Xavier L. Aubert


Xavier L. Aubert studied in Belgium at the Engineering faculty of the University of Louvain, got his University degree in 1978 as "Civil Engineer in Applied Mathematics" and did a PhD in "Numerical Methods for Simulation of Free-Surface Flows" at the same university. He joined Philips Research Laboratories Aachen in September 1991 where he worked during twelve years in the field of Automatic Speech Recognition (ASR), mainly on the conception, design and implementation of large vocabulary decoding algorithms, in the statistical framework of Hidden Markov Models. Most of his publications (~25) are related to these ASR years. After the Philips session of ASR activities, he joined the Medical Signal Processing group where he worked mainly on 2 health-related topics, namely, non-invasive cuff-less techniques for estimating the arterial blood pressure and unobtrusive sleep monitoring at home based on mechanical sensing. Since October 2010, he has been transferred to the Philips Research Laboratory in Eindhoven, where he is currently working in the Circadian Rhythm Management project dealing with non-invasive methods for predicting the circadian phase of a subject in real-life. Recently he got a poster presented at the EBRS 2011 conference held in Oxford, August 2011, under the title "Influence of the Sleep-Wake Cycle on Phase Predictions using a Human Circadian Pacemaker Model". His latest publication co-authored with D. Sellitri is titled "Rest Interval Segmentation from Activity and Light Measurements, and its use in Circadian Phase Estimation", issued in May 2012.  

  • Medical signal processing techniques
  • Search algorithms
  • Circadian pacemaker modeling
  • Ballisto-cardiography
  • Automatic speech recognition
Provide support based on my expertise gained over 25 years of research in the field of applied sciences, with a particular emphasis on numerical simulation methods. Prevent re-inventing the wheel by a continuous follow-up of the scientific literature. Try making things simpler and transparent.