‘This might be one of the biggest contributions in my field, or a waste of time’
UCD’s Dr Anthony Ventresque is working to use AI and software solutions for those previously left out of the technology conversation.
While originally studying philosophy, Dr Anthony Ventresque’s master’s thesis on Turing’s Theory of Computation unlocked the world of computer science to him.
He went on to complete a computer science undergraduate, master’s and PhD at the University of Nantes. Following his PhD, he took up a position at Singapore’s Nanyang Technology University, before deciding to move to Dublin in 2012.
After becoming a lecturer at University College Dublin (UCD) in 2015, he is now director of the UCD Complex Systems Lab at the School of Computer Science.
What inspired you to become a researcher?
I was lucky to grow up in a household of curious and creative people, and could have gone in many different directions including history, engineering, philosophy. I was lucky that there were lots of opportunities for me to develop my love of these things – tinkering in my dad’s workshop fixing cars and motorbikes, building an automatic gate at eight years of age, fixing tractors with my granddad and building fences on his farm.
My mum was also a huge influence, feeding my knowledge by bringing me to museums and castles in the Loire, sometimes even to archaeological digs. I also had amazing teachers, many of whom I still keep in touch with.
Can you tell us about the research you’re currently working on?
Currently, I am focused on artificial intelligence (AI)-based algorithms for complex systems. I am doing this through my collaboration with companies such as Microsoft and Zeeko, an Irish edtech firm headquartered at NovaUCD. I am also collaborating on this topic with fellow UCD researchers in other disciplines, such as applying machine learning to problems in archaeology, computational linguistics to chatbots, search and optimisation to software engineering. In particular, I’m looking at how to make these AI software systems scale and be more robust.
Chatbots, for instance, are of particular interest to me as they are typically complex software systems that require software developers to think about how they code in new ways, combining multiple services (eg language understanding, user intent extraction), large-scale infrastructures and noisy user inputs. Recently, Brad Smith, president of Microsoft, highlighted a chatbot that my team developed for refugees in one of his blogposts.
Another area where my group has made great progress is at the intersection between computer vision and machine learning. We have implemented a prototype of Sign Language Virtual Interpreter, showcasing how using complex compositions of cognitive services in the cloud can lead to high-value applications.
In your opinion, why is your research important?
Software nowadays is infinitely more complex. Coders rarely write code from scratch; they build systems that are made up of a patchwork of other code snippets – software libraries, APIs, services. I’ve heard them described as ‘Frankenstein-type algorithms’. The chance of failures increases exponentially in these heterogeneous environments. Software bugs are more difficult to detect and fix in advance of deployment.
My research is focusing on advancing software development and testing, which is critical as we increasingly allow these complex software systems to make more decisions for us in our daily lives. In some ways, we have never known less about how they really work.
What commercial applications do you foresee for your research?
Conversational agents – aka chatbots – are a hot topic at the moment. Gartner predicted that by 2020, 25pc of medium-to-large enterprises will have deployed product chatbots. They are easy to integrate with many tools and platforms (eg Skype and Facebook) and they use a mode of interaction (text and speech) that is natural to human users.
Ben Gomes, head of search at Google, recently announced that deployment of language technologies will be the next big leap for search engines. He gave many compelling examples of how speech interfaces will open up access to computational systems for all different types of users. These include those who traditionally may have been excluded from interacting with the internet or a computer through a disability, or even disadvantaged by speaking a language that requires more complex keyboard interactions. Making sure the software applications and systems are accessible to everyone is something that I anticipate will become very important to industry as it reaches for greater market share.
What are some of the biggest challenges you face as a researcher in your field?
Science is a challenging domain where you have to take risks – calculated risks, but risks nonetheless. For instance, I am currently working on a very complex and scientifically deep research question. I have spent months trying to crack it, with little success. This might be one of the biggest contributions in my field, or a complete waste of time – something we academics rarely know in advance.
One of the great strengths of the SFI research centres, such as Lero, is that they provide a virtual collaborative environment, which gives you access to loads of smart researchers for bouncing ideas off. This is the best way to advance science and overcome challenges.