Papers
1
Total Citations
161
H-Index
1
About
Kung is a pioneering figure in neural network simulation and high-performance computing, best known for developing one of the fastest back-propagation algorithms of its era. His landmark 1988 paper, "Neural network simulation at Warp speed: how we got 17 million connections per second" (161 citations), introduced a breakthrough approach to training neural networks using a linear array of processors on the Warp systolic array computer. By achieving an unprecedented 17 million connections per second, Kung demonstrated how parallel architectures could dramatically accelerate neural network computations, setting new benchmarks for speed and efficiency. This work not only advanced the practical implementation of back-propagation but also influenced the design of subsequent hardware-accelerated machine learning systems. Kung's contributions bridged the gap between theoretical neural network research and real-time simulation, inspiring generations of researchers in both artificial intelligence and computer architecture. His legacy endures in modern GPU-based deep learning frameworks, which owe a debt to his early innovations in parallel neural computation.
Research Focus
Key Achievements
Top Papers
- 1