and federated learning technologies.
Since AlphaGo, we can already see how powerful and how large the potential of machine learning could be. What if there is a way to potentially engage billions of devices and tens of thousands of developers to make the future machine learning even more powerful yet decentralized? DML Protocol is the key for this revolution.
Machine learning market is huge, but three big limitations
The International Data Corporation foresees that the industry’s revenue will reach over USD210 billion in 2020. Tech giants have heavily invested and gained remarkable achievements in machine learning. However, the current machine learning development is hindered by:
Inaccessibility of Private Data
Traditional machine learning requires datasets to be uploaded to a dedicated server. Due to privacy concern, massive amount of private data stored in individual devices is untapped.
Centralization of Processing Power
Nowadays, machine learning is mainly conducted through a centralized computer, which its processing power is usually limited or confined to the processors of a single machine.
Limitation of Models and Algorithms Development
Only large corporations can afford investing huge initial capital and resources to build in-house machine learning models and algorithms, or acquire tailor-made ones from consultancy firms to apply machine learning in their own business.
DML advances machine learning development by returning power to all ecosystem contributors
We created an open source infrastructure and network which data, processing power and models/algorithms development will all be decentralized.
By deploying encrypted algorithms into all individual devices and conduct machine learning locally, there is no need to extract or upload any private data to a third party server.
All private data will be kept within the devices and only the local prediction results will be transferred. Usage of untapped private data is unleashed as a result.
DML Procotol will utilize the idle processing power of the devices to perform on-device machine learning.
The supply of algorithms will be crowdsourced in our developer community. Customer such as corporates, research institutes, governments or NGOs can simply search or request suitable machine learning algorithms in DML Algorithms Marketplace.
In order to promote machine learning development, we will also hold various machine learning competitions/blockathons to attract and reward our community talents.
In the next articles, we will further explore different components of Decentralized Machine Learning protocol, including the participants, the function of decentralized nodes and the blockchain smart contracts. You will also meet our team and understand why we are building DML Protocol.
Join the Community
We will announce all the latest and exclusive news in our community. So chat with the team and our growing community by joining our Telegram channel.