Machine Learning with Elliott
We speak to Elliott, a PhD student at the University of Bristol, about his research and his short and sweet videos available on the Taught By Humans website.
As part of our new short and sweet series, Elliott has made us 5 minute videos on topics ranging from machine learning to graphic design. So, this week we spoke to him about the PhD research that inspired these videos and his experience of producing video content communicating his knowledge with the public, for your benefit!
Over to Elliott!
What is the subject of your PhD and can you summarise why you are interested in that subject?
In my PhD I’m trying to explore how we allow humans to work with teams of robots. Specifically, I study swarm robotics which takes inspiration from swarms in nature. One example of this is the behaviours of bees and how they can form really interesting complex behaviours without the need for any leader in the hive. The bees work with their neighbours to solve different problems like foraging for nectar. In swarm robotics, we borrow these principles to generate teams of robots that don’t need a designated leader. I really enjoy the connection of swarm robotics to biology and nature to help us think about how to design effective teams.
A specific problem that I have been exploring is how we can allow a human to work alongside a swarm of robots in order to help them with some task. I think this is really important to consider when we think about real-world deployment of swarms, where human interaction will be required. When thinking about humans as part of the swarm system, we want the human to be able to help monitor and control the swarm if something isn’t going quite right. This creates a really interesting new dimension to these problems, thinking about how humans can observe what the swarm is doing, and what tools should the human have to be able to help the swarm. These are the fundamental questions I’ve been trying to explore during my PhD to promote affective human-swarm teaming.
If someone were looking to learn about your subject, how would you recommend they go about it?
Whilst I learnt a lot of what I know through academic papers which I often find through Google Scholar, I don’t recommend this as the best jumping off point. I think the best way to dive into some of these topics in more detail would actually be through Youtube. There are some great channels out there covering topics in AI and robotics. I really like the channel Primer, which explores ideas in artificial evolution along with many other topics. They also use really nice and easy to understand simulations demonstrating different experiments.
You can also find my topics related to robotics given in the form of TED talks. Specifically you can find a TED talk on swarm robotics given by Dr Sabine Hauert on Youtube. I also personally find videos or presentations an easier way to digest new ideas rather than reading papers or articles. If you then want to find some more specific details on a certain topic, for example artificial evolution, you can find many great books covering these topics in great detail, whilst still being easy to follow.
What topic did you choose for your short and sweet videos and why?
For my short and sweet videos I covered some of the main themes I have explored during my PhD. This includes Swarm robotics, Artificial Evolution, and Behaviour Trees. These three areas link together very nicely and cover entire fields of research in their own right. I find them all really fascinating areas which explore how we think about designing robots and AI. In making each of these short and sweet videos I aimed to highlight why I find these areas so interesting and powerful as a roboticist. Whilst these relate nicely to robotics, these methodologies can be applied to a wide range of problems, which I think makes knowledge of these areas really valuable.
In addition to these videos, I also covered topics on graphic design with Inkscape and a short introduction to the fundamentals of photography. I have really enjoyed learning more about graphic design over several years and I have found learning more about Inkscape has been very beneficial to me in easily creating impactful figures. Whilst I learnt to use Inkscape originally for my PhD, I have also used it to help design elements of my own website. I think having good skills in graphic design can be useful if you want to create anything. I also really enjoyed talking about photography which is something I’ve become very passionate about over the last couple of years. When starting with my first camera, it took me some time to properly understand how to control my images. But as I properly developed my understanding I now know exactly how to achieve the look I want which was something I wanted to share.
What made you choose to take part in the content creation scheme for Taught By Humans’ short and sweet series?
I loved the idea of short and sweet videos. It can sometimes be very daunting knowing where to start with a new subject, especially with more complex ideas. I have in the past had this exact problem when trying to learn about certain areas within machine learning. Because of this, I think the short and sweet videos are a great jumping off point, offering a wide range of different areas to explore.
Your videos are brilliant and very informative, what advice would you give to those interested in content creation?
I think content creation can feel a little daunting initially given the amount of work needed to plan what you want to create. My advice would be to take small steps to build up what you want to create. It’s easy to just think about how much work it will take to reach the final product, but thinking about the small steps you need to take to get your content out there to be much more affective. Part of this is planning exactly what you want to convey and having some form of a plan of the steps you need to take to make it happen. I found in creating my short and sweet videos, having a clear set of points I wanted to cover in each video really useful and made the process easier.
If you would like to see more about Elliott, you can find him on Twitter and LinkedIn. You can also read his paper on evolving supervision of swarm robotics, and take a look at his photography page on Instagram.
You can also enjoy a piece of Elliott’s short and sweet videos about machine learning by watching his video on artificial evolution at https://vimeo.com/580447045.
