Faculty of Science - Leading creativity and innovation in the sciences


Take 10 with... Dr Ciprian Giurcaneanu


ciprian-giurcaneanu

Dr Ciprian Giurcaneanu from the Department of Statistics gives us 10 minutes of his time to discuss his research about statistical signal processing.

1.      Describe your research

Statistical signal processing, with a special emphasis on model selection.

2.      Now describe it in layman’s terms!

First description: Most of my recent research is related to the concept termed renormalized maximum likelihood (RNML), which is a model selection method rooted in information theory. It was introduced by Rissanen as the most advanced embodiment of the Minimum Description Length principle. I have proposed novel RNML-based solutions for analysis of multivariate time series, hypothesis testing, adaptive filtering of time series, spectral estimation, construction of irregular histograms, histogram-based entropy estimation.

Second description: The scientists “transcribe” the beauty of the nature into ugly equations which are named mathematical models. I am trying to find the best mathematical model for the original beauty.       

3.      Describe some of your day-to-day research activities

Just three major activities: reading, writing and coding. The key point is to allocate the right proportion of time to each of them.

4.      What do you enjoy most about your research

The fact that it forces me to be always in competition with myself: Today I have to do something better than what I have published yesterday.

5.      Tell us something that has surprised you in the course of your research.

Sometimes my mathematical proofs contradict my own intuition about a certain phenomenon.

6.      What questions have emerged as a result?

Have I used a wrong assumption in my proofs?

7.     How have you approached any challenges you’ve faced in your research?

I allow myself to think about a certain problem for a given period of time (say… few weeks). During this time, I search for the results published by other scientists, I do simulations, and I do calculations. Even if I do not have a complete solution for the problem I investigate, I write a short report with all my findings. After a while, I have a look at these reports: sometimes they help me to find the solution, sometimes they help me to understand how naïve I was.

8.      What kind of impact do you hope your research will have?

Model selection is used in very many applications, but the researchers tend to employ the traditional methods. The newly proposed approaches can potentially lead to better results in most of these applications. This will only happen if the complicated mathematical formulae of the new selection rules will be derived for many classes of models.

9.      If you collaborate across the faculty or University, who do you work with and how does it benefit your research?

I collaborate with the research group of Professor David E Williams (Chemical Sciences) and with Associate Professor Stevan Berber (Electrical and Computer Engineering). The collaboration is beneficial because it gives me the chance to work on new problems. Finding a common language with researchers from other scientific communities is not an easy task, but it is funny.

10.    What one piece of advice would you give your younger, less experienced research self?

Do not take too seriously the advice coming from the senior researchers (including this one).

 

Read more about our research in the Take 10 with... series