SPIKE NEURAL NETWORK CIRCUIT INCLUDING RADIATION SOURCE

• US 20190311244A1
• Filed: 04/08/2019
• Published: 10/10/2019
• Est. Priority Date: 04/09/2018
• Status: Active Grant

• Current Assignee: Electronics and Telecommunications Research Institute
• Original Assignee: Electronics and Telecommunications Research Institute
• Inventors: OH, Kwang IL, CHOI, Byounggun, KANG, Tae Wook, KIM, Sung Eun, PARK, Seong Mo, LEE, Jae-Jin
• Granted Patent: US 11,403,515 B2
• Time in Patent Office: Days
• CPC Class Codes: G06N 3/049 Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
• CPC Class Codes: G06N 3/063 using electronic means

Abstract

Provided is a spike neural network circuit. The spike neural network circuit includes an axon configured to generate an input spike signal, a synapse including a first transistor for outputting a current according to a weight and a second transistor connected to the first transistor and outputting the current according to an input spike signal, a neuron configured to compare a value according to the current output from the synapse with a reference value and generate an output spike signal based on a comparison result, and a radiation source attached to a substrate on which the synapse is formed, configured to output radiation particles to the synapse, and configured to increase magnitudes of threshold voltages of the first and second transistors of the synapse.

Claims

1. A spike neural network circuit comprising:

an axon configured to generate an input spike signal;

a synapse comprising a first transistor configured to output a current depending on a weight and a second transistor connected to the first transistor and configured to output the current depending on the input spike signal;

a neuron configured to compare a value according to the current output from the synapse with a reference value and generate an output spike signal based on a comparison result; and

a radiation source attached to a substrate in which the synapse is formed and configured to output radiation particles to the synapse and increase magnitudes of threshold voltages of the first and second transistors of the synapse.

2. The spike neural network circuit of claim 1, further comprising a capacitor charged by the current output from the synapse,

wherein the value according to the current is a voltage of the capacitor, and

wherein each of the first and second transistors of the synapse is a PMOS and is formed in an N-well area.

3. The spike neural network circuit of claim 2, wherein electron-hole pairs are generated in a junction area of the substrate and the N-well area by the radiation particles, and holes of the electron-hole pairs are located in the N-well area.

4. The spike neural network circuit of claim 3, wherein the N-well area is connected to a first power supply voltage through an N-type diffusion area, an ohmic contact, and a resistor connected to the ohmic contact, and a voltage of the N-well area is higher than the first power supply voltage by the resistor.

5. The spike neural network circuit of claim 3, wherein the N-well area is connected to a first power supply voltage through an N-type diffusion area and an ohmic contact,

wherein the radiation source overlaps the first and second transistors in a plan view and does not overlap the N-type diffusion area in a plan view, and

wherein a voltage of the N-well area is higher than the first power supply voltage by a resistance of an overlapping area of the N-well area overlapping the N-type diffusion area in a plan view.

6. The spike neural network circuit of claim 1, further comprising a capacitor discharged by the current output from the synapse,

wherein the value according to the current is a voltage of the capacitor, and

wherein each of the first and second transistors of the synapse is an NMOS and is formed in a P-well area.

7. The spike neural network circuit of claim 6, wherein electron-hole pairs are generated in a junction area of an N-well area surrounding the P-well area and the P-well area by the radiation particles and electrons of the electron-hole pairs are located in the P-well area.

8. The spike neural network circuit of claim 7, wherein the P-well area is connected to a second power supply voltage through a P-type diffusion area, a ohmic contact, and a resistor connected to the ohmic contact, and a voltage of the P-well area is lower than the second power supply voltage by the resistor.

9. The spike neural network circuit of claim 7, wherein the P-well area is connected to a second power supply voltage through a P-type diffusion area and an ohmic contact,

wherein the radiation source overlaps the first and second transistors in a plan view and does not overlap the ohmic contact in a plan view, and

wherein a voltage of the P-well area is lower than the second power supply voltage by a resistance of an overlapping area of the P-well area overlapping the P-type diffusion area in a plan view.

10. The spike neural network circuit of claim 1, wherein the weight represents a correlation between the axon and the neuron.

11. The spike neural network circuit of claim 1, wherein the radiation source overlaps the synapse and the radiation source does not overlap the axon and the neuron in a plan view.

12. A spike neural network circuit comprising:

an axon circuit configured to generate a plurality of input spike signals;

a synapse circuit comprising transistors configured to output a plurality of weights depending on the plurality of input spike signals;

a neuron circuit configured to compare a value determined by the weights output from the synapse circuit with a reference value and generate output spike signals based on a comparison result; and

a radiation source attached to a substrate in which the synapse circuit is formed and configured to output radiation particles to the synapse circuit and increase magnitudes of threshold voltages of the transistors of the synapse circuit.

13. The spike neural network circuit of claim 12, wherein the transistors of the synapse circuit are turned on depeendingon the plurality of input spike signals and output currents corresponding to the weights, and

wherein the transistors of the synapse circuit are turned off dependingon the plurality of input spike signals.

14. The spike neural network circuit of claim 13, further comprising a capacitor charged or discharged by the currents corresponding to the weight.

15. The spike neural network circuit of claim 12, wherein the synapse circuit is formed on a first surface of the substrate, and

wherein the radiation source is attached to a second surface facing the first surface of the substrate and overlaps the synapse circuit in a plan view, and

wherein the radiation source outputs the radiation particles in a direction from the second surface to the first surface.