# US 2019/0318238 A1 · Artificial intelligence optimized telecommunications systems

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

- Number: US 2019/0318238 A1
- Document ID: US20190318238A1
- Status: Published application
- Kind: A1
- Application number: 15/952,887
- Filed: 2018-04-13
- Published: 2019-10-17
- Earliest claimed: 2018-04-13
- Assignees: Adtran, Inc.
- Applicants: ADTRAN, Inc.
- Inventors: Armand Nokbak Nyembe; Sheila Knight; Michael Arnold; Ramya Raghavendra; Jeremy Lyon; Venkata Mallikarjunarao Kosuri; Zack Whaley
- Web page: https://patentgrep.com/patent/US20190318238A1

## CPC

- G06N 3/092: Reinforcement learning
- G06N 5/043: Distributed expert systems; Blackboards
- G06F 40/30: Semantic analysis
- G06N 3/0499: Feedforward networks
- G10L 15/16: using artificial neural networks
- G10L 15/1815: {Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning}
- G10L 15/22: Procedures used during a speech recognition process, e.g. man-machine dialogue
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING

## Abstract

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for an artificial intelligence system. In one aspect, a system includes multiple artificial intelligence skill agents that have each been trained to perform different actions in a telecommunications system. The system also includes an orchestrator agent that interacts with each of the artificial intelligence skill agents and coordinates which of the artificial intelligence agents performs actions in response to user inputs. The orchestrator agent receives a user input and determines an intent expressed by the user input. The orchestrator agent transmits an instruction to an artificial intelligence skill agent to perform an action that provides a response to the intent. In response to receiving the instruction from the orchestrator agent, the artificial intelligence skill agent performs the action when the artificial intelligence skill agent is capable of carrying out the action.

## Claims

### Claim 1

An artificial intelligence system, comprising:
a plurality of artificial intelligence skill agents, including one or more processors, that have each been trained to perform different actions in a telecommunications system; and
an orchestrator agent that interacts with each of the plurality of artificial intelligence skill agents and coordinates which of the plurality of the artificial intelligence agents performs actions in response to user input, wherein the orchestrator agent includes one or more processors and a computer readable medium storing instructions that upon execution cause the orchestrator agent to perform operations comprising:
receiving a user interface input that was input through a user interface;
determining an intent expressed by the user interface input;
identifying, from among the plurality of artificial intelligence skills agents, a given artificial intelligence skill agent that has been trained to perform an action that provides a response to the intent expressed by the user interface input; and
transmitting an instruction to the given artificial intelligence skill agent to perform the action;
wherein, in response to receiving the instruction from the orchestrator agent, the given artificial intelligence skill agent processes the instruction, including:
performing the action when the given artificial intelligence skill agent is capable of carrying out the action; and
responding to the orchestrator agent with negative feedback when the given artificial intelligence skill agent is incapable of carrying out the action.

### Claim 2 (depends on 1)

The system of claim 1, wherein determining an intent expressed by the user interface input comprises:
providing an input based on the user interface input to an intent mapping neural network, wherein the intent mapping neural network has a plurality of intent mapping neural network parameters, and the intent mapping neural network is configured to process the input in accordance with the intent mapping neural network parameters to generate an intent mapping output; and
determining an intent based on the intent mapping output.

### Claim 3 (depends on 2)

The system of claim 2, wherein:
the orchestrator agent receives a reward based on progress towards results that respond to the user interface input, and
the intent mapping neural network is trained using reinforcement learning techniques based on the received reward, comprising adjusting current values of the intent mapping neural network parameters.

### Claim 4 (depends on 1)

The system of claim 1, wherein determining an intent expressed by the user interface input comprises processing the user interface input using natural language processing techniques.

### Claim 5 (depends on 1)

The system of claim 1, wherein when the given artificial intelligence skill agent is incapable of carrying out the action, the given artificial intelligence skill agent performs operations comprising:
determining a new action that provides a response to the intent expressed by the user interface input, wherein the new action is different than the action;
performing the new action when the given artificial intelligence skill agent is capable of carrying out the new action.

### Claim 6 (depends on 1)

The system of claim 1, wherein the artificial intelligence system performs operations comprising determining whether a multi-tiered access control condition has been met prior to the given artificial intelligence skill agent performing the action, including:
determining whether a network operator that submitted the user input has permission to access the given artificial intelligence skill agent; and
determining whether the given artificial intelligence skill agent has permission to access data required to perform the action.

### Claim 7 (depends on 6)

The system of claim 6, wherein the given artificial intelligence skill agent performing the action is conditioned on the multi-tiered access control condition being met.

