Neural nets supplied synapse signals obtained by digital-to-analog conversion of plural-bit samples
First Claim
1. A system comprising:
- means for generating in serial order sets of P temporally aligned digital input signal samples, P being an integer greater than one, each of which temporarily aligned digital input signal samples has a plurality of bits B in number, each of which bits has one of two conditions which two conditions are represented by respective electric signal levels;
means for converting each set of P temporally aligned digital input signal samples to a corresponding set of temporally aligned analog input signal samples, which means for converting includesa plurality P in number of digital-to-analog converters, each for converting a respective one of said P temporally aligned digital input signal samples to a corresponding one of said temporally aligned analog input signal samples; and
a neural net layer, responsive to said corresponding sets of P temporally aligned analog input signal samples being applied thereto as synapse input signals, for generating in serial order respective analog samples of each of a number M of axonal output signals.
0 Assignments
0 Petitions
Accused Products
Abstract
Digital synapse input signals which are temporally aligned with each other are converted to analog form for application to a neural net layer. Several digital synapse input signals can be written into digital memory at different times, to be read out in temporal alignment for conversion to analog form for application to a neural net layer. The digital synapse input signals can originate from apparatus for raster-scanning a field or successive fields of video samples, for example, with the neural net layer providing filters for recognizing shapes in a field of the video samples. The neural net layer can be either a pre-programmed type or a learning type, and its axonal outputs may be digitized for use in subsequent digital processing.
-
Citations
40 Claims
-
1. A system comprising:
-
means for generating in serial order sets of P temporally aligned digital input signal samples, P being an integer greater than one, each of which temporarily aligned digital input signal samples has a plurality of bits B in number, each of which bits has one of two conditions which two conditions are represented by respective electric signal levels; means for converting each set of P temporally aligned digital input signal samples to a corresponding set of temporally aligned analog input signal samples, which means for converting includes a plurality P in number of digital-to-analog converters, each for converting a respective one of said P temporally aligned digital input signal samples to a corresponding one of said temporally aligned analog input signal samples; and a neural net layer, responsive to said corresponding sets of P temporally aligned analog input signal samples being applied thereto as synapse input signals, for generating in serial order respective analog samples of each of a number M of axonal output signals. - View Dependent Claims (2)
-
-
3. A system comprising:
-
means for generating in serial order sets of P temporally aligned digital input signal samples, P being an integer greater than one; means for converting each plurality Q in number of sets of P temporally aligned digital input signal samples most recently generated, as obtains each time a new one of said sets of P temporally aligned digital input signal samples is generated, to a corresponding group of temporally aligned analog input signal samples P times Q in number; and a neural net layer, responsive to said groups of temporally aligned analog input signal samples applied thereto as synapse input signals, for generating in serial order respective analog samples of each of a number M of axonal output signals. - View Dependent Claims (4, 5, 6)
-
-
7. A system including a neural net layer, said system comprising:
-
apparatus for performing a number M of weighted summations in the analog regime of temporally aligned analog synapse input signal samples, to generate M respective weighted summation results, being included in said neural net layer; circuitry for supplying a plurality P in number of streams of digital samples manifested in electric signal form, said streams of digital samples being herein and hereafter identified by respective ones of the consecutive numbers first through Pth, the samples in each of said first through Pth streams being supplied in a respective serial order; digital-to-analog conversion apparatus for generating groups of temporally aligned analog input signal samples, each said group of temporally aligned analog input signal samples being generated responsive to the plurality Q in number of most recently supplied samples in each of said first through Pth streams of successive digital samples, and each said group of temporally aligned analog input signal samples being supplied as temporally aligned analog synapse input signal samples to said apparatus for performing a number M of weighted summations in the analog regime; and circuitry included in said neural net layer for non-linearly processing said M respective weighted summation results to obtain respective axonal output responses. - View Dependent Claims (8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40)
-
Specification