ABSTRACT: From 1985 until the 1995 break-up of AT&T, Bell Labs in Holmdel, NJ was a hotbed of activity in machine learning and neural net research. Enormous talent gathered in the Holmdel group, including future Turing Prize winners, Yann LeCun and Yoshua Bengio, and National Academy member Vladimir Vapnik.
While at Holmdel, Bernhard Boser, Isabelle Guyon, and Vapnik invented Kernel Support Vector Machines, a widely used classification method. During the same period, LeCun pioneered the trained Convolutional Neural Nets that revolutionized pattern recognition and image analysis. Nearly all members of the group contributed to creating practical products that derived from their research, including high-volume character recognition technology used in check-reading.
This talk will describe the events that led to the formation of the Holmdel group and will highlight some of the achievements. We will also show how ConvNets, similar to those invented at Holmdel in 1989, were used 30 years later, at the same Holmdel site, to steer self-driving cars.
BIO: Larry Jackel is a physicist, machine learning pioneer, and former robotics program manager who served as Head, Adaptive Systems Research Department at AT&T Bell Labs in Holmdel, NJ from 1985 to 1995. During his tenure, his department incubated some of the foundational breakthroughs in modern artificial intelligence. He recruited, mentored, and managed AI pioneers—including Yann LeCun, Vladimir Vapnik, Yoshua Bengio, and Corinna Cortes—whose work directly led to the invention of Support Vector Machines (SVM) and early Convolutional Neural Networks (CNNs), like LeNet. From 1986 to 1996, Jackel ran the influential Snowbird Workshop (originally titled "Neural Networks for Computing"); in 1997 Yann LeCun took over and renamed it "Machines That Learn". Jackel’s department also built physical hardware chips like Net32K and ANNA to run mixed analog-digital computations for neural network convolutions. They successfully applied these systems to commercial optical character recognition (OCR), revolutionizing automated bank check and ATM processing for the financial industry.
Following the break-up of AT&T, Jackel was a Division Manager at AT&T Labs until 2002, where he pioneered projects like AT&T WorldNet webmail. From 2003 to 2007, he was a DARPA Program Manager responsible for conceiving pivotal autonomous ground vehicle initiatives like the Learning Applied to Ground Robots (LAGR) and Learning Locomotion programs.
Since 2015/2016, he has been a consultant for NVIDIA (contributing to their PilotNet end-to-end deep learning driving system) and as a machine learning advisor for the Toyota Research Institute (TRI). He holds a PhD in Experimental Physics from Cornell University and is an APS and IEEE Fellow.
Faculty Host: Debasis Mitra