DETAILED ACTION
This action is responsive to Applicant’s reply filed April 27th 2026. This action is made non-final.
Status of the Claims
Claims 1-3, 5, 8, 10 and 12 are amended.
Claim status is currently pending and under examination for Claims 1-13 of which the independent claim is 1.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on April 27th 2026 has been entered.
Response to Amendment
Applicant’s amendments to the Claims have overcome each and every 112(b) and 101 rejections previously set forth in the Final Office Action mailed February 27th 2026.
Applicant’s arguments regarding the art rejections are moot in view of the new grounds of rejection necessitated by applicant’s amendment.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The following are the references relied upon in the rejections below:
Suzuki (US 20180336426 A1)
Chauvin, Christine, Farida Saïd, and Sabine Langlois. "Does the Type of Visualization Influence the Mode of Cognitive Control in a Dynamic System?." International Conference on Intelligent Human Systems Integration. Cham: Springer International Publishing, 2019.
Gupta (US 20150049193 A1)
Gaspar, John G., and Daniel V. McGehee. "Driver brake response to sudden unintended acceleration while parking." Transportation research interdisciplinary perspectives 2 (2019): 100039.
Wu, Junjie, et al. "Local decomposition for rare class analysis." Proceedings of the 13th ACM SIGKDD international conference on Knowledge discovery and data mining. 2007.
Lo (US 20180345958 A1)
Claims 1, 3-5, 9 and 12-13 are rejected under 35 U.S.C. 103 as being unpatentable over Suzuki / Chauvin / Gupta / Gaspar / Wu / Lo.
With respect to claim 1, Suzuki teaches:
a computer program product comprising computer-readable instructions for training a machine … that, when executed in a computer system including one or more computers, cause the computer system to ([Abstract] “An image processing system includes: a vehicle; and an information processing device, … a second image acquiring unit configured to acquire a plurality of second images for machine learning”
[0017] computer-readable non-transitory recording medium, program; [0018] computer)
receive data … by a machine driven by a human or robot driver (Suzuki discloses “the camera CM that is an exemplary image pickup device is equipped in a vehicle CA. The camera CM shoots the periphery of the vehicle CA and generates an image. For example, as illustrated, the camera CM picks up an image of a forward sight of the vehicle CA. Next, the image generated by the camera CM is acquired by an image acquiring device IM” [0037]. See Figure 1 depicting a vehicle being driven with a camera and image acquiring device.);
generate preprocessed data … ([0057] “the parameter is determined after image processing of the image IMG2. For example, by investigating the image IMG2 after image processing of brightness or the like, the image processing system can estimate weather, time or the like.”);
condition the preprocessed data by grouping the preprocessed data based on machine-control parameters (The Examiner interprets “machine-control parameters” according to its broadest reasonable interpretation (BRI) in view of the Applicant’s specification at Page 3, lines 13-16 as encompassing parameters related to driving a vehicle or a driving environment as disclosed by Suzuki.
[0055-0057] “the image processing system performs grouping of the images IMG2 and analyzes a balance, on the server SR side. First, a parameter by which the analysis is performed is previously set in the server SR. The parameter is a condition when the images IMG2 are picked up, or the like. Specifically, the parameter is each time when the images IMG2 are picked up, each weather when the images IMG2 are picked up, or the like. For example, the parameter is determined after image processing of the image IMG2 … the vehicle speed may be determined based on data of a speed sensor. In addition, the parameter of weather may be determined by acquisition of meteorological data or the like from an external device”
See [0057-0061] describing other parameters obtained from processing images.
[0062-0063] “the image processing system sets images IMG2 having an identical parameter, to one group. Then, the image processing system analyzes the balance about whether the number of images is roughly equal among groups … the image processing system groups the images IMG2 into three conditions: “condition 1”, “condition 2” and “condition 3”. Specifically, suppose that the parameter is “weather”. Further, suppose that the “condition 1” is a group “fair”, the “condition 2” is a group “cloudy”, and the “condition 3” is a group “rainy”. In the figure, the abscissa axis indicates the condition, and the ordinate axis indicates the number of images”
See Figure 3 depicting a graph of various groups and their corresponding number of images.
Images are collected and then processed to determine a parameter for the images, therefore the images are preprocessed data. Parameters describe conditions during operation of a vehicle such as weather or headlight settings (see [0059]), therefore conditions describe a driving environment and are related to driving a vehicle (and therefore parameters are machine-control parameters).),
wherein grouping comprises: grouping the [preprocessed] data into groups corresponding to each of the machine-control parameters (Each weather condition is a parameter (‘machine-control parameter’) and images are grouped by parameter (“condition 1”, “condition 2”, “condition 3”).),
determine at least one data imbalance within the groups … of the conditioned data ([0064] “there is a bias to the “condition 1”, and the image processing system outputs an analysis result of a poor balance. Specifically, on the basis of the “condition 1”, the number of the images of the “condition 2” is less than the number of the images of the “condition 1”, by a difference DF1. … when there is no difference or when the difference is equal to or less than a predetermined value, the image processing system outputs an analysis result of a good balance. In the case of the analysis result of a poor balance, the image processing system determines adjustment of the balance”
See Figure 3 depicting how the differences in the number of images of each condition (group) are used to determine data imbalances.);
balance the conditioned data for which an imbalance was determined ([0067] “the image processing system requests an image from the server SR side to the vehicle side. That is, in the case of the example shown in FIG. 3, in step SB05, the image processing system adjusts the balance such that the difference DF1 and the difference DF2 are reduced. Specifically, in the case of the example shown in FIG. 3, the image processing system requests images satisfying the “condition 2”, to the vehicle CA, for reducing the difference DF1. Similarly, the image processing system requests images satisfying the “condition 3”, to the vehicle CA, for reducing the difference DF2”),
and train a second machine … based on sensor measurements … included in the group of the balanced data (Suzuki discloses images IMG2 can capture vehicle speed “by optical flow processing of the image IMG2, the image processing system can estimate vehicle speed or the like. For the determination of the parameter, sensor data or the like may be used. For example, the vehicle speed may be determined based on data of a speed sensor” [0057].
