Abstract
In this paper, we propose a new digital implementation of neural networks which operates in synchronous pulse-mode and allows massively-parallel processing. The basic signaling scheme is devised based on the Manchester encoding of pulse-stream signals. This pulse mode coding allows signed addition with a single OR gate and signed multiplication with a single AND gate and a two-to-one multiplexer. Each neuronal element keeps a set of registers which hold the information of weights. The activation function is implemented as a look-up-table (LUT). This structure provides programmability in weights, the number of inputs, and activation function of a neuron.
Original language | English (US) |
---|---|
Pages | 65-70 |
Number of pages | 6 |
State | Published - 1993 |
Event | Proceedings of the 1993 Artificial Neural Networks in Engineering, ANNIE'93 - St.Louis, MO, USA Duration: Nov 14 1993 → Nov 17 1993 |
Other
Other | Proceedings of the 1993 Artificial Neural Networks in Engineering, ANNIE'93 |
---|---|
City | St.Louis, MO, USA |
Period | 11/14/93 → 11/17/93 |