# US 5,710,869 A · Daisy chain circuit for serial connection of neuron circuits

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

- Number: US 5,710,869 A
- Document ID: US5710869A
- Status: Granted
- Kind: A
- Application number: 08/485,337
- Filed: 1995-06-07
- Granted: 1998-01-20
- Earliest claimed: 1994-07-28
- Assignees: International Business Machines Corporation
- Inventors: Catherine Godefroy; Andre Steimle; Pascal Tannhof; Guy Paillet
- Web page: https://patentgrep.com/patent/US5710869A

## CPC

- G06N 3/063: using electronic means
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Each daisy chain circuit is serially connected to the two adjacent neuron circuits, so that all the neuron circuits form a chain. The daisy chain circuit distinguishes between the two possible states of the neuron circuit (engaged or free) and identifies the first free "or ready to learn" neuron circuit in the chain, based on the respective values of the input (DCI) and output (DCO) signals of the daisy chain circuit. The ready to learn neuron circuit is the only neuron circuit of the neural network having daisy chain input and output signals complementary to each other. The daisy chain circuit includes a 1-bit register (601) controlled by a store enable signal (ST) which is active at initialization or, during the learning phase when a new neuron circuit is engaged. At initialization, all the Daisy registers of the chain are forced to a first logic value. The DCI input of the first daisy chain circuit in the chain is connected to a second logic value, such that after initialization, it is the ready to learn neuron circuit. In the learning phase, the ready to learn neuron's 1-bit daisy register contents are set to the second logic value by the store enable signal, it is said "engaged". As neurons are engaged, each subsequent neuron circuit in the chain then becomes the next ready to learn neuron circuit.

## Claims

### Claim 1

In each neuron circuit in a neural network comprised of a plurality of neuron circuits, a daisy chain circuit for selecting between two neuron states (free or engaged), said daisy chain circuit including:
means for receiving an input signal (DCI);
means for generating an output signal (DCO) said output signal being available as a subsequent neuron's daisy chain circuit input, such that the neuron circuits of the neural network may be linked together by their daisy chain circuits to form a chain;
register means (601) for storing said input signal under the control of a store enable signal (ST), said store enable signal being active during a learning phase of a neuron, the logic value of said input signal being loaded into said register; and
reset means connected to said register for forcing a first logic value therein at initialization, and
said receiving means of the first daisy chain circuit in a chain of said neuron circuits being connected to a first potential (VH) such that the first input signal of the chain is equal to a second logic value, wherein
said reset means comprises an AND gate (602), the output of said AND gate (602) being connected to the input of said register means and said input signal generated by the preceding daisy chain circuit being ANDed with a reset signal.

### Claim 2 (depends on 1)

The daisy chain circuit of claim 1 wherein the state of the neuron circuit (NC) is according to the following table:
TABLE ______________________________________ DCI DCO NC state ______________________________________ L L free H L first free (ready to learn) H H engaged L H reserved. ______________________________________

### Claim 3 (depends on 1)

The daisy chain circuit of claim 1 further including:
a two-way OR gate (603), the output of said register means being connected to one input of said two way OR gate, a test signal (ALL) being connected to the other input and said two way OR gate's output being said Daisy Chain output DCO; and
said test signal selectively forcing said Daisy Chain output to a second logic value, thereby forcing the neuron circuit into the engaged state.

### Claim 4 (depends on 1)

The daisy chain circuit of claim 1 further including a two-way XOR gate (605), said input signal (DCI) being connected to one input of said two-way XOR gate, said two-way XOR gate's other input being connected to the output of the daisy chain circuit, the output of said two-way XOR gate being a ready to store signal (RS).

### Claim 5 (depends on 5)

The daisy chain circuit of claim 5 wherein in each neuron circuit said ready to store signal is applied to the neuron circuit's weight memory during the recognition phase, said weight memory having the components of the input vector to the neural network written only into the ready to learn neuron circuit's weight memory.

### Claim 6 (depends on 1)

The daisy chain circuit of claim 1 further including a two-way AND gate (606), said Daisy Chain output (DCO) being a fire input of said two-way AND gate and a neuron circuit selection signal (NS) being a second input of said two way AND gate, the output of said two way AND gate being a control signal (CO), said control signal selecting only engaged neuron circuits in the neural network.