In-wallet Plug-and-play Machine Learning Brought To You By GNY
Overview of this news release.
– GNY’s system allows enterprises to securely and collaboratively unlock the hidden value in their data through an integration of commercial-grade machine learning and blockchain.
– Our latest wallet upgrade allows you to experience a selection of our in-wallet plug-and-play machine learning contracts.
– You can now test drive our retail, location and fraud ML contracts, plus learn about our Jupyter Notebook interface whose aim is to facilitate outside developers craft ML to their own specific needs.
Sitting in a retail company’s historical data are patterns that can predict their next top seller, the outlet locations that will sell the most inventory next month, and even what transactions are fraudulent. Identifying these patterns, with hundreds of variables, and millions of data points, is too complex for the human mind. Tasks of this complexity require machine learning.
One of the pillars of the GNY blockchain platform is our system to allow enterprises to securely unlock the hidden value in their data through advanced machine learning. Our patent-pending ML technology within the GNY platform was created to allow developers access powerful machine learning contracts through answering custom-designed questions with code they build, or to simply use one of our pre-designed “plug-and-play” tools for common data-driven questions. Questions like: what will my top sales items be tomorrow?
Also within our updated wallet there is an option to run plug-n-play ML contracts with your own data (instruction on that to follow shortly), plus build custom GNY ML contracts using Jupyter Notebooks. Allowing Jupyter notebook access through the GNY wallet will facilitate outside developers who wish to deep dive into the world of decentralised Machine learning and craft it to their own needs. This is an important step for the future adoption of the GNY tech.
A closer look at GNY’s latest ML contract demos.
Within the wallet (under the Machine Learning tab) there are three ML contract demos. For your convenience step-by-step GNY ML tech notes are included beneath each demo to guide you through testing them. However before you jump in let’s take a look at each ML contract individually.
1. The Retail Demo.
This is a sales prediction model for retail stores which allows the user to predict what their next top selling item/s will be.
The Data science behind the retail demo.
Leveraging a deep learning model that considers the L1 regularization we have achieved sales forecasting accuracy rates of 86%.
The products at the subject retail store have been finely categorized. With classification GNY’s deep learning is able to establish correlations between a person’s and a product’s attributes. GNY’s Deep Learning maps these input attributes to outputs. It finds the correlations.
It is known as a universal approximator, because it can learn to approximate an unknown function f(x) = y between any input x and any output y, assuming they are related at all by correlation or causation, for example.
In the process of learning, a neural network finds the right f, or the correct manner of transforming x into y.
2. The Location Demo.
This prediction model for retail stores allows the user predict what store locations will sell the most inventory next month.
The Data science behind the location demo.
As a retailer GNY’s ML will predict the location of your main customers using a multi-objective DBSCAN spatial clustering algorithm to find the optimal clusters using the spatial data collected in the sales area. We carry out Spatial Data Mining with a K-means algorithm that begins with an initial group of randomly selected centroids, which are used as the beginning points for every cluster, and then performs iterative (repetitive) calculations to optimize the positions of the centroids. It halts creating and optimizing clusters when the centroids have stabilised — there is no change in their values because the clustering has been successful.
3. The Fraud Demo.
This prediction model pinpoints credit card fraud using a neural net trained to know a person type, amount, and frequency of transactions.
The Data science behind the fraud demo.
Because of confidentiality issues, we cannot provide the original features and more background information about the data. Therefore we transform all sales into features ‘Time’ and ‘Amount’. Feature ‘Time’ contains the seconds elapsed between each transaction and the first transaction in the dataset. The feature ‘Amount’ is the transaction. These features can be used for neural net learning. Feature ‘Class’ is the response variable and it takes value 1 in case of fraud and 0 otherwise. Given the class imbalance ratio, we measure the accuracy using the Area Under the Precision-Recall Curve.
Begin your testing of GNY’s latest ML contract demos.
Please note that we are currently in the process of optimising this for mobile users, so tablet or pc interaction is currently recommended.
– Create a free account if you do not have one already, then log into the wallet by entering your 12 word passphrase and clicking the blue login button.
– Navigate across to the buttons on the left hand side (Home, Transfer, Delegates, Assets, Machine Learning), and click on the Machine learning button.
– You will then be presented with 3 choices in the centre of the screen.
– Click on Run GNY ML Demos then select your preferred demo.
Step-by-step GNY ML tech notes are included beneath each demo to guide you through testing them.
We hope you enjoy testing our latest in-wallet plug-and-play machine learning contracts and that they allow you to see where the GNY platform is headed.
This is just an early version of the wallet; we have many more ML contracts, functions, and optimisations on the way, so stay tuned.
Also, please join our community on Telegram and let us know your feelings on the test interface & process. We would love to hear from you.
Built in typescript, the GNY platform is your….