### Claim 8 (depends on 7)

The system of claim 7, wherein in response to the determination of whether the multi-tiered access control condition has been met:
the given artificial intelligence skill agent performs the action when the multi-tiered access control condition has been met; and
the artificial intelligence system responds to the user input without the given artificial intelligence skill agent performing the action when the multi-tiered access control condition has not been met.

### Claim 9

A computer-implemented method comprising:
performing, by an orchestrator agent that interacts with each of a plurality of artificial intelligence skill agents and coordinates which of the plurality of the artificial intelligence agents performs actions in response to user inputs, operations comprising:
receiving a user interface input that was input through a user interface;
determining an intent expressed by the user interface input;
identifying, from among a plurality of artificial intelligence skills agents, a given artificial intelligence skill agent that has been trained to perform an action that provides a response to the intent expressed by the user interface input;
transmitting an instruction to the given artificial intelligence skill agent to perform the action; and
performing, by the given artificial intelligence skill agent, and in response to receiving the instruction from the orchestrator agent, operations comprising:
performing the action when the given artificial intelligence skill agent is capable of carrying out the action; and
responding to the orchestrator agent with negative feedback when the given artificial intelligence skill agent is incapable of carrying out the action.

### Claim 10 (depends on 9)

The method of claim 9, wherein determining an intent expressed by the user interface input comprises:
providing an input based on the user interface input to an intent mapping neural network, wherein the intent mapping neural network has a plurality of intent mapping neural network parameters, and the intent mapping neural network is configured to process the input in accordance with the intent mapping neural network parameters to generate an intent mapping output; and
determining an intent based on the intent mapping output.

### Claim 11 (depends on 10)

The method of claim 10, wherein:
the orchestrator agent receives a reward based on progress towards results that respond to the user interface input, and
the intent mapping neural network is trained using reinforcement learning techniques based on the received reward, comprising adjusting current values of the intent mapping neural network parameters.

### Claim 12 (depends on 9)

The method of claim 9, wherein determining an intent expressed by the user interface input comprises processing the user interface input using natural language processing techniques.

### Claim 13 (depends on 9)

The method of claim 9, wherein when the given artificial intelligence skill agent is incapable of carrying out the action, the given artificial intelligence skill agent performs operations comprising:
determining a new action that provides a response to the intent expressed by the user interface input, wherein the new action is different than the action;
performing the new action when the given artificial intelligence skill agent is capable of carrying out the new action.

### Claim 14 (depends on 9)

The method of claim 9, further comprising determining whether a multi-tiered access control condition has been met prior to the given artificial intelligence skill agent performing the action, including:
determining whether a network operator that submitted the user input has permission to access the given artificial intelligence skill agent; and
determining whether the given artificial intelligence skill agent has permission to access data required to perform the action.

### Claim 15 (depends on 14)

The method of claim 14, wherein the given artificial intelligence skill agent performing the action is conditioned on the multi-tiered access control condition being met.

### Claim 16 (depends on 15)

The method of claim 15, further comprising:
in response to the determination of whether the multi-tiered access control condition has been met:
performing the action, by the given artificial intelligence skill agent, when the multi-tiered access control condition has been met; and
responding to the user input without the given artificial intelligence skill agent performing the action when the multi-tiered access control condition has not been met.

### Claim 17

A non-transitory computer storage medium storing instructions executable by a data processing apparatus and that upon such execution cause the data processing apparatus to perform operations comprising:
receiving a user interface input that was input through a user interface;
determining an intent expressed by the user interface input;
identifying, from among a plurality of artificial intelligence skills agents, a given artificial intelligence skill agent that has been trained to perform an action that provides a response to the intent expressed by the user interface input;
transmitting an instruction to the given artificial intelligence skill agent to perform the action;
wherein, in response to receiving the instruction, the given artificial intelligence agent performs operations comprising:
performing the action when the given artificial intelligence skill agent is capable of carrying out the action; and
responding with negative feedback when the given artificial intelligence skill agent is incapable of carrying out the action.

### Claim 18 (depends on 17)

The non-transitory computer storage medium of claim 17, wherein determining an intent expressed by the user interface input comprises:
providing an input based on the user interface input to an intent mapping neural network, wherein the intent mapping neural network has a plurality of intent mapping neural network parameters, and the intent mapping neural network is configured to process the input in accordance with the intent mapping neural network parameters to generate an intent mapping output; and
determining an intent based on the intent mapping output.

### Claim 19 (depends on 18)

The non-transitory computer storage medium of claim 18, wherein:
a reward is received based on progress towards results that respond to the user interface input, and
the intent mapping neural network is trained using reinforcement learning techniques based on the received reward, comprising adjusting current values of the intent mapping neural network parameters.

### Claim 20 (depends on 17)

The non-transitory computer storage medium of claim 17, wherein determining an intent expressed by the user interface input comprises processing the user interface input using natural language processing techniques.