[0080-0081] “the image processing system determines whether the balance of images to be used for the machine learning is good. That is, when the balance is adjusted in step SA04, step SA05, step SB05, step SB06 and the like shown in FIG. 2, the bias as shown in FIG. 3 is small, and the images to be used for the machine learning are often well balanced. In such a case, the image processing system determines that the balance of the images to be used for the machine learning is good … the image processing system proceeds to step SB0702.”).
However, Suzuki does not teach arranging sub-groups in a hierarchy, which is taught by Chauvin:
wherein at least some of the groups comprise sub-groups and the groups and the sub-groups are arranged in a hierarchy (Chauvin discloses Figure 2 (reproduced below) on P. 754 depicting a dendrogram for clustering (‘arranging’) vehicle variables into clusters (‘groups’) and sub-clusters (‘sub-groups’).
PNG
media_image1.png
808
1218
media_image1.png
Greyscale
Chauvin discloses “Figure 2 shows the dendrogram for the clustering of vehicle variables. … Examination of the composition of the clusters of variables and their factor loadings (squared correlations between variables and the first principal component of their parent group), suggests that Cluster 1 (in blue) refers to the speed of the vehicles, Cluster 2 (in red) to the smoothness of the actions upon the acceleration pedal and Cluster 3 (in green) to the smoothness of the actions upon the steering wheel” (P. 754, Sec. 3, ¶3-4).);
for at least some groups corresponding to statistical machine-control parameters, grouping associated data … (Chauvin discloses vehicle variables are arranged into clusters (‘groups’), see Figure 2 above. Cluster 1 (‘group’) contains vehicle variables related to speed and Cluster 2 contains vehicle variables related to acceleration (therefore, Clusters 1 and 2 are groups corresponding to statistical machine-control parameters). Clusters 1 and 2 are further divided (grouped) into sub-clusters (sub-groups).);
Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to modify the method of Suzuki with the hierarchical clusters disclosed by Chauvin to split data into smaller groups. By splitting data into smaller groups, a more accurate and interpretable representation of data can be achieved, leading to better data management.
Furthermore, the combined method of Suzuki/Chauvin, does not teach grouping statistical machine-control parameters into bins, which is taught by Gupta:
grouping associated data into bins representing ranges of numeric values between a minimum and maximum of corresponding machine-control parameters (Gupta discloses Figure 10 (reproduced below) depicting steering angles (‘statistical parameters’) sorted into bins based on increments of negative and positive degrees.
PNG
media_image2.png
714
766
media_image2.png
Greyscale
Gupta discloses “FIG. 10 shows a histogram of thirty consecutive frame sets lying in the same steering angle range observed during the normal driving along urban routes by multiple drivers. The steering angle has been partitioned into bins 160 from -180 degrees to +180 degrees in varying increments of 6 degrees or more for this experiment. The way the angles are partitioned is determined by an external function in which the central angles are (such as about -6 to +6 degrees) divided into two central bins with a width of six degrees” [0091].);
Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to modify the combined method of Suzuki/Chauvin with the bins disclosed by Gupta to bin continuous variables. By binning continuous variables into discrete categories, data is easier to understand and analyze, thereby leading to a better understanding of underlying relationships.
Furthermore, the combined method of Suzuki/Chauvin/Gupta does not teach grouping associated data into clusters corresponding to sub-parameters of corresponding machine-control parameters, which is taught by Gaspar:
and for at least some groups corresponding to contextual machine-control parameters, grouping associated data into clusters corresponding to sub-parameters of corresponding machine-control parameters (The Examiner interprets “contextual machine-control parameters” according to its BRI in view of the Applicant’s specification. The Examiner notes that the applicant provides no meaningful definition of this term in their specification. Accordingly, the Examiner interprets this term as encompassing a brake response as disclosed by Gaspar. A brake response describes the amount of brake force applied by a driver while operating a vehicle, defined as Hard, Gradual, or Light (therefore break responses are qualitative descriptions of the operation of a vehicle and therefore are contextual machine-control parameters).
Gaspar discloses “The cluster analysis resulted in three distinct response clusters, seen in the cluster dendrogram in Fig. 4. By visualizing individual brake response curves by cluster affiliation, as in Fig. 5, these three clusters were defined based on the following characteristics: Hard Braking: braking with greater than 125lbs of force within the first 1.5 s following the SUA event. Brake Pumping/Gradual Braking: braking to a maximum of approximately 125lbs brake force occurring over the 5 s following the SUA event. Light Brake Press: brakeforce less than 75lbs over the span of 5 s following the SUA event. Half the participants responded with light braking, less than 75lbs brake force, whereas just three drivers responded with hard braking” (P. 4, Sec. 3.1, ¶2).
Gaspar discloses Figure 4 on P. 4 (reproduced below) depicting a cluster dendrogram of brake responses. Brake responses are arranged into Hard Brake, Gradual Brake, and Light Brake clusters (and therefore the clusters are groups corresponding to contextual machine-control parameters). The clusters are further divided into sub-clusters (and therefore these sub-clusters correspond to sub-parameters since the sub-clusters further divide the clusters corresponding to contextual machine-control parameters).
PNG
media_image3.png
733
757
media_image3.png
Greyscale
);
Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to modify the combined method of Suzuki/Chauvin/Gupta with the cluster dendrogram of Gaspar to arrange qualitative data into more granular groups. By arranging qualitative data into more granular groups, a more accurate and interpretable representation of data can be achieved, leading to better data management and interpretability.
