aiTech Trend Interview with Martin Goodson, CEO of Evolution AI

aiTech Trend Interview with Martin Goodson, CEO of Evolution AI

What do you do at, and what do you build with your team?

I am CEO and Chief Scientist which means I run the company and also lead R&D. Our products perform data extraction from documents and are based on deep learning algorithms we have developed in-house. We are very R&D focused and we have many University collaborations – including with UCL and the University of Southampton, which has a great computer vision department.

What are some of the industrial sectors that Evolution AI caters to?

Mainly financial services – especially Asset Management. These are industries that run on large volumes of documents and it’s critical to extract important financial data points accurately. Previous methods of automation, such as optical character recognition (OCR), do not work well here so much of this work is still done manually.

What are some of the common challenges customers approach Evolution AI with?

A typical challenge would be a requirement to extract the data from many thousands of balance sheets. This was very difficult to automate before modern deep learning methods arose. It involves quite sophisticated reasoning capabilities to extract data from complex documents like balance sheets.

It involves an understanding of language and visual relationships – deep learning is very well suited to this.

What are some of the distinctive features of the Evolution AI that differentiates you from your competitors?

Our very high accuracy (which we guarantee) and our ability to work with very complex financial documents.

Can you give us a sneak peek into some of the upcoming product upgrades that your customers can look forward to?

We have just released a new solution specifically to deal with Quarterly Reporting packages in the Private Equity industry. We are very excited about the technology that went into it because it was previously very difficult to deal with complex tables which span many pages, for example.​

We have also recently developed a new version of our algorithms that needs no training data at all, it’s a zero-shot learning algorithm. This will be released into the production system within months.

Where do you see the biggest areas of improvement for Deep Learning in the Finance sector?

Deep learning still needs significant computing power (and energy!). We’d love to see an increase in efficiency there.

How do you keep pace with the rapidly growing AI Solutions product for businesses?

I’m not sure I understood the question – but if you are asking about keeping pace with the development in AI, our firm runs a networking group called London Machine Learning. (It is the largest community of Machine Learning practitioners in Europe). I learn a lot from our members and invited guests.

What is that one quote that has stayed with you throughout your professional life?

“I fear not the man who has practiced 10,000 kicks once, but I fear the man who has practiced one kick 10,000 times.”

Bruce Lee

Bio for Martin Goodson at Evolution AI:

Dr. Martin Goodson is Chief Scientist and CEO of Evolution AI, and a specialist in natural language processing and deep learning. He is the Chair of the Royal Statistical Society Data Science Section, the membership body for data science in the UK, and runs the largest machine learning community in Europe, London Machine Learning.

Bio for Evolution AI:

Enterprises worldwide spend over $40Bn annually on manually extracting data from documents. This extraordinary sum represents hundreds of millions of human hours spent on tedious work which is also slow and prone to errors. The incumbent technology solution, OCR (optical character recognition) has been around for decades. It’s simplistic, rule-based, and performs poorly on many real-life documents. We founded Evolution AI in 2015 to solve this problem. Evolution AI revolutionized data extraction by building a deep-learning machine that closely mimics how humans read documents. Humans not only read the characters of the page, we heavily rely on visual cues from graphic design (tables, font types, and sizes, colors, bold/italic ​, etc) and context. In AI terms, these translate to Visual Reasoning – a branch of computer vision that focuses on how objects visually relate to each other – and NLP (Natural Language Processing). In short, our AI weaves together text, vision, and semantics to extract data from documents – just like humans do.


Related post

Exit mobile version