Blog: GNY: Machine Learning on the Blockchain.
EliteX Exchange’s new coin listing.
By Arjuna @ EliteX Media
Since GNY (www.gny.io) has the distinction of being EliteX Exchange’s first and newest listing (aside from our base pairs), we thought it would be useful for our new and growing cohort of users to know a little bit more about this cutting edge project. Machine learning is one of the hottest areas of tech development today, and the team behind GNY are right at the forefront. Glancing at the blurb on their website, it states: “GNY (pronounced “gee-nie”) is the next generation of our proven machine learning predictive platform for the blockchain”. But, many of us lay folk may only have a vague idea about what machine learning is. So, we thought: why not ask!?
We got in touch and had a chat with Zach Barnett, Head of Social Impact and Cosmas Wong, Co-Founder and CEO of GNY, starting with the question: What is machine learning?
CM: Machine Learning uses complex combinations of algorithms to find repeatable patterns in large amounts of data structures. Often these patterns help someone understand what actions or circumstances lead to a specific action, for example why a customer buys a specific product.
Machine Learning is like an assistant that can be directed to solve for why certain things happen, by iteratively and continuously trawling through large amounts of data that changes all the time. This process is so data intensive and laborious, that only a machine can complete it efficiently and effectively.
The learning component of our machine learning system is linked to feedback loops. When the algorithms makes an incorrect prediction, that outcome is recorded and the system learns from that mistake, makes corrections, and continues to improve its predictive process. This self learning is a continuously adjusting process that can only be accomplished by a machine.
The site mentions that GNY is the next generation of their machine learning platform. Does this suggest that there was an iteration of the GNY project prior to its migration to blockchain? If so, what are the advantages of implementing GNY on blockchain? What does building it on blockchain enable that it otherwise couldn’t achieve?
CM: Our company started as Grey Jean Technologies and was born in 2015 to answer a specific question: how can the vast amounts of digital data left behind by people be organised and analysed to provide business, marketing, and social impact solutions that are predictive, adaptive, and responsive?
So we built a machine learning technology that could identify repeatable patterns, and learning components that constantly adapt to our ever changing behaviour. Genie, the original ML platform, predicts a user’s behaviour so they can be targeted with contextually relevant messages that drive desired actions. It does this by responding to millions of independent variable features while using hundreds of machine learning algorithms in parallel, in order to predict defined business values.
But there was a problem. The customers that could afford this level of service were limited, and frankly we weren’t interested in only using the technology in strictly commercial settings. The desire to make this technology more accessible was really exciting to us, and that’s when I connected with Richard, and started wondering… Could we decentralise this technology and software using blockchain?
The benefit of moving the machine learning to the blockchain are threefold:
1. Consistency of data. Data comes in all different structures, qualities, and continuities. Building the data collection, storage, and analysis of data into a blockchain allows our customers to control for consistency of data more reliably than ever before. Consistency of data correlates directly with the effectiveness of the ML’s efficiency.
2. Accessibility. Most ML services are priced out of reach for most mid-sized companies. Our solution will allow users to download the code for installing a GNY side chain on their private servers. The chain will have our proprietary ML software coded into it. That way you can optimize and analyze your data in the security of your private server at a much more accessible price.
3. Security. Existing competitive services generally require that the customers send them all of their data and then they will have predictions and data delivered back to them. In the era of hacks and data breaches this kind of regular migration of sensitive data makes most people nervous because it exposes the data to security risks.
What about use cases? The material on your site divides into three broad domains of application: sales and marketing, publishing, and fraud detection. Can you say a little bit about how it may be used in such cases?
ZB: The common theme for all use cases is that GNY analyzes huge amounts of data and then is able to predict next actions for your customers. This allows you to make suggestions that are relevant and timely and efficient. The ability to make the right suggestion at the right time is the holy grail of marketing. Personalised communication with customers at scale and efficiency.
Publishing and marketing have very similar needs in this regard. It’s about having a deeper understanding of your customers at the individual level and being able to personalize communications, offers, and suggestions based on their histories. Our experience in both retail and publishing demonstrations have allowed us to greatly improve the relevancy of suggested items for purchase, or articles to view next.
Fraud detection relies on a variation of the goal of pattern detection. With fraud detection we are looking at abnormal patterns. Potential fraud produces a pattern that is out of step with the understood patterns associated with approved and correct usage of a product, drug, or supply chain. Machine Learning can help alert companies when a customer’s or supply chain’s activity is abnormal and needs to be investigated further.
How does GNY improve upon traditional methods of market segmentation and analysis? How does it help to find the relevant segments/data points?
ZB: The problem with traditional segmentation is that it is a very unsophisticated method of determining consumer choices and doesn’t change as a consumer’s behavior does. Segmentation buckets consumers into very simplistic descriptors. Why a consumer does something is highly complex. It involves multiple factors and segmentation doesn’t also deal enough of them or with weights of those specific factors. It is static. Once a person is put into a “segment”, it hardly changes even if that person’s circumstances, tastes, bahavior, or tastes diversify. Only ML can deal with the number of data points and the continually changing weights of those data points effectively.
Earlier you mentioned social impact, and an interest in GNY not being limited only to commercial settings. What about non-profit applications, like research, social and environmental change?
ZB: Social impact applications of GNY is an area that the entire team is incredibly excited about. We are in discussion with non-profits about using our ML blockchain platform to model environmental conditions and improve operating systems for social justice organisations. Most social justice organisations suffer from restricted resources and have a hard time unlocking the hidden value in their data. Our Magic Wish Technology Grant was designed to help provide use cases for social impact organisations and inspire them to realise their missions. More information about that can be found on our site www.gny.io.
The GNY token, enabling decentralised machine learning on the blockchain, will give users access to its platform utilising a pay as you go model. Smart APIs bridge between on different blockchain networks including Ethereum, Asch, and of course Lisk. Soon to run on its own DPOS (delegated proof of stake) network, GNY will provide developers with the tools to build their own sidechains built around artificial intelligence. To that end, GNY is also building Lisk Machine Learning (LML), open toolsets for AI-empowered sidechains built on the Lisk ecosystem. For anyone who wishes to explore further, you can read its technical and use case white papers here. Feel free to join the official GNY telegram channel: https://t.me/GNYioBlockchain, which is open to everybody including potential developers and clients. You can also follow the lastest GNY developments and news on twitter.
From all of us at EliteX Exchange, thank you to the GNY team for giving us your time, and answering our questions. Those of you who have signed up to our exchange already, will receive their airdop of GNY this coming Monday (20th). If you missed out this time around, there will be second round GNY airdrop in the near future. All you have to do is sign up to the exchange: www.elitex.io. Keep an eye on our site and social media, more information about it will be announced soon!
Disclaimer: This article is for information purposes only, and does not constitute investment or trading advice.