Tell us about yourself — what’s your story and what are you studying?
Back in high school (I studied at the Philippine Science High School - Main Campus), I was active in competitive mathematics both locally and internationally, having competed in the International Math Olympiad (IMO). This led me to pursue a combined BA/MA degree in Mathematics at Johns Hopkins University in Baltimore, Maryland, USA. As I got deeper into my studies, my interests shifted to the applied side of math like Computing (computer science). I decided to take up further studies with an MSc in Computing at Imperial College London (ICL).
Initially, I intended to only stay in ICL for a year before moving on to other work. However, the research component of my MSc went very well and I got along with my supervisor, so I decided to pursue a PhD. For my PhD, I focused on Adversarial Machine Learning – the field that studies the security and privacy of machine learning algorithms.
Much of my work is based on computer vision, which is used in object and face recognition, autonomous driving, biomedical imaging, and the like. Computer vision itself has so many applications that it is a vast field of research. The security aspects I explore primarily concern Adversarial Examples – seemingly benign-looking images that were designed to fool the machine learning algorithms. This is quite dangerous if, for example, you have an algorithm that automatically filters acceptable or unacceptable images and videos. You can use Adversarial Examples to alter the images/videos and trick the filter into letting through malicious content or to trick it to block benign content, thus disrupting the service. Another interesting aspect of machine learning security comes in privacy where these algorithms memorise the data given to them. This is a violation of data privacy as this information can often be extracted using certain techniques. These present a high-level view of my research, and you can imagine that these aspects of security and privacy extend beyond computer vision to other data types like sound, speech, and text.
For more examples, you can see my code on GitHub: (1) (2)
Why did you choose the UK as a study destination?
Initially, it was for the silliest reason—I found out that a master’s degree in the UK is typically only one year, and I didn’t want to study for too long again, so I thought it was a great opportunity! It is a bit ironic that I later decided to pursue a PhD. A lot of the appeal was also in the City of London where the university was located. From my previous experiences, I found that I preferred a city-type setting where it was easier to get around and there were always things to do. And being where it is, London is where many of the best researchers in the UK and Europe gather.
What draw you to the field of computing, particularly in machine learning?
Coming from a mathematics background, I became interested in applying what I’ve learned in more tangible ways. Computing (computer science) seemed like a natural step forward. Previously, the logic behind a lot of automation was done with conditional or rules-based code (e.g., if X then do Y). This method is easy if the data you work with is relatively simple (e.g. if the price is less than 100, then buy). However, as the data becomes more complex (e.g. multivariable financial data with time components), it becomes more difficult for algorithms to keep up if you only use conditional statements or rules.
The nice thing about machine learning is that you give it a set of inputs, and instruct the algorithm to match those inputs with the correct outputs. The core learning algorithm then learns internal parameters so it matches the correct solution as best as it can, given the inputs. So now you can do predictions or classifications on more complex data without having to tell any specific rules; you just need to have the data (loosely speaking). Many of these learning algorithms are mathematical, so these were still familiar to me, and it was nice to see the results when moving from designing the algorithm to the application in practice.
What role or impact do you think will computing (and science and technology in general) and computer scientists like you play in the promotion of peace and maintenance of security?
Computing currently plays a huge role in modern science and technology. It is the backbone of our technology infrastructure, and things like machine learning and artificial intelligence drive a lot of innovation. It’s difficult to talk about technology without mentioning the impact that social media and the wider internet has had on the distribution and consumption of information. After all, technology is a lever that can amplify many great and not-so-great things. As scientists, we are working better to understand not just the technical aspects, but also the ethical implications of the technologies that we design, as these can become very powerful tools for maintaining peace and improving the lives of people.
From the cybersecurity point of view, what's important for us to raise awareness of best practices and common threats to the public. The latter would include phishing scams, social engineering attacks, faults in the machine learning algorithms, and misinformation. Highlighting these threats would go a long way in making people vigilant and maintaining security.
What has been the most exciting part of your experience studying and living in the UK?
For my field of research, it was exciting since I was able to attend the top machine learning and cybersecurity conferences as well as meet with the leading people in the field all within the UK. Not to mention, I’ve met many great people in London from all kinds of different backgrounds. I appreciated the diversity in cultures and open exchange of ideas in the UK. I’m also a bit of a foodie, so London, in particular, was really great because of the diverse range of cuisines and good food choices!
What advice would you give to Filipinos who aspire to pursue an advanced degree in computing in the UK?
It is useful to think of what you’re studying (or plan to study) in the context of the problems it can solve. Sometimes we can focus too much on the academic or technical side of the work that we lose sight of what makes it valuable in practice. A lot of funding in the UK is based on projects and research grants geared towards achieving specific goals or solving specific problems, so when finding potential mentors or research groups, it is to your advantage if you can align your interests with the problem/s that they are trying to tackle.
As mentioned, our research group RISS at Imperial College London is on the lookout for high-quality PhD students. Our main research topics are machine Learning learning and cybersecurity, so please reach out to my email at k.co(at)imperial.ac.uk. Our research goals are focused on enabling systems to be resilient to malicious compromise and to continue to operate when they have been partially compromised. Even if your interests are not necessarily these, feel free to reach out anyway since I may be able to connect you with someone who can guide you to the right program or university!