# US 2021/0241174 A1 · Computer implemented machine learning system and a method for operating the machine learning system for determining a time series

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

- Number: US 2021/0241174 A1
- Document ID: US20210241174A1
- Status: Published application
- Kind: A1
- Application number: 17/138,685
- Filed: 2020-12-30
- Published: 2021-08-05
- Earliest claimed: 2020-02-03
- Assignees: Robert Bosch GmbH
- Applicants: Robert Bosch GmbH
- Inventors: Martin Schiegg; Muhammad Bilal Zafar
- Web page: https://patentgrep.com/patent/US20210241174A1

## CPC

- G05B 17/02: electric
- G06N 20/00: Machine learning
- G06F 17/18: for evaluating statistical data {, e.g. average values, frequency distributions, probability functions, regression analysis (forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04)}
- G06N 3/044: Recurrent networks, e.g. Hopfield networks
- G06N 3/0442: characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
- G06N 3/045: Combinations of networks
- G06N 3/047: Probabilistic or stochastic networks
- G06N 3/0475: Generative networks
- G06N 3/088: Non-supervised learning, e.g. competitive learning
- G06N 3/09: Supervised learning
- G06N 3/094: Adversarial learning
- G05B: CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS (systems for controlling or regulating non-electric variables G05D; systems for regulating electric or magnetic variables G05F; control devices or systems insofar as characterised by mechanical features only G05G)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A machine learning system and method of operating a machine learning system for determining a time series, comprising providing an input for a first in particular generative model depending on a probabilistic variable, determining an output of the first model in response to the input for the first model, the output of the first model characterizing the time series. The first model comprises a first layer that is trained to map input for the first model determined depending on the probabilistic variable to output characterizing intermediate data, and a second layer that is trained to map the intermediate data to the time series depending on an output of a third layer of the first model. The output of the third layer characterizes a physical constraint to a machine state. Values of the time series or of the intermediate data are constrained by the output of the third layer.

## Claims

### Claim 1

A computer implemented method of operating a machine learning system for determining a time series, the method comprising the following steps:
providing an input for a first generative model depending on a probabilistic variable, the probabilistic variable being noise; and
determining an output of the first model in response to the input for the first model, wherein the output of the first model characterizes the time series, the first model including a first layer that is trained to map the input for the first model determined depending on the probabilistic variable to output characterizing intermediate data, the first model further including a second layer that is trained to map the intermediate data to the time series depending on an output of a third layer of the first model, the output of the third layer characterizing a physical constraint to a machine state, and wherein values of the time series or the intermediate data are constrained by the output of the third layer.

### Claim 2 (depends on 1)

The method according to claim 1, further comprising:
providing the input for the first model depending on a conditional variable;
wherein the first layer and/or the second layer are trained to map the input for the first model determined depending on the conditional variable and the probabilistic variable to output characterizing the intermediate data.

### Claim 3 (depends on 2)

The method according to claim 2, further comprising:
providing an input for the third layer depending on at least one state of a machine and/or the conditional variable; and
determining an output of the third layer in response to the input for the third layer.

### Claim 4 (depends on 3)

The method according to claim 3, wherein the physical constraint for a time step is determined depending on the time series in a previous time step, and/or a force applied in the environment, and/or a force applied by or to the machine in an environment.

### Claim 5 (depends on 2)

The method according to claim 2, wherein the conditional variable includes at least one of: an environment variable or an internal machine state.

### Claim 6 (depends on 2)

The method according to claim 2, further comprising:
providing an input for a second model depending on the conditional variable and/or the time series; and
determining an output of the second model in response to the input for the second model;
wherein the output of the second model characterizes a score indicating a degree of reliance for the time series, wherein the second model is trained to map the input for the second model determined depending on the conditional variable and the time series to output of the second model characterizing the score indicating the degree of reliance for the time series.

### Claim 7 (depends on 2)

The method according to claim 2, further comprising:
providing the conditional variable as a continuous or discrete first series of values over time within a time period and/or providing the probabilistic variable as a continuous or discrete second series of values over time within the time period;
determining by the first model, a continuous or discrete third series of values for the time series depending on the values of the first series and the second series; and
determining by the second model the score depending on the values of the first series and the third series;
wherein the first model is a first Recurrent Neural network and the second model is a second Recurrent Neural network.

### Claim 8 (depends on 1)

The method according to claim 1, wherein the time series characterizes a machine state series and the intermediate data characterizes a machine state change series.

### Claim 9 (depends on 3)

The method according to claim 3, wherein the time series characterizes a position over time of the machine in an environment and the intermediate data characterizes a change of speed and/or angle over time of the machine in the environment.

### Claim 10 (depends on 8)

The method according to claim 8, wherein the machine state characterizes a velocity and the intermediate data characterizes an unconstrained acceleration or the machine state characterizes a gear of a transmission and the intermediate data characterizes an unconstrained gear.

### Claim 11 (depends on 3)

The method according to claim 3, wherein the time series is determined depending on output of the first model and the second model in response to training data defining input data for the first model and the second model, wherein the output of the second model characterizing the score indicating the degree of reliance for the time series is determined, and wherein at least one parameter of the first model and/or the second model is determined depending on the score.

### Claim 12 (depends on 11)

The method according to claim 11, further comprising:
providing input data including the time series, the conditional variable, the intermediate data, and at least one state of the machine.

### Claim 13

A machine learning system, comprising:
a first model; and
a second model;
wherein the machine learning system is configured to determine a time series, machine learning system configured to:
providide an input for the first model depending on a probabilistic variable, the probabilistic variable being noise; and
determine an output of the first model in response to the input for the first model, wherein the output of the first model characterizes the time series, the first model including a first layer that is trained to map the input for the first model determined depending on the probabilistic variable to output characterizing intermediate data, the first model further including a second layer that is trained to map the intermediate data to the time series depending on an output of a third layer of the first model, the output of the third layer characterizing a physical constraint to a machine state, and wherein values of the time series or the intermediate data are constrained by the output of the third layer.

### Claim 14

A non-transitory computer readable storage medium on which is stored a computer program for operating a machine learning system for determining a time series, the computer program, when executed by computer, causes the computer to perform the following steps:
providing an input for a first generative model depending on a probabilistic variable, the probabilistic variable being noise; and
determining an output of the first model in response to the input for the first model, wherein the output of the first model characterizes the time series, the first model including a first layer that is trained to map the input for the first model determined depending on the probabilistic variable to output characterizing intermediate data, the first model further including a second layer that is trained to map the intermediate data to the time series depending on an output of a third layer of the first model, the output of the third layer characterizing a physical constraint to a machine state, and wherein values of the time series or the intermediate data are constrained by the output of the third layer.