Machine learning with partial inversion

Patent metadata

Field	Value
Publication number	US-11080588-B2
Application number	US-201715786177-A
Country	US
Kind code	B2
Filing date	Oct 17, 2017
Priority date	Oct 17, 2016
Publication date	Aug 3, 2021
Grant date	Aug 3, 2021

Abstract

An example embodiment may involve a machine learning model representing relationships between a dependent variable and a plurality of n independent variables. The dependent variable may be a function of the n independent variables, where the n independent variables are measurable characteristics of computing devices, and where the dependent variable is a predicted behavior of the computing devices. The embodiment may also involve obtaining a target value of the dependent variable, and separating the n independent variables into n−1 independent variables with fixed values and a particular independent variable with an unfixed value. The embodiment may also involve performing a partial inversion of the function to produce a value of the particular independent variable such that, when the function is applied to the value of the particular independent variable and the n−1 independent variables with fixed values, the dependent variable is within a pre-defined range of the target value.

First claim

What is claimed is:

1. A system comprising:
one or more computing devices disposed within a computational instance of a remote network management platform, wherein the computational instance is dedicated to a remote network managed by the remote network management platform;
a machine learning model representing relationships between a dependent variable and a plurality of n independent variables, wherein the dependent variable is a function of the n independent variables, wherein the n independent variables are measurable characteristics of the one or more computing devices, and wherein the dependent variable is a predicted behavior of the one or more computing devices; and
a computing device including a processor and memory, wherein the computing device is disposed within the remote network management platform, and wherein execution, by the processor, of program instructions stored in the memory causes the computing device to perform operations comprising:
obtaining a target value of the dependent variable,
separating the n independent variables into n−1 independent variables with fixed values and a particular independent variable with an unfixed value,
performing a partial inversion of the function to produce one or more values of the particular independent variable such that, when the function is applied to any of the one or more values of the particular independent variable and the n−1 independent variables with fixed values, the dependent variable is within a pre-defined range of the target value of the dependent variable, wherein performing the partial inversion to produce the one or more values of the particular independent variable comprises:
selecting a series of candidate values of the particular independent variable; and
outputting one or more of the series of candidate values of the particular independent variable that, when applied as inputs to the function along with the fixed value of the n−1 independent variables, produces a respective output value of the dependent variable falling within the pre-defined range of the target value of the dependent variable; and

providing the one or more outputted candidate values of the particular independent variable for display or to another software application executing within the remote network management platform.

2. The system of claim 1 , wherein selecting the series of candidate values for the particular independent variable comprises:
determining that applying the function with a particular candidate value and the n−1 independent variables with fixed values results in a particular output value that is within the pre-defined range of the target value of the dependent variable; and
selecting at least one further candidate value to be closer to the particular candidate value than any other of the candidate values.

3. The system of claim 1 , wherein evaluating each respective candidate value of the series of candidate values comprises:
determining the respective candidate values in accordance with a binary search over a range of the series of candidate values.

4. The system of claim 1 , wherein selecting the series of candidate values for the particular independent variable comprises:
randomly selecting the series of candidate values.

5. The system of claim 1 , wherein performing the partial inversion of the function comprises:
using a non-linear solver software application to determine the one or more values of the particular independent variable.

6. The system of claim 1 , wherein the operations comprise:
monitoring a measurable characteristic of the one or more computing devices corresponding to the particular independent variable.

7. The system of claim 6 , wherein the operations comprise:
determining that the measurable characteristic of the one or more computing devices is within a threshold range of at least one of the one or more values of the particular independent variable; and
generating and sending an alert related to the measurable characteristic.

8. The system of claim 7 , wherein generating and sending the alert related to the measurable characteristic comprises:
transmitting the alert by way of email, voice call, or text message.

9. The system of claim 1 , wherein the measurable characteristics of the one or more computing devices relate to processor utilization of the one or more computing devices, memory utilization of the one or more computing devices, or network traffic received by the one or more computing devices.

10. The system of claim 1 , wherein the predicted behavior of the one or more computing devices is an outage or software application crash related to the one or more computing devices.

11. The system of claim 1 , wherein the operations comprise:
separating the n independent variables into a second set of n−1 independent variables with fixed values and a second particular independent variable with an unfixed value; and
performing a second partial inversion of the function to produce one or more values of the second particular independent variable such that, when the function is applied to any of the one or more values of the second particular independent variable and the second set of n−1 independent variables with fixed values, the dependent variable is within the pre-defined range of the target value of the dependent variable.

12. A method comprising:
determining, by a computing system, a machine learning model representing relationships between a dependent variable and a plurality of n independent variables, wherein the dependent variable is a function of the n independent variables, wherein the n independent variables are measurable characteristics of one or more computing devices of a remote network management platform, and wherein the dependent variable is a predicted behavior of the one or more computing devices;
obtaining, by the computing system, a target value of the dependent variable;
separating, by the computing system, the n independent variables into n−1 independent variables with fixed values and a particular independent variable with an unfixed value;
performing, by the computing system, a partial inversion of the function to produce one or more values of the particular independent variable such that, when the function is applied to any of the one or more values of the particular independent variable and the n−1 independent variables with fixed values, the dependent variable is within a pre-defined range of the target value of the dependent variable, wherein performing the partial inversion of the function comprises:
iteratively selecting a plurality of candidate values for the particular independent variable;
evaluating each respective candidate value of the candidate values by applying the function to the respective candidate value and the n−1 independent variables with fixed values to obtain a respective output value; and
selecting the one or more values of the particular independent variable with corresponding respective output values that are within the pre-defined range of the target value of the dependent variable; and

providing, by the computing system, the one or more values of the particular independent variable for display or to a software application executing within the remote network management platform.

13. The method of claim 12 , wherein selecting the plurality of candidate values for the particular independent variable comprises:
determining that applying the function with a particular candidate value and the n−1 independent variables with fixed values results in a particular output value that is within the pre-defined range of the target value of the dependent variable; and
selecting at least one f

Assignees

• Servicenow Inc

Inventors

• Jayaraman Baskar
• Thakur Aniruddha
• Govindarajan Kannan

Classifications

• H04L41/16 Primary

  using machine learning or artificial intelligence · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N20/00 Primary

  Machine learning · CPC title

• G06F17/11

  for solving equations {, e.g. nonlinear equations, general mathematical optimization problems (optimization specially adapted for a specific administrative, business or logistic context G06Q10/04)} · CPC title

• G06F11/3684

  for test design, e.g. generating new test cases · CPC title