Signal Processing

Rice Digital Signal Processing LogoToday’s sensors, signal processing, and machine learning systems are under increasing pressure to accommodate:

  • ever larger and higher-dimensional data sets, including high-resolution images and video, volumetric data, three-dimensional (3-D) video, 4-D+ light-fields, and beyond;
  • ever faster capture, sampling, and processing rates;
  • ever lower power consumption in order to permit remote, networked, battery operation for long periods;
  • communication over ever more difficult channels;
  • radically new sensing modalities.

Fortunately, over the last decades, there has been an enormous increase in computational power  thanks to Moore’s Law, which provides a new angle to tackle these challenges. We are currently on the verge of moving from a digital signal processing (DSP) paradigm, where signals are represented using periodic samples to a computational signal processing (CSP) paradigm where signals are measured and represented using dimensionality reduction techniques.  The concepts of sparsity,  randomization, and optimization play starring roles.

Current research projects include:

  • computational imaging to enable new kinds of sensing modalities
  • machine learning, including new kinds of deep learning systems
  • machine learning on educational data to close the learning feedback loop
  • sparsity-based signal processing and compressive sensing

For our latest results, see the DSP publications archive and my Google Scholar page.

Multi-university research projects based at Rice University include the ARO MURI on Opportunistic Sensing.

Support for these projects has come from NSF, ONR, DARPA, AFOSR, ARO, AFRL, DOE, NGA, EPA, NATO, and a number of industrial sponsors, including the Texas Instruments Leadership University Program.

Additional links

Comments are closed.