The world of online banking is heading towards a digital reality where questions are no longer answered by humans, but by computers. In the last year alone the number of banks focusing on developing technology to create realistic conversations between humans and robots has risen exponentially. As chatbot technology gets smarter, thanks to machine learning techniques, banks are developing new ways to interact with customers, that can understand human emotions and recognise intentions, all whilst engaging in conversation. For the development of truly intelligent virtual assistants and chatbots, deep learning and natural language processing (NLP) technologies are key. Adam McMurchie, SME Technician Subject Matter Expert at Lloyds Banking Group, has a good understanding of where the gaps and challenges are currently within deep learning, with a background in Financial Services, Technological Development and Neurocomputing. On a panel of leading innovators within financial services, Adam will be joining us at the Deep Learning in Finance Summit (1-2 June), to further discuss and explore the implications of NLP for online banking.
I asked Adam a few questions ahead of the summit to learn more about his work, and the impact deep learning is having on the financial sector.
Please introduce yourself and what you do.
My name is Adam McMurchie, I work developing applications that run the banking systems in the Cards & ATM space. I have a physics degree, also I am a polyglot & translator with 10 years experience. I have an interest in natural language, natural systems and am passionate about emerging technologies that have a positive social impact and bring people together.
How did you get into deep learning?
I had been working on broad spectrum translation software for a couple of years and had been looking into emerging technologies that could assist me improve the efficiency of language processing.
Which industries or areas do you feel deep learning will have the most beneficial impact?
On a high level I think Deep Learning will be most beneficial in the energy, medicine and the services industry.
In the energy sphere, we will see a world where excess green energy is distributed via smart networks, that thanks to pattern recognition aren’t simply reactive but predictive - and to a high degree of accuracy. This could be a huge asset in the fight against climate change.
In medicine, IBM Watson and similar DL applications are able to analyse far more information than doctors and standard analytical tools, drawing correlations and diagnosis from millions of patient records, publications and spotting trends in a sea of data that could otherwise be missed by experts.
Finally Deep Learning is beginning to solve the issue of ‘dumb automation’, from smart chatbots to tailored customer services honed via big data and pattern recognition.
What advancements in deep learning would hope to see in the next 3 years?
I would say the holy grail in deep learning is also where the main bottle neck lies at the moment, which is in transfer learning (the ability for A.I to apply pre-learned knowledge to a completely new task).
So for example AlphaGo which recently won the last intellectual game to be dominated by humans (go), is not designed to apply its learnings to say driving a car or treating patients.
I hope to see progress in the next three years addressing this problem, essentially using both stochastic gradient descent and a genetic selection method - which may be the missing link towards building a genuine general learning system.
How beneficial will natural language processing (NLP) become within financial services?
When implemented effectively, NLP will completely transform banking landscape. Sophisticated analytics programs are great for highlighting important relationships in xls’s and relational databases - but in terms of usable data this only accounts for about 20%.
The real issue for businesses is how to extrapolate useful information from unstructured data - such as an ocean of social media posts, images, email, text messages, audio files, Word documents, PDFs and other sources that make up the other 80 percent of data which can’t be understood by traditional computer tools and methods.
Personally, I think people are far more likely to share their experience on social media than a customer satisfaction survey. So even in this regard, NLP could be invaluable to companies for understanding how customers really feel.
How disruptive will deep learning applications be for the banking industry?
Deep learning adoption is increasing rapidly. It will have a profound polarising effect - banking industries and fin tech’s that embrace DL will be propelled forward making huge gains and companies that don’t will be left behind wondering where their customers went and why they have higher fraud than their competitors.
Simply put, Deep Learning will accelerate the rise and fall of companies in the banking sector - companies that are slow to respond, with ‘missed learning opportunities’ may no longer simply suffer lower ROI and profits, but could face annihilation.
Other confirmed speakers include Jakob Aungiers, Head of Quantitative Research Development, HSBC; Blaise Ngonmang, Chief Data Officer, AXA; Oded Luria, Data Scientist, Citi; Michael Natusch, Global Head of Artificial Intelligence, Prudential; Diego Klabjan, Professor, Northwestern University; Harshwardhan Prasad, Sr Manager of Financial Risk Management, Ernst & Young, and more. View more speakers and topics here.
Early Bird passes expire on Friday 7 April. Book your space now.