# US 2021/0342132 A1 · Multi Objective Optimization of Applications

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

- Number: US 2021/0342132 A1
- Document ID: US20210342132A1
- Status: Published application
- Kind: A1
- Application number: 16/863,001
- Filed: 2020-04-30
- Published: 2021-11-04
- Earliest claimed: 2020-04-30
- Assignees: International Business Machines Corporation
- Applicants: International Business Machines Corporation
- Inventors: Srinivasan Parthasarathy; Fabio A. Oliveira; Sushma Ravichandran
- Web page: https://patentgrep.com/patent/US20210342132A1

## CPC

- G06F 8/443: {Optimisation}
- G06F 11/3006: {where the computing system is distributed, e.g. networked systems, clusters, multiprocessor systems (multiprogramming arrangements G06F9/46; allocation of resources G06F9/50)}
- G06F 11/302: {where the computing system component is a software system}
- G06F 11/3409: {for performance assessment}
- G06N 3/09: Supervised learning
- G06N 5/04: Inference or reasoning models
- G06N 20/00: Machine learning
- G06N 3/08: Learning methods
- G06N 20/10: using kernel methods, e.g. support vector machines [SVM]
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A system, computer program product, and method are provided for orchestrating a multi objective optimization of an application. A set of two or more key performance indicators (KPIs) and one or more parameters associated with the application are received. A machine learning (ML) based surrogate function learning model in combination with an acquisition function is leveraged to conduct one or more adaptive trials. Each trial consists of a specific configuration of the one or more parameters. A pareto surface of the KPIs of the application is computed based on the observations of KPI values from each adaptive trial. The pareto surface is explored and an optimal operating point is selected for the application. The application is then executed at the selected operating point.

## Claims

### Claim 1

A computer system comprising:
a processing unit operatively coupled to memory;
an artificial intelligence (AI) platform operatively coupled to the processing unit, the AI platform configured to orchestrate a multi-objective optimization of a cloud application, the AI platform comprising:
an input manager to receive a set of two or more key performance indicators (KPIs) associated with the application, each KPI in the set of KPIs associated with a KPI value;
a trial manager to leverage a machine learning (ML) based surrogate function learning model in combination with an acquisition function to conduct an adaptive trial execution of the application with one or more parameters;
the trial manager to observe output from each trial execution and to compute a pareto surface of the observed output;
an optimization manager to explore the computed pareto surface and select an optimal application operating point corresponding to configuration of the one or more parameters; and
the optimization manager to execute the application at the selected optimal operating point.

### Claim 2 (depends on 1)

The computer system of claim 1, wherein the ML based surrogate function learning model maps each configuration of the one or more parameters to the two or more KPI values.

### Claim 3 (depends on 2)

The computer system of claim 2, wherein the ML based surrogate function learning model is implemented using a neural network, Gaussian Progress regression, kernel regression techniques, or a combination thereof.

### Claim 4 (depends on 1)

The computer system of claim 1, wherein the acquisition function selects a second configuration of one or more second parameters for a second trial execution of the application.

### Claim 5 (depends on 4)

The computer system of claim 4, where the acquisition function is implemented as an expected improvement function, maximum probability of improvement function, upper confidence bound function, or a combination thereof.

### Claim 6 (depends on 1)

The computer system of claim 1, wherein the set of two or more KPI values comprises latency, error rate, throughput, or a combination thereof, and wherein the one or more parameters associated with one or more topological constraints comprises timeout, retry, replication count, or a combination thereof.

### Claim 7 (depends on 6)

The computer system of claim 6, wherein the topological constraints include a hierarchy of micro-services and the one or more parameters are set with respect to the hierarchy, including ensuring that the one or more parameters guarantees a query from a parent micro-service in the hierarchy is active as long as one or more queries from a child micro-service is active.

### Claim 8

A computer program product to orchestrate a multi objective optimization of cloud applications, the computer program product comprising a computer readable storage medium having program code embodied therewith, the program code executable by a processor to:
receive a set of two or more key performance indicators (KPIs) associated with an application, each KPI in the set of KPIs associated with a KPI value;
leverage a machine learning (ML) based surrogate function learning model in combination with an acquisition function to conduct an adaptive trial execution of the application with the one or more received parameters;
observe output from each trial execution and compute a pareto surface of the observed output;
explore the computed pareto surface and select an optimal application operating point corresponding to configuration of the one or more parameters; and
execute the application at the selected optimal operating point.

### Claim 9 (depends on 8)

The computer program product of claim 8, wherein the ML based surrogate function learning model maps each configuration of the one or more parameters to the two or more KPI values.

### Claim 10 (depends on 9)

The computer program product of claim 9, wherein the ML based surrogate function learning model is implemented using a neural network, Gaussian Progress regression, kernel regression techniques, or learning combination thereof.

### Claim 11 (depends on 8)

The computer program product of claim 8, wherein the acquisition function selects a second configuration of one or more second parameters for a second trial execution of the application.

### Claim 12 (depends on 11)

The computer program product of claim 11, where the acquisition function is implemented as an expected improvement function, maximum probability of improvement function, upper confidence bound function, or a combination thereof.

### Claim 13 (depends on 8)

The computer program product of claim 8, wherein the set of two or more KPI values comprises latency, error rate, throughput, or a combination thereof, and wherein the one or more tunable parameters associated with one or more topological constraints comprises timeout, retry, replication count, or a combination thereof.

### Claim 14 (depends on 13)

The computer program product of claim 13, wherein the topological constraints include a hierarchy of micro-services and the one or more parameters are set with respect to the hierarchy, including ensuring that the one or more parameters guarantees a query from a parent micro-service in the hierarchy is active as long as one or more queries from a child micro-service is active.

### Claim 15

A computer implemented method comprising:
receiving a set of two or more key performance indicators (KPIs) associated with an application, each KPI in the set of KPIs associated with a KPI value;
leveraging a machine learning (ML) based surrogate function learning model in combination with an acquisition function to conduct an adaptive trial execution of the application with one or more received parameters;
observing output from each trial execution and computing a pareto surface of the observed output;
exploring the computed pareto surface and selecting an optimal application operating point corresponding to configuration of the one or more parameters; and
executing the application at the selected optimal operating point.

### Claim 16 (depends on 15)

The method of claim 15, wherein the ML based surrogate function learning model maps each configuration of the one or more parameters to the two or more KPI values.

### Claim 17 (depends on 16)

The method of claim 16, wherein the ML based surrogate function learning model is implemented using a neural network, Gaussian Progress regression, kernel regression techniques, or a combination thereof.

### Claim 18 (depends on 15)

The method of claim 15, wherein the acquisition function selects a second configuration of one or more second parameters for a second trial execution of the application.

### Claim 19 (depends on 18)

The method of claim 18, where the acquisition function is implemented as an expected improvement function, maximum probability of improvement function, upper confidence bound function, or a combination thereof.

### Claim 20 (depends on 15)

The method of claim 15, wherein the set of two or more KPI values comprises latency, error rate, throughput, or a combination thereof, and wherein the one or more parameters associated with one or more topological constraints comprises timeout, retry, replication count, or a combination thereof.