Furthermore, the combined method of Suzuki/Chauvin/Gupta/Gaspar does not teach determining at least one data imbalance within the groups and the sub-groups of the conditioned data, and balancing a number of data points between bins or clusters of a group or sub-group, which is taught by Wu:
determine at least one data imbalance within the groups and the sub-groups of the conditioned data (Wu discloses balanced subclasses (‘sub-groups’) are produced from large classes (‘groups’), “Specifically, for a data set with an imbalanced class distribution, we perform clustering within each large class and produce sub-classes with relatively balanced sizes … Since the clustering is conducted independently within each class but not across the entire data set, we call it local clustering … By exploiting local clustering within large classes, we can decompose the complex concepts, e.g., nonlinear-separable concepts for linear classifiers, into relatively simple ones, e.g., linearly separable concepts. Another effect of local clustering is to produce subclasses with relatively uniform sizes. In addition, for data sets with highly skewed class distributions, we further integrate the over-sampling technique into the COG scheme and propose the COG with over-sampling technique … COG has the ability to divide imbalanced classes into relatively balanced and small sub-classes” (P. 814-815, Sec. 1).);
wherein balancing the data comprises balancing a number of [data points] between bins or clusters of a group or sub-group (Wu discloses above over-sampling is performed to balance sub-classes. See Figure 2 on P. 816 depicting creating balanced sub-classes using the COG (Classification using Local Clustering) procedure.);
Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to modify the combined method of Suzuki/Chauvin/Gupta/Gaspar with the balancing method disclosed by Wu to balance hierarchical data in a top-to-bottom approach. By balancing hierarchical data from top-to-bottom, it can be ensured that high-level minority classes are equally represented in a dataset since they are balanced first and their information is preserved as the balancing process continues down the hierarchy to more specific subclasses, thus leading to an accurately represented and balanced dataset.
Furthermore, the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu does not teach training a machine to execute a driving maneuver corresponding to a group of balanced data based on sensor measurements associated with a subset of a plurality of trajectories included in the group of the balanced data, which is taught by Lo:
training a machine to execute driving maneuvers based on trajectory data … ([0002] “Autonomous vehicles use a variety of on-board sensors and computer systems to detect nearby objects and use such detections to make control and navigation decisions”
[0004] “An autonomous or semi-autonomous vehicle system can use the collision prediction subsystem to determine a high likelihood of a collision taking place, and in response, automatically perform a specified action to prevent the collision before it takes place. For example, in response to determining a high confidence that a collision will take place, the autonomous or semi-autonomous vehicle system can automatically apply vehicle brakes to prevent the collision.”
[0003] “uses the trajectories of the predicted motion of the object clusters to compute a confidence indicating a likelihood of a future vehicle collision”
An autonomous vehicle uses trajectories to compute a likelihood of a future vehicle collision. In response to a high likelihood of a collision, the autonomous vehicle performs an action (braking) to prevent the collision, therefore executing driving maneuvers based on trajectory data. It is implied an autonomous vehicle is trained if it can predict vehicle collisions and execute an action based on that prediction.),
receive data associated with a plurality of driving maneuvers executed by a machine driven by a human or robot driver ([0033] “The on-board collision detection subsystem 114 includes components that identify objects within a vicinity of the vehicle, analyze motion of the identified objects within recently collected sensor data to predict subsequent motion of the identified objects, and determines if the predicted motion will cause a collision with the vehicle”
[0027] “the vehicle 102 can autonomously apply the brakes if a prediction indicates that a human driver is about to collide with a detected object, e.g., a pedestrian, a cyclist, another vehicle.”
Collected sensor data (‘receive data’) is used to analyze motion of an identified object. An identified object can be a vehicle and the vehicle’s motion is determined based on sensor data, therefore the collected sensor data is associated with a plurality of driving maneuvers since driving maneuvers must be executed by a vehicle (‘machine’) being driven in order for the vehicle to be in motion.),
wherein the data is obtained from a plurality of sensors ([Abstract] “Laser obstacle points derived from recent sensor readings of one or more sensors of a vehicle are initially obtained”);
generate preprocessed data comprising a plurality of trajectories comprising sequences of sensor measurements based on synchronizing the data obtained from the plurality of sensors ([0042] “the occupancy grid 402 can represent the time axis using multiple frames that each correspond to a discrete time point within a period of time during which the laser data is collected. As an example, the occupancy grid 402 can include 12 frames for laser data collected over a two-second time period. In this example, each frame can represent spatial information for laser obstacle points at a discrete time point over the two second time period, i.e., each subsequent frame is associated with a time point that is incremented by 20 ms relative to the time point of a prior frame. As discussed in detail below, the system tracks the changes in coordinate locations of laser obstacle points between frames of the occupancy grid 402 to identify motion of corresponding objects over the time period during which the laser data is collected”
[0009] “clustering the laser obstacle points into one or more object clusters within the vicinity of the vehicle in the occupancy grid”
[0003] “system identifies object clusters in the space-time occupancy grid that are within the vicinity of the vehicle. … uses the trajectories of the predicted motion of the object clusters to compute a confidence indicating a likelihood of a future vehicle collision”
Sensors are used to collect laser obstacle points. Laser obstacle points are represented in an occupancy grid and are organized by frames associated with incremented discrete time points, therefore synchronizing sequences of sensor measurements (laser obstacle points). An occupancy grid is used to identify object clusters and object clusters are used to determine trajectories, therefore generating preprocessed data (trajectories of predicted motion of object clusters).),
wherein each trajectory corresponds to a driving maneuver of the plurality of driving maneuvers (Trajectories of predicted motion of object clusters are determined from object clusters identified from laser obstacle points collected from sensor data. Sensor data (laser obstacle points) is used to determine an object’s motion, therefore sensor data is associated with a plurality of driving maneuvers since driving maneuvers must be executed by a vehicle being driven in order for the vehicle (object) to be in motion (and therefore each trajectory corresponds to a driving maneuver).);
grouping the plurality of trajectories into groups corresponding … ([0061] “generates object cluster data 514, which specifies tracked information for each of the identified object clusters C01-04. As shown, the object cluster data 514 specifies a cluster identifier, a predicted number of identified objects within the object cluster, i.e., based on a determined size of the object clusters, and a cluster trajectory. The cluster trajectory is represented as a series of coordinate locations that are identified for an object cluster”
Lo discloses Figure 6 (reproduced below) depicting a table representing object cluster data. The object cluster data has object clusters (‘groups’) represented by CLUSTER ID. Each object cluster has a cluster trajectory that represents the trajectory of the identified objects in each object cluster. Therefore, trajectories are grouped into groups (object clusters).
