RU2019105275A - Method of coding information in impulse neural networks - Google Patents

Method of coding information in impulse neural networks Download PDF

Info

Publication number
RU2019105275A
RU2019105275A RU2019105275A RU2019105275A RU2019105275A RU 2019105275 A RU2019105275 A RU 2019105275A RU 2019105275 A RU2019105275 A RU 2019105275A RU 2019105275 A RU2019105275 A RU 2019105275A RU 2019105275 A RU2019105275 A RU 2019105275A
Authority
RU
Russia
Prior art keywords
pulse
sequences
input
partial
intensity
Prior art date
Application number
RU2019105275A
Other languages
Russian (ru)
Other versions
RU2748257C2 (en
RU2019105275A3 (en
Inventor
Михаил Ефимович Мазуров
Original Assignee
Михаил Ефимович Мазуров
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Михаил Ефимович Мазуров filed Critical Михаил Ефимович Мазуров
Priority to RU2019105275A priority Critical patent/RU2748257C2/en
Publication of RU2019105275A3 publication Critical patent/RU2019105275A3/ru
Publication of RU2019105275A publication Critical patent/RU2019105275A/en
Application granted granted Critical
Publication of RU2748257C2 publication Critical patent/RU2748257C2/en

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/04Architecture, e.g. interconnection topology
    • G06N3/049Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/06Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
    • G06N3/063Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons using electronic means
    • G06N3/065Analogue means

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Artificial Intelligence (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • Computational Linguistics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Neurology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Image Analysis (AREA)
  • Radar Systems Or Details Thereof (AREA)

Claims (1)

Способ кодирования информации в импульсных избирательных нейронных сетях при распознавании объектов и их интенсивности, когда используют импульсные последовательности постоянной амплитуды и длительности и со средней частотой следования импульсов пропорциональной интенсивности входного сигнала, основанный на изменении параметров или структуры входных импульсных последовательностей, отличающийся тем что, в качестве единичного символа информации для кодирования используют всю целиком парциальную импульсную последовательность из числа всех входных или ее отсутствие; создают кластеры парциальных импульсных последовательностей, избирательно настроенных на позиционную существенную входную информацию; импульсные последовательности, которые соответствуют информативно несущественной входной информации, полагают нулевыми; далее производят безынерционное суммирование мгновенных временных значений всех парциальных импульсных последовательностей каждого из образованных избирательных кластеров и получают в каждом регистрирующем нейроне после суммирования импульсные последовательности с чередующимися пиками амплитуды, равными сумме пиков парциальных импульсных последовательностей, что автоматически реализуют, благодаря фундаментальным свойствам равномерных почти периодических функций; далее возбуждают пиками импульсов суммарных последовательностей пороговый ждущий автогенератор, который преобразует пики суммарных импульсных последовательностей на его входе в импульсы или последовательности импульсов на выходе; если интенсивность объекта на входе изменяется, то синхронно изменяется «средняя частота» всех парциальных последовательностей на входе, что приводит к изменению «средней частоты» импульсной последовательности на выходе импульсного автогенератора, равной почти периоду парциальных импульсных последовательностей на его входе, при этом точность позиционного распознавания остается инвариантной (независимой) относительно интенсивности входного сигнала объекта.A method of coding information in pulse selective neural networks when recognizing objects and their intensity, when using pulse sequences of constant amplitude and duration and with an average pulse repetition rate proportional to the intensity of the input signal, based on a change in the parameters or structure of input pulse sequences, characterized in that, as a single symbol of information for encoding use the entire partial pulse sequence from among all input or its absence; create clusters of partial pulse sequences selectively tuned to positional essential input information; pulse sequences that correspond to informatively insignificant input information are assumed to be zero; then inertialess summation of instantaneous time values of all partial impulse sequences of each of the formed selective clusters is performed and impulse sequences with alternating amplitude peaks equal to the sum of the peaks of partial impulse sequences are obtained in each recording neuron after summation, which is automatically implemented due to the fundamental properties of uniform almost periodic functions; then, a threshold waiting oscillator is excited by the peaks of the total pulse sequences, which converts the peaks of the total pulse sequences at its input into pulses or pulse trains at the output; if the intensity of the object at the input changes, then the "average frequency" of all partial sequences at the input changes synchronously, which leads to a change in the "average frequency" of the pulse sequence at the output of the pulse generator, which is almost equal to the period of the partial pulse sequences at its input, while the accuracy of positional recognition remains invariant (independent) with respect to the intensity of the input signal of the object.
RU2019105275A 2019-02-26 2019-02-26 Information encoding method in pulsed neural networks RU2748257C2 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
RU2019105275A RU2748257C2 (en) 2019-02-26 2019-02-26 Information encoding method in pulsed neural networks

