# US 2021/0174185 A1 · Output circuits for an analog neural memory system for deep learning neural network

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2021/0174185 A1
- Document ID: US20210174185A1
- Status: Published application
- Kind: A1
- Application number: 17/181,656
- Filed: 2021-02-22
- Published: 2021-06-10
- Earliest claimed: 2018-08-27
- Assignees: Silicon Storage Technology, Inc.
- Applicants: Silicon Storage Technology, Inc.
- Inventors: Hieu Van Tran; Vipin Tiwari; Mark Reiten; Nhan Do
- Web page: https://patentgrep.com/patent/US20210174185A1

## CPC

- G06N 3/065: Analogue means
- G06F 3/0688: {Non-volatile semiconductor memory arrays}
- G06F 17/16: Matrix or vector computation {, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization (matrix transposition G06F7/78)}
- G06N 3/044: Recurrent networks, e.g. Hopfield networks
- G06N 3/0442: characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/08: Learning methods
- G11C 27/02: Sample-and-hold arrangements (G11C27/04 takes precedence)
- G06N 3/045: Combinations of networks
- H10B 41/30: characterised by the memory core region
- H10D 30/6892: {having at least one additional gate other than the floating gate and the control gate, e.g. program gate, erase gate or select gate}
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G11C: STATIC STORES (semiconductor memory devices H10B)
- H10B: ELECTRONIC MEMORY DEVICES
- H10D: INORGANIC ELECTRIC SEMICONDUCTOR DEVICES

## Abstract

Numerous output circuits are disclosed for an analog neural memory system for a deep learning neural network. In one embodiment, an adaptable neuron circuit receives output current from a neuron and converts it into a voltage. In another embodiment, a current sample and hold circuit samples an input current and generates an output current. In another embodiment, a voltage sample and hold circuit samples an input voltage and generates an output voltage.

## Claims

### Claim 1

An adaptable neuron circuit coupled to a neuron in a neuromorphic memory array, the adaptable neuron circuit comprising:
a sample-and-hold circuit for sampling, in a first mode, a neuron current and storing a voltage on a gate of a transistor, and in a second mode, generating a mirrored current of the neuron current; and
a variable resistor for drawing the mirrored current during the second mode and generating an output voltage based on the mirrored current, the output voltage indicating a value stored in the neuron.

### Claim 2 (depends on 1)

The adaptable neuron circuit of claim 1, wherein the transistor is a PMOS transistor.

### Claim 3 (depends on 1)

The adaptable neuron circuit of claim 1, wherein the sample-and-hold circuit comprises a first switch, a second switch, and a third switch, and wherein in the first mode, the first switch is closed and couples the gate of the transistor to the neuron, the second switch is closed and couples the neuron to a source of the transistor, and the third switch is open.

### Claim 4 (depends on 3)

The adaptable neuron circuit of claim 3, wherein in the second mode, the first switch is open, the second switch is open, and the third switch is closed and couples the source of the transistor to the variable resistor.

### Claim 5

A current sample and hold circuit for a neuron output for a neural network, comprising
an input transistor comprising a first terminal, a second terminal coupled to ground, and a gate;
a capacitor comprising a first terminal and a second terminal;
an output transistor comprising a first terminal providing an output current, a second terminal coupled to ground, and a gate;
a first switch; and
a second switch;
wherein in a first mode, the first switch is closed and couples an input current to the first terminal of the input transistor and the gate of the input transistor and the second switch is closed and couples the first terminal of the input transistor to the first terminal of the capacitor and the gate of the output transistor, and in a second mode, the first switch is open and the second switch is open and the capacitor discharges into the gate of the output transistor.

### Claim 6 (depends on 5)

The current sample and hold circuit of claim 5, wherein the second terminal of the capacitor is coupled to ground.

### Claim 7 (depends on 5)

The current sample and hold circuit of claim 5, wherein the input transistor is an NMOS transistor.

### Claim 8 (depends on 7)

The current sample and hold circuit of claim 7, wherein the output transistor is an NMOS transistor

### Claim 9 (depends on 5)

The current sample and hold circuit of claim 5, wherein the input current is provided by a neuron in a neuromorphic memory array.

### Claim 10

A sample and hold circuit for a neuron output for a neural network, comprising:
a switch;
a capacitor comprising a first terminal and a second terminal coupled to ground; and
an op amp comprising a first input terminal coupled to the first terminal of the capacitor and a second terminal coupled to an output of the op amp, the output of the op amp providing an output voltage;
wherein a first mode, the switch is closed and couples an input voltage to the first terminal of the capacitor, and in a second mode, the switch is open and the capacitor discharges into the op amp.
