GSRC Student Profile:
Mohammed Shoaib
|
 |
(For more contact info, please log in.)
Princeton University
Advisor: Naveen Verma
GSRC theme: drivers
Expected graduation: Jun, 2013 |
Research Overview: Ultra-low-power integrated circuits in sensing, computing, and communication platforms for biomedical applications.
Biomedical devices are becoming key computational interfaces in advanced health-care platforms. In state-of-the-art algorithms extremely accurate inferences from biological signals are routinely made in real time. However, these algorithms have overshot power budgets in severely battery constrained applications by several orders of magnitude. Machine learning has been an important driver for these applications. This has been effectuated by the emergence of global patient databases. These databases are instrumental in the systematic development of efficient models to de-embed complex physiological states from biological signals using machine learning.
My research focuses on bridging the algorithmic advances with efficient hardware platforms which can support machine learning computations across a large class of biomedical applications. I am interested in looking at how we can trade-off aspects of “flexibility” for “efficiency with just enough reconfigurability” to adapt to the various computational needs of the algorithms. I have been looking at enabling aspects from the perspective of both computation and communication.
To bring the biomedical devices closer to the patient in a wearable or implantable system, computation and communication in these applications need to be counterbalanced. For the computational optimization, I have proposed an efficient classifier accelerator for Support Vector Machine (SVM) algorithms which form the inference backbone across a wide class of medical algorithms. In the future, I am interested in looking at optimized communication strategies for these systems incl. low data rate encoding and transmission.
|
