He was known for work involving something called convolutional neural networks, which he had developed for software that banks use to scan and read checks.    From Wordnik.com.  [NYT > Home Page] Reference

Viterbi encoding or half-rate convolutional coding?.    From Wordnik.com.  [Making Light: Heads Up For Our Friends at Random House] Reference

Thus there is a significance in the convolutional pattern of the brain.    From Wordnik.com.  [The Brain and the Voice in Speech and Song] Reference

What made this feat possible was the depth of the student's convolutional neural network.    From Wordnik.com.  [ACM TechNews] Reference

A convolutional neural network achieves 99. 26% accuracy on a modified NIST database of hand-written digits.    From Wordnik.com.  [The Code Project Latest Articles] Reference

VLIW (very-long-instruction-word) ConnX 545CK DSP has eight-way SIMD units, dual 128-bit load/stores, and a Viterbi convolutional coder accelerator.    From Wordnik.com.  [Engineering Hardware-Software] Reference

A newer class of interated short convolutional codes that are less complex than the Viterbi algorithm on the long convolutional codes that would be required for the same performance.    From Wordnik.com.  [RCR Wireless News >> FCC] Reference

He developed the sequential decoding techniques for convolutional codes that made error-free communication possible with relatively low computing power - enabling the subsequent development of modern strategies used by the Internet, cellular phones, and deep-space transmissions.    From Wordnik.com.  [Media Newswire] Reference

In their experiments, the students, led by professor Tom Li, discovered that the optimal number of layers for musical genre recognition was three convolutional (or "thinking") layers, with the first layer taking in the raw input data and the third layer outputting the genre data.    From Wordnik.com.  [ACM TechNews] Reference

The students used a convolutional network to.    From Wordnik.com.  [ACM TechNews] Reference

All 802. 11a/g features are supported, including data scrambling, forward error correction via convolutional encoding at rates from 1/2 to 3/4, sub-channel coding of BPSK, QPSK, 16QAM or 64QAM, both short - and long-training sequences, pilot insertion, and the IFFT/FFT required to convert to/from time-domain RF I/Q signals.    From Wordnik.com.  [Electronicstalk - electronics industry news] Reference