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
RU2019105275A RU2748257C2 (en) 2019-02-26 2019-02-26 Information encoding method in pulsed neural networks

Publications (3)

Publication Number Publication Date
RU2019105275A3 RU2019105275A3 (en) 2020-08-26
RU2019105275A true RU2019105275A (en) 2020-08-26
RU2748257C2 RU2748257C2 (en) 2021-05-21

Family

ID=72233764

Family Applications (1)

Application Number Title Priority Date Filing Date
RU2019105275A RU2748257C2 (en) 2019-02-26 2019-02-26 Information encoding method in pulsed neural networks

Country Status (1)

Country Link
RU (1) RU2748257C2 (en)

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2007071070A1 (en) * 2005-12-23 2007-06-28 Universite De Sherbrooke Spatio-temporal pattern recognition using a spiking neural network and processing thereof on a portable and/or distributed computer
RU2597497C2 (en) * 2015-01-13 2016-09-10 Михаил Ефимович Мазуров Single-layer perceptron based on selective neurons
RU2598298C2 (en) * 2015-02-09 2016-09-20 Михаил Ефимович Мазуров Near-real impulse neuron
RU2597496C1 (en) * 2015-02-24 2016-09-10 Михаил Ефимович Мазуров Single-layer perceptron, simulating real perceptron properties
US10671912B2 (en) * 2016-09-13 2020-06-02 Sap Se Spatio-temporal spiking neural networks in neuromorphic hardware systems
US10922608B2 (en) * 2017-03-08 2021-02-16 Arm Ltd Spiking neural network

Also Published As

Publication number Publication date
RU2748257C2 (en) 2021-05-21
RU2019105275A3 (en) 2020-08-26

Similar Documents

Publication Publication Date Title
CN108926333B (en) System and method for heart rate estimation
CN114093377B (en) Splitting normalization method and device, audio feature extractor and chip
MX2021003066A (en) Improving return signal detection in a light ranging and detection system with pulse encoding.
JP2017501387A (en) Method and device for obtaining when a low light pulse passes a threshold
RU2019105275A (en) Method of coding information in impulse neural networks
Liu et al. Ustc-nelslip system for dcase 2018 challenge task 4
DE60239341D1 (en) SYSTEM AND METHOD FOR SIGNALFORM PROCESSING
CN108880698A (en) Communication means and device based on cetacean cry pulse time length
Malik et al. Automatic threshold optimization in nonlinear energy operator based spike detection
Khandelwal et al. FMSG-NTU submission for DCASE 2022 Task 4 on sound event detection in domestic environments
Park et al. A fast and accurate convolutional neural network for LPI radar waveform recognition
Keshavarzi et al. A new method for detection of complex pulse repetition interval modulations
CN103033802B (en) Precise time delay processing device of laser pulse echo detecting instrument
CN108254064A (en) A kind of optical fiber vibration sensing detection method and device
US20190012515A1 (en) System and method for detecting spikes in noisy signals
RU2547095C1 (en) Radio pulse detector
Cho The Malware Detection Using Deep Learning based R-CNN
Madhu et al. Radar emitter identification using signal noise and power spectrum analysis in deep learning
RU2304061C1 (en) Automatic cab signaling adaptive signal receiver
US9471814B2 (en) Reader and method for reading data
DE50206345D1 (en) METHOD FOR ALLOCATING A PULSE COURSE TO A DIFFERENT RETURN TIME FROM A MULTIPLE OF PULSES AND DEVICE FOR THEIR IMPLEMENTATION
CN108960032B (en) Tristable logic stochastic resonance method
CN102315851B (en) Counting, phase-detecting and decoding device and method
RU2821272C1 (en) Method of detecting pulse signals in discrete-continuous recording of electromagnetic radiation
RU2586825C1 (en) Method and device for measuring frequency of rotation

Legal Events

Date Code Title Description
FZ9A Application not withdrawn (correction of the notice of withdrawal)

Effective date: 20200915

MM4A The patent is invalid due to non-payment of fees

Effective date: 20210227