BIGMATH: Mathematical morphology for the prediction of face expression transition
My name is Rongjiao Ji, I am currently a first-year PhD student in the PhD school in Mathematical Sciences, Universita’ degli Studi di Milano (Italy), and I am enrolled in the Marie Skłodowska-Curie Action – Innovative Training Network/European Industrial Doctorate BIGMATH. I received my master degree in Probability and Statistics at the Mathematics and Applications Department, Institute Superior Tecnico, University of Lisbon (Portugal). Before that, I got my bachelor degree in Information and Computing Science, Mathematics department, Dalian University of Technology (China).
In my bachelor’s last year, I took part in a one-year exchange program in which I finished my bachelor thesis and one-year credits of my master, and then, directly entered the second year of master. Afterwards, I was a research assistant at Signal Processing and Image Group under a one-year research reward in the Institute for Systems and Robotics.
Why I applied to the BIGMATH Marie Skłodowska-Curie fellowship
For complicated high-dimensional high-volume datasets, new models based on the data’s latent structures are in huge demand. The scientific goal of BIGMATH, to address the major challenges that the Big Data era is posing to mathematical research, naturally combines mathematics and data science in which fields I have great interests, and fits nicely my previous research experience.
What I will do during the fellowship
The project where I am involved is called “Mathematical morphology for the prediction of face expression transition”. Human face expression conveys informative but subtle messages, which attracts great interests of researchers trying to give convincing answers from different profiles. From the perspectives of mathematics and computer graphics, we intend to give a realistic description of facial expression, by constructing digital versions based on 4D data scanned from real humans. This project deals with the modeling of the shape and movements of a human face in their connection with the underlying emotions which are generating the expressions. The project will primarily focus on: a) the study of the relationship between face movements, expressions and emotions, with the aim to recognize automatically which ‘mixture of emotions’ is connected with some specific face movements; b) the study of the transition distributions between different facial expressions, and simulation of expressions and of ‘change of emotions’ in a realistic way. I will contribute to developing innovative stochastic geometric models and statistical methods in non-Euclidean spaces, and apply them to the real data provided by 3Lateral.