Ricardo Cachucho

Hi, I am currently PhD candidate in the LIACS data mining group at the university of Leiden, working under the supersision of Arno Knobbe. My background involve a bachelor in economics and a master in data analytics at the Economics Faculty of Porto (FEP). Originally inspired to study time series after studying evolutionary economics (see, path dependency theories), for my PhD I decided to focus on data mining techniques for time series. Some of my current research interests and projects are:

Mining Time Series
For the last 3 years I have been focusing on algorithms and tools that can help us understanding temporal phenomena, both for supervised and unsupervised settings. I have been fundamentally interested in the possibility of modelling and understanding evolving systems (group dynamics, body, mind, structures, businesses... ) by the use of data. This fundamental question lead me to more practical tasks of data collection (sensors), data preparation, feature construction and selection, mining algorithms and its applications. We are developing in-house a software for time series, that can help in some of the tasks mentioned above.

Activity Recognition and Healthcare Monitoring
The problem of activity recognition has insterested me since I decided to explore it as a topic for my master thesis. In this problem, the task is to classify a person’s activities into a finite set of classes. Activities can be predicted using body-worn or environmental sensors, for example measuring physiological parameters (e.g. heart rate), acceleration or position in space. I have explored this problem in multiple publications. With our partners at the LUMC, we are developing an activity recognition model which will serve multiple projects (GOTO, SWITCHBOX and AGO studies). More to come soon.

Sports Analytics
Most modern professional sport teams are producing data in abundance. Sensors, automatic diary data loggers and physical check-ups are more and more commun in sports communities. Recently we collected bio-signs (heart rate, breath rate, temperature... ), accelerometry and location data from a marathon athlete, during his trainings and marathon event. Part of this data was used for my publication at Ubicomp 2014 (see publications). More challenges to come soon.

Structural Health Monitoring
Infrastructures like roads and bridges are built to endure harsh conditions, such as heavy duty traffic or extreme weather conditions. However, on the long run harsh conditions take their toll and deteriorate the infrastructures over time. InfraWatch is a structural health monitoring project to study how the quality of infrastructure assets evolve over time, involving the collaboration between Leiden University and Delft University of Technology. At LIACS, we are mining the data from sensor networks installed in infrastructures, especially the network installed by our industrial partner Strukton at the Hollandse Brug. Recently, we have been experimenting with a ground based radar (FastGBSAR) from Metasensing.

The topics and projects mentioned above are only indicative of my current activities. Seen from a broader perspective, I am interested in practical data mining and machine learning challenges, especially if developed with/for multidisciplinary projects.

 

Ricardo Cachucho