Decentralized Machine Learning-DML Protocol

Posted: March 26, 2018 by Tyler Swope

Decentralized Machine Learning-DML Protocol


https://decentralizedml.com/
Click Here for Decentralized Machine Learning Telegram

The Machine Learning Market

Machine learning, and artificial intelligence is a massive market, and is reaching the forefront and the top of people minds with the introduction of blockchain technology. Today, I would like to introduce a new project whose aim is to offer a project that takes machine learning to a whole nother level…that project is decentralized machine learning. The platform they are creating will utilized untapped private data and idle processing power of user devices for artificial intelligence model training while keeping the users data completely safe and private and at the same time providing economic incentives.
Now the machine learning market is massive and will only keep growing into the foreseeable future. In 2016 the revenue from big data was 130 billion USD. The international data corporation foresees that the industry revenue will reach over 210 billion dollars USD by 2020. That nearly 100 percent growth in just 4 years.

Industry Problems

There are some big problems within this industry though. In recent years it has become much cheaper to generate, store, and access data thanks to the internet of things and mobile devices. Now the data is broken down in to publicly and privately accessible data. Of course, the private data is much more relevant and useful, but how current things work tech giants like facebook, google, and amazon have made a hell of a business out of leveraging this private data without the users consent and keeping all the profits made for themselves. We are unaware that this private data is exploited and used, and of course this data is stored on central servers which are prone to hacking. Cambridge Analytica ring a bell to anyone? This is a major example of the problem, our data is being used in nefarious ways without our consent!

The Decentralized Machine Learning Protocol

Decentralized Machine Learning is attempting to change all of that and enable machine learning without exposing your data. The protocol is designed to expand the reaches of machine learning to privately held data stored on users devices. They will ensure that data privacy is protected, by only sharing the aggregated machine learning result and not the data itself. Idle processing power on a participants device is used to compute the algorithms and compute the results.
At the core of the DML protocol is the smart-contract based DML Algorithm marketplace, where any developer can list a machine learning module for sale in a middle-man free environment. Customers acquire the algorithms from the marketplace that best fits their purpose. The customer specifies their needs and criteria such as target segments: geographic and demographic. The criteria input, scope of services provided by each participant and rewards in terms of an utility token i.e. DML token will be written in the blockchain-based smart contracts. The smart contracts will be digitally signed to eliminate default risk and connected among the customers, developers, and the decentralized nodes that are connected to DML protocol.

A Working Product

DML has had some great activity on their product development. There is a working prototype of the DML algo marketplace and DML App for users running on an Ethereum testnet. I love working products, and is a major component of wheter I will invest or not. Booya baby, working product that is going to be very, very useful right away. The DML Protocol Gen 0 The beta is expected to launch upon the completion of DML token generation event, so to provide an immediately utility for the DML Tokens. Customers and developers can then transact machine learning algorithms under the interface of DML Algo Marketplace with the support of smart contracts. Who doesn’t like earning free value for data they own, which in the past used to be taken from us and used to make other people richer?

Decentralized Machine Learning: How It Works For Users and Nodes

So what about the other end of it, how is people data taken from their devices and distributed within the DML ecosystem. The DML protocol adopts a decentralized approach to connect the developers to various decentralized nodes with different functionalities. Each of these nodes will connect to multiple individual data owners devices. This means a massive number of potential data owners can be connected at faster speeds and stable connections. The distributing nodes identify devices for algorithm distributions and send the algorithm to those devices. Unlike traditional data analytics, which private data are transferred from the data owners to the developers or a centralized hub for processing, DML protocol will facilitate machine learning algorithms to be run directly on data owners’ devices such as smartphones without the need to extract any personal data or to store the information elsewhere for processing. As a result, individual private data will be well protected without any raw data leakage from individual devices.

After the machine learning algos are run on the authorized dataset within the device, only the analytical conclusion in the form of local prediction results will be encrypted and sent to the next type of node. A federateded node. Federated nodes aggregate local machine learning prediction results, and then average the aggregated results by federated learning. Federated nodes also validate the quality of the results against the requirements, if they don’t match the defined scope, the node will instruct the devices to re-run the algorithms.

The third and final type of node called the report node receives the aggregated results from the federated node, this node will further average the encrypted results, and create a finally report that is generated and stored in a distributed file system such as IPFS. The report node also updates the smart contracts.

Here is how it looks like when it all comes together. Ton’s of moving parts, but the protocol is very well though out and is it working….how many cryptos can say that! DML tokens are the utility token of the protocol, and of course they are used to incentive users for contributing data and idle processing power. They are also used to incentive the developer community to create new machine learning algorithms. Customers can deploy the algos on the marketplace by purchasing them with DML tokens. Another function is when developers seek to train and enhance the accuracy of their models, and they can reward trainers with DML tokens. A final function is to provide incentives for all of the three types of nodes….can you say masternodes in the future!

Here is something I like, the team is really focused on the future of cross-chain compatibility and interoperability. Their protocol aims to allow integration of new blockchain tech to the currently adopted ones without replacing or overriding them. They wan’t to create a co-existing ecosystem that links and aggregates different blockchains, and new technologies will be created as a side-effect.

Decentralized Machine Learning Team

What about the team, who are the names behind this project? Victor Cheung is the blockchain developer, he is a full-stack web, mobile application, and smart contract developer. He is proficient in solidity, C#, java, php, node.js and more. Micheal Kwok is the project lead director and has founded 2 tech companies, and has experience in early stage startups, especially proficient in business development, digital marketing, and SEO. Jacky Chan is the blockahin and software developer who is a full stack software engineer. He has substantial involvement in Metamasks new UI development, as well as the network visualization dashboard design with Dfininity. Pascal Lejolif is the machine learning engineer and has decades worth of experience in the space. He was the former CTO of Alkia IT services, which specialized in cloud computing, AI and cyber security, These are just a few member the team and advisor list is long and strong, and gives me confidence in this project.

Decentralized Machine Learning Token Sale

Let’s dive into a look at the token distribution and sale. 36 percent of the tokens will be up for sale, 9.9 percent will be given as ecosystem bonuses for public contributors, 8.3 percent will be given for debugging, development and protocol upgrade from the developer community. This means 54.2 percent is going to the public. 19.5 percent is going to the team, advisors, and early contributors, 15 percent is being held in reserves, 8.8 percent for strategic business and research partners, and 2.5 percent for PR and marketing. Total token supply is 330 million. The team is aiming to raise 28000 ethereum during the token sale. No details on the token sale date, but you can get whitelisted by going to their website which is in the description

No Comments

Leave a Reply

Your email address will not be published. Required fields are marked *

X