LIACS Research Seminar
- Thursday 1 September 2016
- Snellius Room 174
Prof. dr. Wessel Kraaij
Quantified self for measuring and influencing well-being at work
Wearable sensors have become commodity gadgets and are used e.g. for sports and health monitoring. In addition, offices and smartphones are increasingly equipped with sensors. In the COMMIT/ project SWELL (www.swell-project.net) we have investigated how an instrumented work environment can help to signal high workloads as a basis for personalized burn-out prevention. In the talk I will present the main ideas developed in the project and zoom in on a study of automatic workload recognition based on classifying the visual features in facial expression. We show that a model that segmenting the test population in similar users helps to improve accuracy, since the facial expression data exhibits strong individual differences.
About the speaker
Wessel Kraaij started March 1 as a part-time professor 'Applied data analytics' at LIACS. He is also a principal scientist at TNO The Hague at the Data Science research group. His current research interests are: digital health, multimedia information retrieval, text mining, sense making from heterogeneous sensor data, privacy respecting analysis of personal data.
Dr. Michael Lew
How far has deep learning come in computer vision?
Computer vision is the ability for a computer to understand the world using visual imagery, similar to the way in which humans can perceive the world through their eyes. In this talk, I will briefly go over the past, present and future of deep learning for computer vision. This will touch upon classic computer vision topics such as edge detection and also the holy grail challenge known as “bridging the semantic gap” where the computer receives an image and should bridge the gap between the low level representation of pixels and output meaningful descriptions in human vocabulary. I also introduce recent advances by the LIACS Media Lab to the state-of-the-art that involve new architectures and knowledge fusion methods in deep neural networks and then move to important challenges for the future of deep learning. Some of the limitations and current capabilities of the deep learning networks will be illustrated with a live LIACS demo.
About the speaker
Michael Lew is co-head of the Computer Systems, Imagery and Media Research Cluster and co-director of the LIACS Media Lab. He is also the Editor-In-Chief of the International Journal of Multimedia Information Retrieval. His research interests are to discover new paradigms for the retrieval and uderstanding of multimedia information using deep learning approaches and data science techniques.