PNG
media_image4.png
613
1166
media_image4.png
Greyscale
),
wherein balancing the data comprises balancing a number of trajectories between bins or clusters of a group or sub-group (The Examiner interprets “balancing” according to its BRI in view of the Applicant’s specification as encompassing removing weighted particle trajectories. This interpretation is consistent with the illustrative descriptions in the Applicant’s specification at (P. 4, Lines 4-11), see excerpt below.
Applicant’s written description at (P. 4, Lines 4-11): “the conditioned data may be searched for finding or determining imbalances with regard to the distribution of the data over one or more groups or subgroups. In the context of data imbalances it may also be referred to deviations. The searching of imbalances preferably includes comparing data groups or sub-groups with each other for finding differences/deviations between the compared groups. Further, the conditioned data may be balanced by removing the determined imbalance(s) and the balanced data may be output.”
[0010] “assigning a weight to the particle based on a correspondence between the particle trajectory and a cluster trajectory of an object cluster that includes the particle.”
[0067] “The particle filter 330 assigns a weight to each particle based on a consistency between the predicted particle trajectory and the predicted object cluster trajectory, which are included in particle weight data 604. For example, the particle filter 330 assigns a weight of “1.0” to particles that reside on a cell location occupied by a laser obstacle point in every frame of the occupancy grid 512, i.e., indicating that particle motion is generally consistent with object cluster motion. In an alternative example, the particle filter 330 assigns a weight of “0.0” to particles that does not reside in a cell occupied by a laser obstacle point in one or more frames of the occupancy grid 512, i.e., indicating that particle motion is not consistent with object cluster motion in such frames of the occupancy grid 512.”
[0071] “the assigned weights can be used to remove or reduce the impact of particle collision scores of particles with inconsistent motion”
[0074] As shown, in the example, the particle filter 330 assigns a weight of “1.0” to particle P01, a weight of “0.4” to particle P03, and a weight of P03 “0.0” to particles P02 and P04. In this example, the particle motion for particle P01 is determined to be entirely consistent with cluster motion of the object cluster C03, the particle motion for particle P03 is determined to be somewhat consistent, i.e., discrepancies in locations in one or more frames of the occupancy grid 512, and the particle motion for particles P02 and P04 are not consistent, i.e., discrepancies in a majority of frames of the occupancy grid 512. Because particles P02 and P04 are assigned weights of “0.0,” trajectories of motion detected for these particles are not factored into any adjustments to the predicted motion of the object cluster C03 as a whole for use in collision detection.”
Particle trajectories and cluster trajectories are compared to determine if particle trajectories are consistent (do not deviate) from cluster trajectories. If a particle trajectory is not consistent with a cluster trajectory (object cluster motion), then the particle trajectory is assigned a weight of 0 to effectively remove it from being used in collision detection.
Particle trajectories and a cluster trajectory are located within the same object cluster (see Figure 6). Therefore, particle trajectories and a cluster trajectory are ‘sub-groups’ of an object cluster (‘group’). By comparing a particle trajectory and a cluster trajectory to determine consistency, this is determining a ‘imbalance’ between sub-groups since searching for imbalances involves comparing sub-groups for finding differences/deviations (the difference/deviation being consistency). By assigning a weight of 0 to an inconsistent particle trajectory, the particle trajectory is effectively removed from being used in collision detection, therefore removing a determined imbalance (and therefore balancing data).);
and train a second machine to execute a driving maneuver corresponding to a group of the balanced data based on sensor measurements associated with a subset of the plurality of trajectories included in the group of the balanced data ([0035] “When a planning subsystem 116 receives the one or more collision predictions 135, the planning subsystem 116 can use the one or more collision predictions 135 to make fully-autonomous or semi-autonomous driving decisions. For example, the planning subsystem 116 can generate a fully-autonomous plan to navigate the vehicle to avoid the trajectory of predicted motion for an object that is identified within the collision predictions 135 to have a high confidence of colliding with the vehicle, i.e., by applying the vehicle's breaks or otherwise changing the future trajectory of the vehicle.”
Particle trajectories with weights of 0.0 are not factored into collision detection (see [0074]), therefore ‘balanced’ data based on sensor measurements is used to execute a driving maneuver. Particle trajectories with non-zero weights (‘trajectories included in the group of balanced data’) are obtained from object clusters (see [0010]) and laser obstacle points (‘sensor measurements’) are clustered into object clusters (see [0009]), therefore executing a driving maneuver corresponding to a group of balanced data (non-zero weighted particle trajectories used in collision detection). It is implied an autonomous vehicle is trained if it can predict vehicle collisions and execute an action based on that prediction.).
Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to modify the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu with the technique disclosed by Lo to train an autonomous vehicle to perform a driving maneuver. By training an autonomous vehicle to perform a driving maneuver, trained autonomous vehicles can travel with minimal human interaction, thus relieving human drivers of some driving-related responsibilities and saving drivers’ time.
With respect to claim 3, the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu/Lo teaches:
the computer program product according to claim 1, wherein the conditioned data have n hierarchically arranged groups and sub-groups, with N indicating the highest level group and N-1, N-2, .. , N-n indicating sub-groups of lower levels (Chauvin discloses Figure 2 (reproduced above) on P. 754 depicting a dendrogram for clustering (‘arranging’) vehicle variables into clusters (‘groups’) and sub-clusters (‘sub-groups’). The larger clusters on top of the dendrogram represent higher levels, while the sub-clusters represent lower levels.),
and each group or sub-group is divided into one or more clusters of data or one or more bins of data (Chauvin discloses Figure 2 depicting a dendrogram with sub-clusters (‘sub-group’). The sub-clusters are the result of splitting (‘dividing’) a larger cluster (‘group’).).
Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to modify the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu/Lo with the hierarchical clusters of Chauvin to split data into smaller groups. By splitting data into smaller groups, a more accurate and interpretable representation of data can be achieved, leading to better data management.
