When nothing is sure, everything is possible!
John Guiver - Research Software Design Engineer @ Microsoft Reseach Cambridge, UK
When nothing is sure, everything is possible!
This wonderful quote from Margaret Drabble could be a tag line for our exhibit. When we embrace uncertainty in everyday life, an inescapable consequence is that any outcome is possible. The same is true if we build mathematical models which embrace uncertainty. Our exhibit aims to help you understand about the science of uncertainty, and how we can quantify it and reason with it at a very large scale to build more intelligent computer programs.
Our choices and decisions are seldom made in a context of certainty, but we try to balance out the uncertainties and make the best decisions we can. To take a simple example, if the Met Office forecasts a '70% chance of rain', we might decide to take a waterproof jacket when going on a walk. But what does '70% chance of rain' really mean? Well, we can imagine a 'Groundhog Day' scenario where we wake each morning on the same day 100 times in succession; then about 70 of those days will show rain, and the remaining will be dry. So both outcomes are possible though one is more likely than the other. But we can go beyond that and say that the 70% probability itself is uncertain! For example it may be 70%, but it could also be 65% or 75% with a lower probability.
Take a look at our Clinical Trial demo to understand uncertainties on probabilities in more detail. And talking about 'Groundhog Day', make sure you check out the Movie Recommender demo which is a lot of fun and provides personalised recommendations of movies whilst showing the science behind in understandable terms. Both those demos are close to my heart as I spent several weeks building and fine-tuning them prior to the exhibition.
We've known how to reason with probabilities for about 250 years, but only recently have we known how to do this at a large scale. Some of the mathematical algorithms to do such reasoning have been developed by our researchers over the last few years. We have also been hard at work on creating a framework to make it much easier to write computer programs that incorporate these ideas. I'll be there on Saturday and Sunday, the last two days of the exhibition, so I look forward to seeing you. Unsure as to whether you'll enjoy it? It's highly probable with some uncertainty!