FIELD: data processing.
SUBSTANCE: invention relates to a method of training a binary neural network for object recognition. In the method, without the use of computational methods and a teacher, a system of clusters of communication channels of detecting neurons is instrumentally created, for which significant communication channels of the recording neurons are connected to informatively significant elements - communication channels of input binary sequences - code combinations, that is, each recording neuron is tuned to informative properties of input signals; levelling integral values arriving at recording neurons via communication channels from input objects; when new objects are added, threshold value of non-linear threshold function of detecting neurons is synchronously changed.
EFFECT: technical result is higher efficiency of training a neural network.
1 cl, 11 dwg
| Title | Year | Author | Number |
|---|---|---|---|
| EFFECTIVE PERCEPTRON BASED ON MCCULLOCH-PITTS NEURONS USING COMPARATORS | 2019 |
|
RU2729554C1 |
| SINGLE-LAYER PERCEPTRON BASED ON SELECTIVE NEURONS | 2015 |
|
RU2597497C2 |
| INFORMATION ENCODING METHOD IN PULSED NEURAL NETWORKS | 2019 |
|
RU2748257C2 |
| NEURON SIMULATION OF REAL NEURON PROPERTIES | 2014 |
|
RU2597495C2 |
| SINGLE-LAYER PERCEPTRON, SIMULATING REAL PERCEPTRON PROPERTIES | 2015 |
|
RU2597496C1 |
| NEAR-REAL IMPULSE NEURON | 2015 |
|
RU2598298C2 |
| NEURON MODEL BASED ON DENDRITIC COMPUTING | 2021 |
|
RU2777262C1 |
| METHOD FOR DIAGNOSING INFORMATION-CONVERTING ELEMENTS OF AIRCRAFT ON-BOARD EQUIPMENT BASED ON MACHINE LEARNING | 2022 |
|
RU2802976C1 |
| NEURON MODEL, REALIZING LOGICAL NONEQUIVALENCE FUNCTION | 2003 |
|
RU2269155C2 |
| METHOD FOR DETECTING OBJECTS IN THE IMAGE OF THE PLAN-SCHEME OF THE CONSTRUCTION OBJECT | 2022 |
|
RU2785821C1 |