With respect to claim 4, the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu/Lo teaches:
the computer program product according to claim 3, wherein the conditioned data is arranged to have at least one of contextual parameters and statistical parameters provided in a group or sub-group (Suzuki discloses grouping weather condition parameters (‘contextual parameters’), “the image processing system sets images IMG2 having an identical parameter, to one group. Then, the image processing system analyzes the balance about whether the number of images is roughly equal among groups … the image processing system groups the images IMG2 into three conditions: “condition 1”, “condition 2” and “condition 3”. Specifically, suppose that the parameter is “weather”. Further, suppose that the “condition 1” is a group “fair”, the “condition 2” is a group “cloudy”, and the “condition 3” is a group “rainy”. In the figure, the abscissa axis indicates the condition, and the ordinate axis indicates the number of images” [0062-0063]. See Figure 3 depicting a graph of various conditions and their corresponding number of images.
Suzuki discloses images IMG2 can capture a vehicle speed parameter (‘statistical parameter’), “by optical flow processing of the image IMG2, the image processing system can estimate vehicle speed or the like. For the determination of the parameter, sensor data or the like may be used. For example, the vehicle speed may be determined based on data of a speed sensor” [0057].),
wherein in the hierarchically arranged conditioned data each bin or cluster of a group of sub-group is connected to one or more bins or clusters of the next lower level sub-group (Chauvin discloses Figure 2 (reproduced above) on P. 754 depicting a dendrogram for clustering (‘sorting’) vehicle variables into clusters (‘groups’) and sub-clusters (‘sub-groups’). The larger clusters on top of the dendrogram represent higher levels, while the sub-clusters represent lower levels. Levels are represented by splitting a cluster into sub-clusters.).
With respect to claim 5, the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu/Lo teaches:
and the statistical parameters include at least one of speed, yaw rate or accelerations (Suzuki discloses a speed parameter, “the image processing system can estimate vehicle speed or the like. For the determination of the parameter, sensor data or the like may be used. For example, the vehicle speed may be determined based on data of a speed sensor” [0057].).
and the received data includes data of a plurality of trajectories driven by the vehicle that is driven by a human or robot driver (Gupta discloses “the vehicle is shown moving in a straight line so as to enable the central vanishing point to be determined. However, when the vehicle turns as a result of a change in its steering angle, the motion of the vehicle can be approximated over relatively short distances (approximately 0.5 to 2 seconds of travel time, depending of vehicle speed) as a straight motion at an angle with respect to the ground Y-axis. Repeating the foregoing process of extracting and tracking the trajectories of feature points for various steering angle ranges as the vehicle moves will enable other vanishing points to be determined, hence enabling the determination of the vanishing line” [0075].).
Gupta teaches extracting and tracking trajectories by moving vehicles is a known method in the art. Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to modify the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu/Lo with the trajectory data disclosed by Gupta because trajectories reveal insights about driver behavior and traffic congestion. By analyzing trajectory data, more information can be learned about driver behavior and traffic congestion, thereby providing data that can be useful to make accurate predictions in autonomous driving.
wherein the contextual parameters include at least one of a left turn, right turn, straight road, complex turn, obstacles, free cruising, lane changing, obstacle following or overtaking (Lo discloses collecting laser obstacle points (‘obstacle’ contextual parameters), “Laser obstacle points derived from recent sensor readings of one or more sensors of a vehicle are initially obtained. The laser obstacle points are projected into a pose coordinate system to generate an occupancy grid of a vicinity of the vehicle. A confidence that any objects represented by the laser obstacle points are on a trajectory that will collide with the vehicle is determined by applying a particle filter to the occupancy grid.” [Abstract].).
Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to modify the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu/Lo with the laser obstacle points disclosed by Lo to use obstacle data to predict vehicle collision. By using obstacle data to predict vehicle collision, a machine learning model can be trained to detect obstacles, thereby enabling an autonomous vehicle to detect obstacles and predict safe paths.
With respect to claim 9, the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu/Lo teaches:
the computer program product according to claim 1, wherein the balancing of the conditioned data is performed level-by-level from a highest level group to a lowest level sub-group (Wu discloses balanced subclasses (‘sub-groups’) are produced from large classes (‘highest level group’), “Specifically, for a data set with an imbalanced class distribution, we perform clustering within each large class and produce sub-classes with relatively balanced sizes … Since the clustering is conducted independently within each class but not across the entire data set, we call it local clustering … By exploiting local clustering within large classes, we can decompose the complex concepts, e.g., nonlinear-separable concepts for linear classifiers, into relatively simple ones, e.g., linearly separable concepts. Another effect of local clustering is to produce subclasses with relatively uniform sizes. In addition, for data sets with highly skewed class distributions, we further integrate the over-sampling technique into the COG scheme and propose the COG with over-sampling technique … COG has the ability to divide imbalanced classes into relatively balanced and small sub-classes” (P. 814-815, Sec. 1). See Figure 2 on P. 816 depicting creating balanced subgroups using the COG (Classification using Local Clustering) procedure.).
Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to modify the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu/Lo with the balancing method of Wu to balance hierarchical data in a top-to-bottom approach. By balancing hierarchical data from top-to-bottom, it can be ensured that high-level minority classes are equally represented in a dataset since they are balanced first and their information is preserved as the balancing process continues down the hierarchy to more specific subclasses, thus leading to an accurately represented and balanced dataset.
With respect to claim 12, the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu/Lo teaches:
the computer program product of claim 1 installed on a distributed computing resource (Suzuki discloses “the image acquiring device IM includes communication parts such as an antenna and a processing integrated circuit (processing IC), and sends the image to an external device such as the server SR, through a network NW” [0038]. Suzuki further discloses “the image processing system acquires second images on the server SR side. The second images are images previously input, images periodically sent from the vehicle CA, … the image processing system performs grouping of the images IMG2 and analyzes a balance, on the server SR side” [0054-055]. See Figure 1 depicting an image processing system configured to send images to a server (‘distributed computing resource’) for processing.).
