Artificial Intelligence and the Economy | BlueDot DI helps to deliver smart marketing

Artificial Intelligence and the Economy features machine-learning computer models in Jamaica. These models are computer algorithms, or smart apps, that seek to give computers the ability to learn like children for a variety of tasks.

Here, we highlight how an author's work may solve a particular set of real-world tasks or problems. By doing this, we aim to encourage more local research and development in artificial intelligence.

Today, we will highlight machine learning applied to smart marketing, that is, the science of equipping businesses with the ability to smartly detect consumer desires. This is work being done by Adrian Dunkley at Blue Dot Data Intelligence, the first data science/machine learning-focused company in Jamaica. I was extremely surprised to find that a company of this level exits in the country.

Jordan: Briefly explain your role at your company BlueDot Data Intelligence (DI), including what data science is and how it relates to machine-learning, as well as what your company does.

Adrian: BlueDot DI is an analytics and data science consulting company. My role at BlueDot is to oversee the data science process and machine-learning development.

I like the quote popularised by DJ Patil, the first US Chief Data Scientist: "A data scientist is that unique blend of skills that can both unlock the insights of data and tell a fantastic story via the data."

As a data scientist, I combine business understanding and expertise, data, visualizations, statistical analysis and machine-learning techniques to provide measurable value to organisations.

Machine learning is the development of algorithms that allow computers to find patterns and make decisions by analysing data. The idea is that instead of explicitly defining every rule for every situation, which would be tedious and take a very long time, you define some parameters (a goal, a reward and the computer follows these rules to learn from data). The trick is to translate the business goal into a form a computer can understand.

You can think of machine learning as an integral tool used when conducting data science.

What do you like about machine learning?

Machine learning is all about feedback. You build a model, test it and validate the outcome. If the model's performance does not meet your requirements, you review your approach and your data, make adjustments and test again. You may not always get the results you planned for, but you will always learn something.

Now, once you've developed a machine-learning model that predicts well or discovered amazing insights that were hidden in a sea of data, then you feel like a boss. The next step is to test your creation under real-world conditions. When you see your model has led to improvements in operational efficiency, profits or a reduction in losses for a company, you feel like a big boss.

Let's talk about a large problem you helped to solve at BlueDot, with the help of machine learning. Briefly describe the problem (or problem space as we say in the artificial intelligence field) as well as tell us briefly about the solution you employed.

A problem we have identified is 'blind' marketing. Some companies launch a campaign or a new product without having the benefit of data analytics. Then the impact of these endeavours isn't measured against the effort put into them, return on effort or return on investment.

At BlueDot, we have assisted companies with marketing and brand imaging, using machine learning and artificial intelligence. Our approach uses neural networks to analyse how consumers use and relate to products, video commercial and radio advertisements.

 

Method used to solve problems

 

Let's get into somewhat technical territory. What class of learning algorithm(s) did you decide to leverage to solve the problem mentioned earlier?

We probably need a whole other interview to discuss that, but to keep it short, we rely on a combination of techniques, including Convolution Neural Networks (CNN) and Recurrent Neural Networks (RNN). These approaches were inspired by the structure and processes in the human brain. CNNs is the approach that allows your smart device to recognise your face in an image, and RNNs are good at predicting variables with a temporal behaviour what will happen next.

Jordan: In week three's article featuring the talented 19-year-old Jamaican, Leon, readers can view more details on learning algorithms like CNNs, and see why these learning algorithms have become easier to apply in recent times. Crucially, many other Jamaican individuals can acquire similar talents, due to the ease of access of these powerful learning algorithms.

What are some of the shortfalls of your solution you seek to get rid of in the future?

Like with all data-driven solutions the quality, availability and the relevance of your data is fundamental. We have developed a data-collection methodology that focuses on publicly available data, surveys and focus groups.

How does society benefit from your solution?

We want Jamaicans to have more control of what products they are presented with, by letting companies understand what Jamaicans really want and what they don't want. We also want Jamaicans to be the ones to get the most out of Brand Jamaica, a topic for another discussion.

Can you describe some of your larger machine learning-related personal projects?

I have multiple projects running right now. My passion projects are the development of an AI to support football coaches. I call it an AI Assistant Coach. It predicts the angle and direction of a goalie's dive for the penalty shot, measures the strength of football formations based on player movements, and tracks player performance.

The next project is the development of an unsecured credit risk model architecture to increase the ease of access to funding for lower-income earners and gamifying good financial management.

Stay tuned for next week, where we seek to highlight more Jamaicans applying machine learning. Remember to tell us about your machine-learning projects.

- Send feedback to editorial@gleanerjm.com, or jordanmicahbennett@gmail.com.