IEEE Signal Processing Society
Distinguished Lecturer Program

http://www.ieee.org/organizations/society/sp/dlprogram.html

Professor Robert M. Gray
http://ee.stanford.edu/~gray/
2006

If you are renting an apartment in Paris, you might consider this warning based on our unfortunate experience and financial loss at the hands of a rental agency.


Bio

Lecture Topics: (See below for further detail and example slides.)

  1. Extracting discrete information from a continuous world: Quantization, Compression, and Classification
  2. Packet speech on the Arpanet: A history of early linear predictive coded (LPC) speech and its accidental impact on the Internet Protocol
  3. Mentoring for Engineering Academia

Contact Information

Robert M. Gray
Lucent Technologies Professor of Engineering
261 Packard Electrical Engineering Building
350 Serra Mall
Stanford, CA 94305
USA
P: +1 650 723 4001 (Stanford)
email: rmgray at stanford.edu

Extracting discrete information from a continuous world: Quantization, Compression, and Classification

Scientists and engineers often seek to measure, communicate, store, process, reproduce, or analyze signals encountered in the real world. Most such signals are inherently continuous or analog in nature, yet increasingly the means for communicating, storing, and manipulating such information are discrete or digital. Generally something is lost when continuous information is converted into discrete approximations, so a natural goal is to preserve as much of the original information as possible. This is the general problem of quantization, a technique that historically has cropped up in a variety of branches of signal processing, taxonomy, physics, mathematics, and statistics as well as playing a key role as the interface between a continuous world and digital processing. Quantization traditionally has been used to model analog to digital conversion, Shannon source coding, and data compression. Viewed generally, quantization also models the extraction of information from signals, including statistical classification, clustering methods, and machine learning. This talk will describe the fundamentals of quantization along with examples and recent research topics in theory and application.

This topic comprises one or more lectures and the technical level is adjustable from undergraduate to graduate student and expert audiences.

Packet speech on the Arpanet: A history of early linear predictive coded (LPC) speech and its accidental impact on the Internet Protocol

There are two variations to this talk. The first and shorter variation (approximately 50 minutes):

Voice over IP (VoIP) is popularly considered as a recent technology, but real-time packet speech was demonstrated on the ARPANet over three decades ago --- --before the IP protocol existed. This talk sketches the early history of both linear predictive coding (LPC) of speech and of network protocols for realtime low bitrate speech coding for the ARPANet. Beginning with a brief technical survey of the many approaches and developments of LPC, most of the talk is a narrative of the history of LPC and of the first real time successful packet speech demonstration on the ARPANET in 1974 and its impact on the development of the Internet protocol (IP). The talk expands on the the article ``Digital Speech and the Internet Protocol: The 1974 Origins of VoIP,'' IEEE Signal Processing Magazine, Vol. 22, July 2005, pp. 87--90.

The second variation is the current version of a talk originally entitled California Coding: Early LPC Speech in Santa Barbara, Marina del Rey, and Silicon Valley 1967-1982 which expands the subject matter to sketch the historical and technical threads of the California portion of the story of the development of linear predictive methods for speech coding along with joint and parallel work in Japan, New Jersey, Massachusetts and Texas. The focus is on the 1970s, but the story begins earlier and the narrative covers through the early 1980s. Personalities, institutions, and milestones are considered along with technical developments and interpretations. With the benefit of hindsight, a brief technical tour of the basics of linear predictive coding is provided to provide context for the history. The primary personalities considered are Glen J. Culler, John Parker Burg, John D. Markel, A.H. (Steen) Gray, Jr., and Danny Cohen. The institutions emphasized include UCSB, SCRL, ISI, Culler Harrison Inc., and Time and Space Processing. The focal events are the first real time LPC speech communication on the ARPArnet in 1974, the first hardware LPC speech boxes, the book "Linear Prediction of Speech" by Markel and Gray, and the appearance of TI's Speak & Spell toy. The technical threads involve several variations and interpretations of LPC and the encounters of early LPC research with the origins of the ARPAnet and Internet and the precursors of wavelet analysis. Anecdotal stories of interractions among researchers in California and foreign (nonCalifornia) institutions such as NTT, Lincoln Labs, Bell Labs and TI are recounted. The talk is based on oral histories, the literature, email and conversations, and the author's memories as a peripheral participant.

Mentoring for Engineering Academia

Dictionary definitions of "mentor" include "experienced and trusted adviser" and "trusted counselor or guide, tutor, coach." Effective mentoring is important to all persons pursuing an advanced degree in engineering and related technical fields, and especially to those beginning academic careers. Studies have shown that women often receive less mentoring and have a more difficult time being selected as proteges than males, especially in male-dominated fields such as engineering, which helps explain why fewer than 8.2% of electrical engineering Ph.D.s in the U.S. are awarded to women and why at the turn of the century there were only about 130 tenured and tenure-track women electrical engineering professors in the US. Although many factors affect these statistics and a variety of approaches exist to improve them, arguably the means by which a small number of people can have the maximum impact is by encouraging, assisting, and supporting those who understand the value of diversity and wish to become leaders in academia. This talk is a distillation of the presentations and discussions of a June 2004 workshop on mentoring for academic careers in engineering which was jointly supported by the Presidential Award for Excellence in Science, Mathematics, and Engineering Mentoring Program of the White House Office of Science and Technology and the Stanford University School of Engineering. The workshop had 80 participants including graduate students, junior through senior faculty, and representatives from a variety of mentoring programs. The conference Proceedings can be found at http://paesmem.stanford.edu in both pdf and html format.

Modified abstract for presentations in Turkey:

Dictionary definitions of ``mentor'' include ``experienced and trusted adviser'' and ``trusted counselor or guide, tutor, coach.'' Effective mentoring is important to all persons pursuing an advanced degree in engineering and related technical fields, and especially to those beginning academic careers. Lack of good mentoring can lead to missed opportunities and wasted potential. Many factors affect student success in general and the development of young faculty in particular and a variety of approaches exist to improve the situation, but arguably the means by which a small number of people can have the maximum impact is by encouraging, assisting, and supporting those who are considering or following an academic career. This talk is a distillation of the presentations and discussions of a June 2004 workshop on mentoring for academic careers in engineering which was jointly supported by the Presidential Award for Excellence in Science, Mathematics, and Engineering Mentoring Program of the White House Office of Science and Technology and the Stanford University School of Engineering. The workshop had 80 participants including graduate students, junior through senior faculty, and representatives from a variety of mentoring programs. The conference Proceedings can be found at http://paesmem.stanford.edu in both pdf and html format.


12 November 2006