With respect to claim 13, the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu/Lo teaches:
an artificial intelligence training control device which at least has a memory unit with the computer program product according to claim 1 stored therein (Suzuki discloses “car navigation device is equipped in the vehicle CA, the car navigation device guides a driver who operates the vehicle CA, through a voice, an image, a combination thereof or the like, such that the vehicle CA turns right at the intersection CR … it is desirable for the car navigation device to perform the guide of the route using a signboard LM placed near the intersection CR as a landmark, as illustrated. In such a case, in the vehicle CA, an image of a forward sight of the vehicle CA is picked up by the image pickup device, and image recognition of the signboard LM on the picked-up image is performed. For such an image recognition, for example, it is desirable to perform machine learning based on the image containing the signboard LM” [0075].
Suzuki discloses “the image processing system determines whether the balance of images to be used for the machine learning is good. … when the image processing system determines that the balance of the images to be used for the machine learning is good (YES in step SB0701), the image processing system proceeds to step SB0702 … In step SB0702, the image processing system performs learning of signboards, or the like. Specifically, in the case of using the signboard LM for the guide as shown in FIG. 4, the image processing system performs the learning using the image containing the signboard LM. Thereby, the image processing system can perform the image recognition of the signboard LM on the image” [0080-0082]. See Figure 1 depicting a image processing system that includes a storage device (memory).),
an input interface (Suzuki discloses an input interface as an image acquiring device, “the image acquiring device IM includes an arithmetic device such as an electronic circuit, an electronic control unit (ECU) or a central processing unit (CPU), and a control device. The image acquiring device IM further includes an auxiliary storage device such as a hard disk, and stores the image acquired from the camera CM” [0038].),
an output interface (Suzuki discloses an output interface as an analyzing unit “the analyzing unit ISF6 performs an analyzing step of analyzing the balance among the plurality of groups that is the balance of the number of the second images that belong to the plurality of groups, and outputting an analysis result ANS. For example, the analyzing unit ISF6 is realized by the CPU SH1” [0122]. See Figure 11 depicting an analyzing unit outputting an analysis result in an image processing system.)
and a display unit (Suzuki discloses “The output device SH4 is a display or the like, and outputs a processing result or the like to the user” [0044].),
wherein the input interface configured to receive one or more data sets from a data source being connected to the input interface via a wired connection or a wireless connection (Suzuki discloses an image acquiring device (‘input interface’) receives images (‘data sets’) from a camera (‘data source’), “image acquiring device IM further includes an auxiliary storage device such as a hard disk, and stores the image acquired from the camera CM. Furthermore, the image acquiring device IM includes communication parts such as an antenna and a processing integrated circuit (processing IC), and sends the image to an external device such as the server SR, through a network NW” [0038]. See Figure 1 depicting a connected camera and image acquiring device.),
the output interface configured to output the balanced data (“the analyzing unit ISF6 performs an analyzing step of analyzing the balance among the plurality of groups that is the balance of the number of the second images that belong to the plurality of groups, and outputting an analysis result ANS” [0122].),
the display unit configured to display at least one of the one or more data sets, the conditioned data or the balanced data to a user (Suzuki discloses “The output device SH4 is a display or the like, and outputs a processing result or the like to the user” [0044]. Suzuki further discloses “FIG. 13 is a diagram showing an exemplary processing result in an image processing system according to an embodiment of the disclosure. The left side of FIG. 13 is a diagram showing an analysis result ANS that is the same balance as that in FIG. 3. When images IMG3 are sent, images satisfying the “condition 2” and images satisfying the “condition 3” are added … the image processing system IS roughly equalizes the respective image numbers for the conditions, as shown in the right side of FIG. 13, for example, and thereby, can improve the balance” [0139-0140]. See Fig. 13 depicting unbalanced condition groups and balanced condition groups.).
The following are the references relied upon in the rejections below:
Feng, Xidong, et al. "Vehicle trajectory prediction using intention-based conditional variational autoencoder." 2019 IEEE Intelligent Transportation Systems Conference (ITSC). IEEE, 2019.
Claims 2 and 8 are rejected under 35 U.S.C. 103 as being unpatentable over Suzuki / Chauvin / Gupta / Gaspar / Wu / Lo / Feng.
With respect to claim 2, the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu/Lo teaches:
the computer program product according to claim 1, wherein, by comparing the number of [data points] of the groups or sub-groups with each other, an imbalance is determined if a difference of the number of [data points] included in the groups or sub-groups compared with each other is larger than a predefined threshold (Suzuki discloses a poor balance (‘imbalance’) is determined by comparing the number of images (‘number of data points’) in each condition (group), “there is a bias to the “condition 1”, and the image processing system outputs an analysis result of a poor balance. Specifically, on the basis of the “condition 1”, the number of the images of the “condition 2” is less than the number of the images of the “condition 1”, by a difference DF1. Furthermore, on the basis of the “condition 1”, the number of the images of the “condition 3” is less than the number of the images of the “condition 1”, by a difference DF2. On the other hand, when there is no difference or when the difference is equal to or less than a predetermined value, the image processing system outputs an analysis result of a good balance. In the case of the analysis result of a poor balance, the image processing system determines adjustment of the balance” [0064].).
However, the combination does not teach comparing the number of trajectories of groups to determine an imbalance, which is taught by Feng:
wherein, by comparing the number of trajectories of the groups or sub-groups with each other, an imbalance is determined if a difference of the number of trajectories included in the groups or sub-groups compared with each other is larger … ((P. 3517, Sec. V-A, Last Paragraph) “Less than 3% of trajectories sampled completely randomly involve lane-change, which means the dataset has a serious problem of data imbalance. Therefore, we first randomly sample only lane-change trajectories and get a sampled-set, and then randomly sample all maneuvers of data as another. Finally, we combine these two sets (we use a ratio of 50%:50%) as our training set. As a result, the amount of lane-change trajectories is roughly equal to lane-keeping.”);
Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to modify the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu/Lo with the technique disclosed by Feng to determine a data imbalance between groups of trajectories. By determining a data imbalance between groups of trajectories, it can be determined which trajectory class is uncommon, thereby enabling machine learning engineers to augment rare trajectory data to help a machine learning model to learn uncommon trajectories and make better predictions.
