The Dock recently hosted its very first Deep Learning for Enterprise Meetup, which featured a blend of networking, keynote speeches, and some deep dive talks where a panel of experts shared their experiences from recent Deep Learning projects, and offered advice on how to leverage methods to build a deep connection between academic and industry.
To better understand this rapidly emerging space, I asked three of our speakers at the event to offer an insight to four highly topical questions:
Dr. Marissa Zhou, Lead Data Scientist at Accenture
One example of how an enterprise can use Deep Learning to be more intelligent is via the use of Chatbots—a computer program with a Deep Learning brain. It can help educate potential customers about a company’s services, or it can be used to recommend the right resources to the right employee at the right time, to help the employee grow.
Deep Learning is just one part of the AI ecosystem. Currently, its primary use is in supervised machine learning—like image classification and pattern recognition. We can see a lot of problems that Deep Learning cannot yet solve and it will continue to evolve with the converging of all other technologies.
I wouldn’t consider AI as "hype"; it has been proven to be effective in solving a lot of problems that human beings have limited ability to solve. For example, companies are using machine learning to predict heart disease, and deep learning to help automate the monitoring and detecting of abnormal heart rhythms.
FinTech, including fraud detection, payment process automation
SmartHome, including better living space and energy efficiency control
Healthcare, including new drugs for diseases, also diagnosis
Peter Fitzpatrick, Business Development Manager at Insight Centre for Data Analytics
Deep Learning is now capable of mining large amounts of data due to faster processing speeds and as it uses neural networks (like neurons in the brain) it’s very good at learning from data and can automatically pick out key attributes. It’s particularly good at detecting images and can provide interesting applications in video analytics in areas such as healthcare and autonomous vehicles.
Deep Learning is one of the components, but you also need recommender systems, decision-making tools and optimisation to solve a lot of the problems.
The initial stages of AI through automation of simple tasks, such as credit checks and fraud detection has landed, but the advanced solutions for autonomous decisions i.e. where the systems make decisions without humans is not quite there yet. The proof of use of AI is yet to deliver in a number of cases.
Tony O'Dowd, Chief Architect at KantanLabs
Deep Learning will impact how we understand consumer behaviour and how we can use data visualisation approaches to provide business knowledge. Deep Learning will also fundamentally redefine how Degree, Master and PhD researchers in the field of computing learn their trade.
At its core, Deep Learning finds hidden patterns in data that can lead to actionable insights. For the past fifty years, we've developed procedural code to automate predefined outcomes and processes, now we have a technology that allows the data to decide these outcomes and these outcomes may not be, at first, obvious or pre-determined to humans.
Deep Learning is an approach to developing many applications of AI, but is not the future of AI itself. For example, Deep Learning is used in both image and speech recognition systems, but a combination of technologies is required for autonomous cars (LIDAR for near field object mapping, RADAR for velocity measurements, GPW for topology mapping etc.).
Personally, I believe we're only at the very beginning of the AI explosion and the intelligence and capabilities of computer systems.
Financial trading; this tech will follow the money
We should also consider search and cybersecurity fields.
Accenture’s Enterprise Insight Studio is applying Deep Learning technology to transform Accenture into an intelligent enterprise.
We are not just experimenting, we are also focusing on deploying Deep Learning technology into production across Accenture's business.
Deep Learning is growing in sophistication at a rapid pace and in order stay competitive, we are continuously evolving Accenture’s methodologies of innovation. Our team have delivered several successful Deep Learning use cases such as generating insights from employee surveys, helping internal audit teams identify hidden non-compliant expenses and frauds and optimizing enterprise content search.