FIELD: design of processors based on neural networks. SUBSTANCE: device has serial circuit of convergence unit, unit converting signals for transmission, divergence unit. These units are equipped with communication channels which are designed as serial circuit of photosensitive resistor and light source. Each communication channel of convergence unit has additional communication channel with additional photosensitive resistor which is connected in parallel to main photosensitive resistor in main communication channel, local memory units, which are designed using optoelectronic integration gates, and elements for feedback to optoelectronic converters of light flow. EFFECT: increased functional capabilities due to possibility of implementation of neural network learning and training algorithms. 2 dwg
Title | Year | Author | Number |
---|---|---|---|
NEURO-SIMILAR CIRCUIT MODULE | 0 |
|
SU1803923A1 |
ELECTRIC-OPTIC COMPUTER FOR SOLVING PARTIAL DIFFERENTIAL EQUATIONS | 0 |
|
SU1807505A1 |
HYDRIDE OPTRONIC CELL | 1991 |
|
RU2025776C1 |
HYBRID TRANSPUTER FOR OPTOELECTRONIC NEURAL PROCESSOR | 1992 |
|
RU2057363C1 |
OPTOELECTRONIC NEURON NETWORK SIMULATOR | 1992 |
|
RU2070334C1 |
OPTOELECTRONIC MODEL OF NEURAL NETWORK | 1991 |
|
RU2074417C1 |
SIGNAL DIVERGENCE ELECTRO-OPTIC UNIT | 0 |
|
SU1802366A1 |
HYBRID TRANSPUTER FOR OPTOELECTRONIC NEURAL PROCESSOR | 1992 |
|
RU2057362C1 |
MEMORY LOCATION | 1991 |
|
RU2050582C1 |
DEVICE FOR SOLVING TRANSFER THEORY PROBLEMS | 0 |
|
SU1420604A1 |
Authors
Dates
1999-09-10—Published
1994-10-14—Filed