John S. Denker

Timeline

1. By 1985

   Caltech, Cornell and Bell Labs

   He is an undergraduate at Caltech, where he founds a small software and electronics company during his junior year, and does his doctoral research at Cornell on a gas of hydrogen atoms a few thousandths of a degree above absolute zero. In 1985, he joins the Adaptive Systems Research Department at AT&T Bell Laboratories. His Bell Labs roles include Distinguished Member of Technical Staff, Department Head, and Division Manager.

2. 1986–1987

   Editing Neural Networks for Computing

   In 1986, he edits Neural Networks for Computing, the American Institute of Physics proceedings of the April 1986 conference at Snowbird, Utah. From 1986 to 1987, he serves as Visiting Professor at the Institute for Theoretical Physics at the University of California, Santa Barbara.

3. 1988

   A zip code digit recognizer

   At NIPS 1988, he is first author of “Neural Network Recognizer for Hand-Written Zip Code Digits”, written with eight coauthors including Hans Peter Graf, L. D. Jackel and Isabelle Guyon. The system combines classical techniques with neural-net methods and is trained and tested on digits derived from zip codes on actual U.S. Mail. It marks the first of four computer-vision milestones in the Bell Labs group's neural-network research.

4. 1989

   Optimal Brain Damage and backpropagation

   At NIPS 1989, he coauthors “Optimal Brain Damage” with Yann LeCun and Sara A. Solla; the method removes unimportant weights from a neural network to gain better generalization, fewer training examples, and improved speed of learning or classification. In December 1989, he also coauthors “Backpropagation Applied to Handwritten Zip Code Recognition” in Neural Computation, with LeCun as first author.

5. 1992–1994

   Transformation distance and classifier comparison

   At NIPS 1992, he coauthors “Efficient Pattern Recognition Using a New Transformation Distance” with Patrice Simard and Yann LeCun; the distance measure can be made locally invariant to input transformations and consistently outperforms the other systems tested on the same handwritten character databases. In October 1994, he is one of eleven authors of “Comparison of classifier methods: a case study in handwritten digit recognition”, published in Jerusalem in the proceedings of the 12th IAPR International Conference on Pattern Recognition.

6. 1995–1999

   Adiabatic circuitry and Moat VPN

   In December 1995, he receives U.S. patent US5477164A for adiabatic dynamic noninverting circuitry, which makes a logic signal and its inverse simultaneously available in dynamic logic circuitry without significant power dissipation. At the 1999 USENIX LISA conference, he coauthors “Moat: A Virtual Private Network Appliance and Services Platform”, describing a virtual private networking system for telecommuters built with off-the-shelf hardware and open-source software.

7. 1996–2014

   Writing See How It Flies

   He writes See How It Flies, an online book on the perceptions, procedures, and principles of flight copyrighted 1996–2005 and 1996–2014, with 20 chapters, 140,000 words, and over 280 color illustrations. His site av8n.com also hosts physics documents written mainly to answer questions from Phys-L, the Forum for Physics Educators. He is certified as a Commercial Pilot, Flight Instructor, and Ground Instructor.

8. 2018–2025

   Physics uncertainty and hardware randomness

   In November 2018, he and Larry Smith coauthor “Guard Digits vs. Roundoff Error vs. Overall Uncertainty” in The Physics Teacher. In 2021, he writes the Turbid User’s Guide on configuring and using the Turbid hardware random number generator, and in January 2025 he coauthors the American Association of Physics Teachers Winter Meeting talk “The Trouble with Significant Figures” with Smith.