IP Library Granted Patent US 11,625,588
Granted Patent B2
US 11,625,588 · App. 16/809,522 · Granted Apr 11, 2023

Neuron circuit and artificial neural network chip

Inventors: Tuo-Hung Hou (Hsinchu, TW); Shyh-Shyuan Sheu (Hsinchu County, TW); Jeng-Hua Wei (Taipei, TW); Heng-Yuan Lee (Hsinchu County, TW); Ming-Hung Wu (Taoyuan, TW)
Assignee: Industrial Technology Research Institute
G06N3/063G06N3/04G06N3/08G11C11/1693G11C11/54G11C13/0004G11C13/0061
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 11,625,588
App. No.
16/809,522
Granted
Apr 11, 2023
Kind
B2
Abstract

A neuron circuit and an artificial neural network chip are provided. The neuron circuit includes a memristor and an integrator. The memristor generates a pulse train having an oscillation frequency when an applied voltage exceeds a predetermined threshold. The integrator is connected in parallel to the memristor for receiving and accumulating input pulses transmitted by a previous layer network at different times, and driving the memristor to transmit the pulse train to a next layer network when a voltage of the accumulated input pulses exceeds the predetermined threshold.

Claims (20)

1. A neuron circuit, comprising:

a memristor, generating a pulse train having an oscillation frequency when an applied voltage exceeds a predetermined threshold; and

an integrator, connected in parallel to the memristor for receiving and accumulating input pulses transmitted by a previous layer network at different times, and driving the memristor to transmit the pulse train to a next layer network when a voltage of the accumulated input pulses exceeds the predetermined threshold.

2. The neuron circuit according to claim 1 , wherein a frequency of the pulse train is related to an operating energy of the memristor.

3. The neuron circuit according to claim 1 , wherein the memristor comprises a magnetoresistive random access memory (MRAM), a variable resistive random access memory (RRAM) or a phase-change memory (PCM).

4. A neuron circuit, comprising:

a memristor, generating a pulse train having an oscillation frequency when an applied voltage exceeds a predetermined threshold; and

a current-voltage converter, connected in parallel to the memristor for receiving an input pulse transmitted by a previous layer network, and driving the memristor to transmit the pulse train to a next layer network when a voltage of the input pulse exceeds the predetermined threshold.

5. The neuron circuit according to claim 4 , wherein a frequency of the pulse train is related to an operating energy of the memristor.

6. The neuron circuit according to claim 5 , wherein the memristor comprises a magnetoresistive random access memory, a variable resistive random access memory or a phase-change memory.

7. An artificial neural network chip, comprising:

a synapse array, comprising n*m synapse elements respectively connecting a plurality of input terminals and a plurality of output terminals, wherein the synapse elements in the same row are connected to the same input terminal, the synapse elements in the same column are connected to the same output terminal, and n and m are positive integers;

a plurality of neuron circuits, an input terminal of each of the neuron circuits is connected to one of the output terminals; and

a control circuit, respectively connected to the input terminals and output terminals of the neuron circuits for adjusting a weight of each of the synapse elements, and monitoring a pulse train transmitted by each of the neuron circuits, wherein

each of the neuron circuits comprises:

a memristor, generating a pulse train having an oscillation frequency when an applied voltage exceeds a predetermined threshold; and

an integrator, connected in parallel to the memristor for receiving and accumulating input pulses input by a previous layer network through the input terminal and passed through the synaptic element, and driving the memristor to transmit the pulse train to a next layer network when a voltage of the accumulated input pulses exceeds the predetermined threshold.

8. The artificial neural network chip according to claim 7 , wherein the integrator is realized by a current-voltage converter, and drives the memristor to transmit the pulse train to the next layer network when a voltage of the input pulses input by the previous layer network through the input terminal and passed through the synaptic element exceeds the predetermined threshold.

9. The artificial neural network chip according to claim 7 , wherein the synapse elements are realized by memristors.

10. The artificial neural network chip according to claim 7 , wherein the memristor comprises a magnetoresistive random access memory, a variable resistive random access memory or a phase-change memory.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Mar 9, 2020
From: HOU, TUO-HUNG; SHEU, SHYH-SHYUAN; WEI, JENG-HUA; LEE, HENG-YUAN; WU, MING-HUNG
To: INDUSTRIAL TECHNOLOGY RESEARCH INSTITUTE
Reel/Frame 052059/0106 →
Priority Claims (1)
TW 108141732 · Nov 18, 2019 · national
Continuity (1)
Related Publication 20210150317A1 · May 20, 2021
Cited By (3)
US 12,536,421 US 12,591,769 US 12,596,921