The Applied Intelligence Research Centre has a diverse team.
My doctoral thesis investigates the application of active learning techniques to judgement-based label elicitation. It is well known that individuals are usually subject to biases when using rating scales. These biases lead to unreliable and incommensurate labels being collected. I am developing a system which reframes questions eliciting ratings on a scale of 1 – N (i.e. how much do you like X); instead requesting pairwise comparisons (do you prefer X or Y?). This system draws on advances in rank aggregation and active learning to optimise the data collection process, resulting in higher quality labels being collected.