NUMERICALLY MORE STABLE TRAINING OF A NEURAL NETWORK ON TRAINING MEASURED DATA PROVIDED AS A POINT CLOUD

Abstract

A method for monitored training of a neural network. In the method, training examples including training measured data and associated training output variables are provided; a spatial region, which contains at least a part of the locations indicated by the training measured data of a training example, is subdivided into a grid made up of adjoining cells; for each cell, values of the measured variables contained in the training measured data for all locations in this cell are aggregated to form values of the measured variables which relate to this cell; these aggregated values of the measured variables are mapped by the neural network on one or multiple output variables; deviations of these output variables from the training output variables are assessed using a predefined cost function; parameters of the neural network are optimized.

Claims

1. A method for monitored training of a neural network, which maps measured data on one or multiple output variables, the measured data assigning values of one or multiple measured variables to locations in the two-dimensional or three-dimensional space, the method comprising the following steps: providing training examples made up of training measured data and associated training output variables; subdividing a spatial region, which contains at least a portion of the locations indicated by the training measured data of a training example, into a grid made up of adjoining cells; aggregating, for each cell of the adjoining cells, the values of the measured variables contained in the training measured data of the training example for all locations in the cell, to form values of the measured variables which relate to the cell; mapping the aggregated values of the measured variables, by the neural network, on one or multiple output variables; assessing deviations of the output variables from the training output variables using a predefined cost function, which is composed in weighted form of contributions of individual cells of the grid, the weight of each contribution being a function of an occupancy of the corresponding cell with locations contained in the training measured data of the training example; and optimizing parameters, which characterize a behavior of the neural network, with a goal that upon further processing of training examples, the assessment by the cost function is expected to improve.

2. The method as recited in claim 1, wherein the weight of the contribution of at least one cell to the cost function: is set to a first positive value when the training measured data of the training example do not indicate a location in the at least one cell, and is set to a second, higher positive value when the training measured data of the training example indicate at least one location in the at least one cell.

3. The method as recited in claim 2, wherein the second positive value is between eight times and twenty times the first positive value.

4. The method as recited in claim 1, wherein a distribution of the weights within the grid is selected in such a way that cells, within which the training measured data of the training example do not indicate a location, overall supply the same contribution to the cost function as cells, within which the training measured data of the training example indicate at least one location.

5. The method as recited in claim 1, wherein at least one weight is also optimized with a goal that upon further processing of training measured data, the assessment by the cost function is expected to improve.

6. The method as recited in claim 1, wherein: the training is repeated for multiple subdivisions of the spatial region into grids having different mesh widths; and that mesh width, for which the training converges on a best assessment by the cost function, is set as an optimum mesh width for live operation of the neural network.

7. The method as recited in claim 1, wherein training measured data including measured variables, which characterize reflections of radar radiation, laser radiation, and/or ultrasonic waves at locations in the space, are selected.

8. The method as recited in claim 7, wherein the training measured data are obtained by observation of a scenery using a first measuring setup and/or from a first perspective; and the training output variables are obtained by observation of the same scenery using a second measuring setup and/or from a second perspective.

9. The method as recited in claim 7, wherein the training measured data and the training output variables are obtained by observation of a scenery using the same measuring setup and/or from the same perspective.

10. The method as recited in claim 1, wherein the training output variables contain classification scores of the training input variables with respect to one or multiple classes of a predefined classification.

11. A method, comprising the following steps: training a neural network which maps measured data on one or multiple output variables, the measured data assigning values of one or multiple measured variables to locations in the two-dimensional or three-dimensional space, the training including: providing training examples made up of training measured data and associated training output variables, subdividing a spatial region, which contains at least a portion of the locations indicated by the training measured data of a training example, into a grid made up of adjoining cells, aggregating, for each cell of the adjoining cells, the values of the measured variables contained in the training measured data of the training example for all locations in the cell, to form values of the measured variables which relate to the cell, mapping the aggregated values of the measured variables, by the neural network, on one or multiple output variables, assessing deviations of the output variables from the training output variables using a predefined cost function, which is composed in weighted form of contributions of individual cells of the grid, the weight of each contribution being a function of an occupancy of the corresponding cell with locations contained in the training measured data of the training example, and optimizing parameters, which characterize a behavior of the neural network, with a goal that upon further processing of training examples, the assessment by the cost function is expected to improve; supplying measured data, to the trained neural network, which are recorded using at least one sensor carried along by a vehicle; and ascertaining an activation signal from the output variables supplied by the neural network.