With respect to claim 8, the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu/Lo teaches:
the computer program product according to claim 1 wherein an imbalance is determined by comparing bins or clusters of at least one group or sub-group with each other for finding the bin or cluster with the lowest number of [data points] in the group or sub-group (Suzuki discloses Figure 14 (reproduced below) depicting the calculated differences (DF1 and DF2) between the number of images belonging to each condition group (‘cluster’).
PNG
media_image5.png
539
1057
media_image5.png
Greyscale
Suzuki discloses condition groups are formed from the weather parameter group, “the image processing system sets images IMG2 having an identical parameter, to one group. Then, the image processing system analyzes the balance about whether the number of images is roughly equal among groups … the image processing system groups the images IMG2 into three conditions: “condition 1”, “condition 2” and “condition 3”. Specifically, suppose that the parameter is “weather” [0062-0063].),
and the balancing is performed by … under-sampling of all other bins or clusters in the group or sub-group to the number of [data points] in the bin or cluster with the lowest number of data points (Suzuki discloses images (‘data points’) are deleted from each condition group (‘clusters’) to match the number of images of the smallest condition group, “for adjusting the balance, a method of deleting data is possible. Specifically, in the case of the same balance as the balance on the left side of FIG. 13, images of the “condition 1” and images of the “condition 2” are deleted such that the number of the images of the “condition 1” and the number of the images of the “condition 2” become equal to the number of the images of the “condition 3”. When the balance is adjusted in such a method, the deleted images are wasted” [0145]. See Figure 14 depicting balanced condition groups after deleting images.).
However, the combination does not teach determining an imbalance between groups of trajectories or balancing groups of trajectories, which is taught by Feng:
wherein an imbalance is determined by comparing bins or clusters of at least one group or sub-group with each other for finding the bin or cluster with the lowest number of trajectories in the group or sub-group ((P. 3517, Sec. V-A, Last Paragraph) “Less than 3% of trajectories sampled completely randomly involve lane-change, which means the dataset has a serious problem of data imbalance. Therefore, we first randomly sample only lane-change trajectories and get a sampled-set, and then randomly sample all maneuvers of data as another. Finally, we combine these two sets (we use a ratio of 50%:50%) as our training set. As a result, the amount of lane-change trajectories is roughly equal to lane-keeping.”),
and the balancing is performed by a random under-sampling of all other bins or clusters in the group or sub-group to the number of trajectories in the bin or cluster with the lowest number of data points (A trajectory lane-change class is determined to only make up 3% of a training dataset (therefore determining an imbalance by comparing groups of trajectory classes). To obtain a 50%:50% ratio of trajectory lane-change samples to trajectory lane-keeping samples, the trajectory lane-change samples are randomly sampled to increase to 50% and the trajectory lane-keeping samples are randomly sampled to decrease to 50% (where decreasing is performing random under-sampling to obtain a balanced training dataset).).
Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to modify the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu/Lo with the technique disclosed by Feng to determine a data imbalance between groups of trajectories. By determining a data imbalance between groups of trajectories, it can be determined which trajectory class is uncommon, thereby enabling machine learning engineers to augment rare trajectories and reduce common trajectories to help models learn both uncommon and common trajectories equally.
The following are the references relied upon in the rejections below:
Ma (US 9244790 B1).
Claim 6 is rejected under 35 U.S.C. 103 as being unpatentable over Suzuki / Chauvin / Gupta / Gaspar / Wu / Lo / Ma.
With respect to claim 6, the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu/Lo teaches the computer program product according to claim 5. However, the combination does not teach “wherein, if a number of clusters and bins per group or sub-group is predefined in a database, the number of clusters is equal to the number of sub-parameters of the contextual parameter associated with the clusters and depending on the number of bins associated with a statistical parameter, the range of values of the statistical parameter is evenly distributed over the bins”, which is taught by Ma:
wherein, if a number of clusters and bins per group or sub-group is predefined in a database, the number of clusters is equal to the number of sub-parameters of the contextual parameter associated with the clusters (Ma discloses “a quantile distribution graph for a set of known working disks 1102 for a particular diagnostic parameter (e.g., RAS) have been created … the Y axis represents the values of the corresponding diagnostic parameter and the X axis represents the distribution of the values of the diagnostic parameter. The values of the diagnostic parameter may be sorted and evenly distributed in a predetermined number of intervals, which may be in percentiles or deciles” (Col. 15, line 63 to Col. 15, line 5).),
and depending on the number of bins associated with a statistical parameter, the range of values of the statistical parameter is evenly distributed over the bins (Ma discloses deciles (‘bins’) are evenly distributed, “In this example, the values are evenly distributed in 10 deciles. A decile represents any of the nine values that divide the sorted data into ten equal parts, so that each part represents 1/10 of the sample or population” (Col. 15, lines 5-8).).
Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to modify the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu/Lo with the technique disclosed by Ma to create consistent bins. By using a predetermined number of intervals, data from different datasets can be distributed consistently therefore ensuring that different datasets are processed in a consistent manner which leads to fair comparisons and reliable results.
The following are the references relied upon in the rejections below:
Hu, Hsiao-Wei, Yen-Liang Chen, and Kwei Tang. "A novel decision-tree method for structured continuous-label classification." IEEE transactions on cybernetics 43.6 (2013): 1734-1746.
Claim 7 is rejected under 35 U.S.C. 103 as being unpatentable over Suzuki / Chauvin / Gupta / Gaspar / Wu / Lo / Hu.
With respect to claim 7, the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu/Lo teaches the computer program product according to claim 1, however the combination does not teach a predefined hierarchical structure of groups and sub-groups, which is taught by Hu:
wherein a hierarchical structure of groups and sub-groups as well as bins and clusters of the groups and sub-groups is predefined (Hu discloses Figure 1 depicting a predefined hierarchy tree (‘hierarchical structure’) consisting of predefined ranges (‘bins’) and multiple levels (‘sub-groups’).
