US 2021/0241092 A1Published application

Knowledge persistent and structurally dynamic neural network

AssigneeBank of America Corporation

InventorsGarrett Thomas Botkin, Matthew Bruce Murray

Life of this patent

Estimating the term…

Loading patent…

Abstract

A method includes receiving a request to solve a problem defined by input information and applying a neural network to generate an answer to the problem. The neural network includes an input level, a manager level including a first manager, a worker level including first and second workers, and an output level. Applying the neural network includes implementing the input level to provide a piece of input information to the first manager; implementing the first manager to delegate portions of the piece of information to the first and second workers; implementing the first worker to operate on its portion of information to generate a first output; implementing the second worker to operate on its portion of information to generate a second output; and implementing the output level to generate the answer to the problem, using the first and second outputs. The method also includes transmitting a response comprising the answer.

Claims20

Dependent claims

An apparatus comprising:

a memory configured to store non-transitory computer readable instructions corresponding to a neural network, the neural network comprising:
an input level configured, when implemented by a hardware processor, to receive input information;
a manager level coupled to the input level, the manager level comprising a first manager configured, when implemented by the hardware processor, to operate on a first piece of the input information;
a worker level comprising a first worker assigned to the first manager and a second worker assigned to the first manager, wherein the first manager is further configured, when implemented by the hardware processor, to:
delegate a first portion of the first piece of the input information to the first worker, wherein the first worker is configured, when implemented by the hardware processor, to operate on the first portion of the first piece of the input information, using a first set of adjustable parameters, to generate a first output;
and
delegate a second portion of the first piece of the input information to the second worker, wherein the second worker is configured, when implemented by the hardware processor, to operate on the second portion of the first piece of input information, using a second set of adjustable parameters, to generate a second output;
and
an output level coupled to the worker level, the output level configured, when implemented by the hardware processor, to generate an output, based at least in part on the first output from the first worker and the second output from the second worker;
and
the hardware processor communicatively coupled to the memory, the hardware processor configured to:
receive a request to solve a problem defined by a set of information;
apply the neural network to the set of information to generate an answer to the problem, wherein the set of information corresponds to the input information received by the input level and the answer to the problem corresponds to the output from the neural network;
and
in response to applying the neural network, transmit a response comprising the answer to the problem.

12 more elements

A method comprising:

receiving a request to solve a problem defined by input information;
applying a neural network to the input information to generate an answer to the problem, the neural network comprising non-transitory computer readable instructions corresponding to:
an input level;
a manager level comprising a first manager;
a worker level comprising a first worker assigned to the first manager and a second worker assigned to the first manager;
and
an output level, wherein applying the neural network comprises:
implementing the input level to receive the input information and to provide a first piece of the input information to the first manager;
implementing the first manager to:
delegate a first portion of the first piece of the input information to the first worker;
and
delegate a second portion of the first piece of the input information to the second worker;
implementing the first worker to operate on the first portion of the first piece of the input information, using a first set of adjustable parameters, to generate a first output;
implementing the second worker to operate on the second portion of the first piece of the input information, using a second set of adjustable parameters, to generate a second output;
and
implementing the output level to generate an output, based at least in part on the first output from the first worker and the second output from the second worker, wherein the output comprises the answer to the problem;
and
in response to applying the neural network, transmitting a response comprising the answer to the problem.

15 more elements

Dependencies are inferred from the claim text.

Facts

Application: 16/781,656Filed 4 Feb 2020
Published: 5 Aug 2021
Term ends: Not estimatedThe source records don’t support an estimate.

Classification8 codes

G06N 3/084
Learning methods: backpropagation, e.g. using gradient descent
G06N 3/04
Neural networks: architecture, e.g. interconnection topology
G06N 3/045
Architecture, e.g. interconnection topology: combinations of networks
G06N 3/0464
Architecture, e.g. interconnection topology: convolutional networks [CNN, ConvNet]

Family

US 12,020,163 B2Related patent
US 2021/0241092 A1This patent

From the US filing history. Foreign family members aren’t listed.

Timeline

Citations

Open elsewhere

Building an agent or pulling many patents? Use the API or MCP server instead of scraping pages. Get an API key