12. The method as recited in claim 11, wherein the vehicle is additionally activated using the activation signal.

13. A non-transitory machine-readable data medium on which is stored a computer program configured for monitored training of a neural network which maps measured data on one or multiple output variables, the measured data assigning values of one or multiple measured variables to locations in the two-dimensional or three-dimensional space, the computer program, when executed by one or multiple computers, causing the one or multiple computers to perform the following steps: providing training examples made up of training measured data and associated training output variables; subdividing a spatial region, which contains at least a portion of the locations indicated by the training measured data of a training example, into a grid made up of adjoining cells; aggregating, for each cell of the adjoining cells, the values of the measured variables contained in the training measured data of the training example for all locations in the cell, to form values of the measured variables which relate to the cell; mapping the aggregated values of the measured variables, by the neural network, on one or multiple output variables; assessing deviations of the output variables from the training output variables using a predefined cost function, which is composed in weighted form of contributions of individual cells of the grid, the weight of each contribution being a function of an occupancy of the corresponding cell with locations contained in the training measured data of the training example; and optimizing parameters, which characterize a behavior of the neural network, with a goal that upon further processing of training examples, the assessment by the cost function is expected to improve.

14. One or multiple computers configured for monitored training of a neural network which maps measured data on one or multiple output variables, the measured data assigning values of one or multiple measured variables to locations in the two-dimensional or three-dimensional space, the one or multiple computers configured to: provide training examples made up of training measured data and associated training output variables; subdivide a spatial region, which contains at least a portion of the locations indicated by the training measured data of a training example, into a grid made up of adjoining cells; aggregate, for each cell of the adjoining cells, the values of the measured variables contained in the training measured data of the training example for all locations in the cell, to form values of the measured variables which relate to the cell; map the aggregated values of the measured variables, by the neural network, on one or multiple output variables; assess deviations of the output variables from the training output variables using a predefined cost function, which is composed in weighted form of contributions of individual cells of the grid, the weight of each contribution being a function of an occupancy of the corresponding cell with locations contained in the training measured data of the training example; and optimize parameters, which characterize a behavior of the neural network, with a goal that upon further processing of training examples, the assessment by the cost function is expected to improve.

Description

BRIEF DESCRIPTION OF EXAMPLE EMBODIMENTS

[0041] FIG. 1 shows an exemplary embodiment of method 100 for training a neural network 1.

[0042] FIG. 2 shows a representation of the transfer of measured data 2 provided as a point cloud into a grid 5.

[0043] FIG. 3 shows an exemplary embodiment of method 200 including the complete action chain up to activation of a vehicle 50.

DETAILED DESCRIPTION OF EXAMPLE EMBODIMENTS

[0044] FIG. 1 is a schematic flowchart of an exemplary embodiment of method 100 for training a neural network 1. Neural network 1 maps measured data 2 on one or multiple output variables 3. These measured data 2 assign values 2c of one or multiple measured variables to locations 2b in the two-dimensional or three-dimensional space.

[0045] In step 110, training examples 4 made up of training measured data 2a and associated training output variables 3a are provided. Training measured data 2a each include values 2c of measured variables which are assigned to locations 2b.

[0046] According to block 111, in particular, for example, training measured data 2a including measured variables 2c are selected, which characterize reflections of radar radiation, laser radiation, and/or ultrasonic waves at locations 2b in the space.

[0047] According to block 111a, in particular, for example, training measured data 2a may be selected which were obtained by observation of a scenery using a first measuring setup and/or from a first perspective. Then, according to block 111b, training output variables 3a may be selected, which were obtained by observation of the same scenery using a second measuring setup and/or from a second perspective.

[0048] According to block 111c, in particular, for example, training output variables 3a may be selected which were obtained by observation of a scenery using the same measuring setup and/or from the same perspective.

