Neural computation by time concentration
First Claim
1. A system for recognizing at least one pattern comprising:
- a plurality of stretchable delay means, each responsive to an applied input signal and each developing a set of different delayed signals, with said different delayed signals in each set differing from each other in delay and duration;
an interconnection network for selectively connecting said delayed signals to a plurality of outputs; and
a network responsive to said plurality of outputs for developing a preselected output signal indicative of recognition of said pattern.
1 Assignment
0 Petitions
Accused Products
Abstract
Apparatus that solves the problem of pattern recognition in a temporal signal that is subject to distortions and time warp. The arrangement embodying the invention comprises a neural network, an input interconnection network, and a plurality of signal modification circuits. A plurality of input leads delivers a preselected characteristic stimulus to associated signal modification units, and in response to an applied stimulus, each signal modification unit develops a plurality of output signals that begins at the time of stimulus application, rises to a peak, and decays thereafter. The mean time delay of each output (time to reach the peak) is different for each of the modification unit output signals. The outputs of the signal modification units are applied to the input interconnection unit wherein connections are made in accordance with the sequences that are to be recognized.
41 Citations
8 Claims
-
1. A system for recognizing at least one pattern comprising:
-
a plurality of stretchable delay means, each responsive to an applied input signal and each developing a set of different delayed signals, with said different delayed signals in each set differing from each other in delay and duration; an interconnection network for selectively connecting said delayed signals to a plurality of outputs; and a network responsive to said plurality of outputs for developing a preselected output signal indicative of recognition of said pattern. - View Dependent Claims (2, 3, 4, 5, 6)
-
-
7. A pattern recognition system responsive to a input signal composed of concatenated pattern components, Xm, and developing a plurality of output signals, Vi, comprising:
-
a feature detector responsive to said input signal for developing output signals DX.sbsb.m (t-τ
), where m is a selected integer in the range 1 to M, M being an integer, developing and output signal that is 1 when pattern component Xm is present within said input signal at time t=τ and
0 otherwise, τ
being a selected constant,an M plurality of conditioners, each responsive to one of said DX.sbsb.m (t-τ
) signals and developing K conditioner output signals, with each of said conditioner output signals being responsive to said DX.sbsb.m (t=τ
) signal and to fk (τ
), where k is an integer in the range 1 to K, K being another integer, and fk (τ
) is a selected stretchable delay function that relates the kth output signal of a conditioner to the conditioner'"'"'s input signal each of said fk (τ
) functions being different from each other; anda decision circuit responsive to output signals of said conditioners that develops an output signal in accordance with an energy function describing the behavior of the circuit that decreases with time during operation of said decision circuit and includes the component ##EQU8## where Ti Xm ;
k is a constant having a contributing value when Xm with delay provided by fk (τ
) contributes to the activation of said signal Vi as part of said pattern recognition, and having a noncontributing value otherwise. - View Dependent Claims (8)
-
Specification