Unlock instant, AI-driven research and patent intelligence for your innovation.

Computer-implemented method for ascertaining data set for training classifier and/or checking performance of classifier

A data set and computer technology, applied to computer components, calculations, instruments, etc., can solve problems such as system failures, mispredictions, misinterpretations, etc.

Pending Publication Date: 2021-09-14
ROBERT BOSCH GMBH
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Here, the algorithms used for this are not perfect and can cause (more or less serious) mispredictions which, in vehicles operating autonomously, can lead, for example, to misinterpretations and system failures

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Computer-implemented method for ascertaining data set for training classifier and/or checking performance of classifier
  • Computer-implemented method for ascertaining data set for training classifier and/or checking performance of classifier
  • Computer-implemented method for ascertaining data set for training classifier and/or checking performance of classifier

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0046] In a first exemplary embodiment, a computer-implemented method is used to generate a data set of image data for training a classifier. The method is in figure 1 is schematically shown in .

[0047] In step 101, a first data set of camera images is provided. For this example, the application case of highly automated driving is chosen. Furthermore, in this exemplary embodiment, the dataset required for the method is extracted from a database of camera images provided with semantic attributes for highly automated vehicles. Here, semantic attributes represent properties of the environment in which the vehicle operates. In this case, the data set is selected such that it covers all environmental properties present in the database.

[0048] In step 102 a variational autoencoder (VAE) is trained with the dataset from step 101 . For this purpose, the data of the first data set are projected into the latent space of the VAE by means of the encoder part of the VAE. The data...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

A computer-implemented method for training a classifier and / or checking classifier performance, the method comprising the steps of providing a first data set of image data (201); training an encoder (301) and a decoder (303) in such a way that data points of the first data set (201) are mapped into a low-dimensional space (302) by means of the encoder (301) and inversely transformed again by means of the decoder (303); ascertaining a first data representation (401), the first data representation (401) being ascertained by applying the encoder (301) to data points of the first data set (201); ascertaining an additional data representation (402) within a convex hull of the ascertained first data representation (401); determining an inverse transformed data set (403) by applying the decoder (303) to the additional data representation (402); solving a second data set based on the inversely transformed data set; and training a classifier and / or checking the performance of the classifier by means of the second data set.

Description

technical field [0001] The invention relates to a computer-implemented method for obtaining a data set for training a classifier and / or checking the performance of a classifier, a device arranged for carrying out the method, a computer program for carrying out the method and A machine-readable storage medium on which the computer program is stored. Background technique [0002] Computer-implemented machine learning methods, in particular neural networks, are frequently used as part of the recognition of the surroundings of partially-automated, highly-automated or fully-automated robots, in particular autonomous vehicles. The algorithms used for this purpose are not perfect here and can cause (more or less serious) mispredictions which, in the case of autonomous vehicles, can lead, for example, to misinterpretations and system failures. [0003] There is a great need to determine priors under which conditions machine learning methods output wrong predictions. To make sure t...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/62G06N3/04
CPCG06N3/045G06F18/24G06F18/214
Inventor S·拉法特尼亚O·维勒斯S·祖德霍尔特
Owner ROBERT BOSCH GMBH