[0049] According to block 112, in particular, for example, training output variables 3a may be selected which include classification scores of training input variables 2a with respect to one or multiple classes of a predefined classification.

[0050] In step 120, the spatial region which contains at least a part of locations 2b indicated by training measured data 2a of a training example 4 is subdivided into a grid 5 made up of adjoining cells 5a.

[0051] In step 130, for each cell 5a, values 2c of measured variables contained in training measured data 2a of training example 4 for all locations 2b in this cell 5a are aggregated to form values 5b of the measured variables which relate to this cell 5a. This is explained in greater detail in FIG. 2.

[0052] In step 140, these aggregated values 5b of the measured variables are mapped by neural network 1 on one or multiple output variables 3.

[0053] In step 150, deviations of these output variables 3 from training output variables 3a are assessed using a predefined cost function 6, so that an assessment 6b results. This assessment 6b is composed in weighted form of contributions 6a of individual cells 5a of grid 5. Weight α of each contribution 6a is dependent on the occupancy of corresponding cell 5a with locations 2a contained in training measured data 2a of training example 4.

[0054] According to block 151, weight α of contribution 6a of at least one cell 5a to cost function 5 may be set to a first positive value a if training measured data 2a of training example 4 do not indicate a location 2b in this cell 5a. According to block 152, this weight α may be set to a second, higher positive value b if training measured data 2a of training example 4 indicate at least one location 2b in this cell 5a.

[0055] According to block 153, a distribution of weights α within grid 5 may be selected in such a way that cells 5a, within which training measured data 2a of training example 4 do not indicate a location 2b, overall provide the same contribution to cost function 5 as cells 5a, within which training measured data 2a of training example 4 indicate at least one location 2b. For example, first positive value a may be set to a=1−N.sub.2b=0, N.sub.2b=0 being the number of those cells 5a in which no location 2b indicated by measured data 2a of training example 4 falls. Second positive value b may then be set to b=1−N.sub.2b>0, N.sub.2b>0 being the number of those cells 5a in which at least one location 2b indicated by measured data 2a of training example 4 falls.

[0056] In step 160, parameters la, which characterize the behavior of neural network 1, are optimized with the goal that upon further processing of training examples 4, assessment 6b by cost function 6 is expected to improve. The finished trained state of the parameters is identified by reference numeral 1a*. These parameters 1a* characterize the behavior of finished trained neural network 1*.

[0057] According to block 161, in this case at least one weight α may also be optimized with the goal that upon further processing of training measured data 2a, assessment 6b by cost function 6 is expected to improve.

[0058] In step 170, the training is repeated for multiple subdivisions of the spatial region into grids 5 having different mesh widths 5c.

[0059] In step 180, that mesh width 5c for which the training converges on best assessment 6b by cost function 6 is set as optimal mesh width 5c* for the live operation of neural network 1.

[0060] FIG. 2 illustrates the transfer of measured data 2 provided as a point cloud into a grid 5. In this example, measured data 2 are recorded using a radar sensor 51, which is carried along by a vehicle 50. Radar sensor 51 detects radar reflections having measured variables 2c, which come from locations 2b, in its detection area 51a.

[0061] If a spatial region that contains at least a part of detection area 51 is divided into a grid 5 including cells 5a, a location 2b from which a radar reflection comes falls in each of some cells 5a. In the example shown in FIG. 2, two locations 2b, from each of which a radar reflection comes, fall in one cell 5a. For each cell 5a, an aggregated value 5b is formed from measured variables 2c for all radar reflections from locations 2b, in particular cell 5a. This aggregated value 5b is zero for most cells 5a because of a lack of locations 2b of radar reflections located therein. The weighting in cost function 6 according to above-described method 100 begins here.

[0062] FIG. 3 is a schematic flowchart of an exemplary embodiment of method 200 including the complete action chain up to the activation of vehicle 50.

[0063] In step 210, a neural network 1 is trained using above-described method 100.

[0064] In step 220, measured data 2 are supplied to trained neural network 1*, which were recorded using at least one sensor 51 carried along by a vehicle 50.

[0065] In step 230, an activation signal 230a is ascertained from output variables 3 supplied by neural network 1*.

[0066] In step 240, vehicle 50 is activated using this activation signal 230a.