PNG
media_image6.png
795
1071
media_image6.png
Greyscale
Hu discloses “the goal is to classify data into classes that are a set of predefined ranges and can be organized in a hierarchy. In the hierarchy, the ranges at the lower levels are more specific and inherently more difficult to predict, whereas the ranges at the upper levels are less specific and inherently easier to predict” (P. 1734, Abstract).
Hu further discloses “a predefined hierarchy tree comprises a set of predefined ranges of expenditures that can be organized in a hierarchy. Each node in a hierarchy tree can be treated as a class label. Because each label is continuous in the data and may belong to multiple class labels, we can further define a label as a hierarchical continuous label. In this example, constructing DTs with hierarchical continuous labels involves building a DT from the data in Table I and simultaneously selecting the most appropriate class label according to the label distribution in the predefined hierarchy tree in Fig. 1” (P. 1735, Sec. 1, First Paragraph).) or is editable by a user, wherein in the latter case the user is at least prompted to input the number of groups and sub-groups as well as the number of bins or clusters per group or sub-group.
Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to modify the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu/Lo with the predefined hierarchy tree of Hu to group continuous data into ranges. By grouping continuous data into predefined ranges, data can be simplified into few, discrete categories which allows for a clearer understanding of patterns and trends.
The following are the references relied upon in the rejections below:
Zhang (US 20210208545 A1)
Claim 10 is rejected under 35 U.S.C. 103 as being unpatentable over Suzuki / Chauvin / Gupta / Gaspar / Wu / Lo / Zhang.
With respect to claim 10, the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu/Lo teaches: the computer program product according to claim 1,
wherein the conditioned data or the balanced data is output to a user … (Suzuki discloses “The output device SH4 is a display or the like, and outputs a processing result or the like to the user” [0044]. Suzuki further discloses “FIG. 13 is a diagram showing an exemplary processing result in an image processing system according to an embodiment of the disclosure. The left side of FIG. 13 is a diagram showing an analysis result ANS that is the same balance as that in FIG. 3. When images IMG3 are sent, images satisfying the “condition 2” and images satisfying the “condition 3” are added … the image processing system IS roughly equalizes the respective image numbers for the conditions, as shown in the right side of FIG. 13, for example, and thereby, can improve the balance” [0139-0140]. See Fig. 13 depicting unbalanced condition groups and balanced condition groups.).
However, the combination does not teach outputting data to a user for validation, which is taught by Zhang:
wherein the [data] is output to a user for validation of the [data processing] operation (Zhang discloses “present the data processing results for user validation via a tag evaluation GUI, through which a user can review, modify, accept, and/or reject the data processing results, and (4) store the user validated data processing results” [0020].
Zhang further discloses “the tag evaluation service 440 may also be linked to a tag evaluation graphical user interface 450 configured to allow user to validate data processing by the tag evaluation service 440. For example, the tag evaluation graphical user interface 450 may be configured to present the valid windows of industrial process data selections, exclusions and/or modifications for user to review, modify, and/or validate. The tag evaluation graphical user interface 450 may be a readable and writeable screen or a webpage assessable by user. User may be allowed to validate the data processing by accepting, rejecting and/or modifying the data processing. Since it is possible that the tag evaluation service 440 may erroneously exclude valid data, it is best for the customer to be the final decider” [0086].).
Zhang teaches displaying data in a graphical user interface for a user to validate is a known method in the art. Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to modify combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu/Lo with the user validation method of Zhang to allow users to confirm data processing results. By allowing users to validate data processing results, users can make adjustments to the data and ensure the results received are suitable for their needs.
The following are the references relied upon in the rejections below:
Herman (US 20160275144 A1)
Claim 11 is rejected under 35 U.S.C. 103 as being unpatentable over Suzuki / Chauvin / Gupta / Gaspar / Wu / Lo / Herman.
With respect to claim 11, the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu/Lo teaches:
the computer program product according to claim 1, wherein the balanced data is output to a user (Suzuki discloses “The output device SH4 is a display or the like, and outputs a processing result or the like to the user” [0044]. Suzuki further discloses “FIG. 13 is a diagram showing an exemplary processing result in an image processing system according to an embodiment of the disclosure. The left side of FIG. 13 is a diagram showing an analysis result ANS that is the same balance as that in FIG. 3. When images IMG3 are sent, images satisfying the “condition 2” and images satisfying the “condition 3” are added … the image processing system IS roughly equalizes the respective image numbers for the conditions, as shown in the right side of FIG. 13, for example, and thereby, can improve the balance” [0139-0140]. See Fig. 13 depicting unbalanced condition groups and balanced condition groups.).
However, the combination does not teach outputting data arranged as concentric rings, which is taught by Herman:
wherein the … data is output to a user such that groups and sub-groups are arranged as concentric rings with different diameters (Herman discloses Figure 3 (reproduced below) depicting a sunburst chart with concentric circles (‘concentric rings’) with differently sized diameters in a graphical user interface.
PNG
media_image7.png
677
948
media_image7.png
Greyscale
Herman discloses “sections 310 are graphically displayed in sunburst chart 303. Sections 310 represent the employee distribution for sections of organization 106. As depicted, the size of each section in sections 310 is based on the number of employees in the section … concentric circles 312 are graphically displayed in sunburst chart 303. Concentric circles 312 represent levels of hierarchy for sections 310. … Concentric circles 312 located toward the center of sunburst chart 303 represent higher levels of hierarchy for sections 310 than concentric circles 312 located toward the outside of sunburst chart 303” [0093-0094].),
and the bins or clusters are segments of the concentric rings (Herman discloses Figure 3 depicting a sunburst chart with concentric circles with differently sized sections (‘bins’).).
Herman teaches displaying a sunburst chart with hierarchical data and concentric rings is a known method in the art. Before the effective filing date of the claimed invention, it would have been obvious to a person of ordinary skill in the art to modify the combined method of Suzuki/Chauvin/Gupta/Gaspar/Wu/Lo with the sunburst chart of Herman to display hierarchical data. By using a sunburst chart to display hierarchical data, the relative size of each category and subcategory can be quickly compared by a user, thus improving user interpretability of the data and helping users make informed data-driven decisions.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Palanisamy et al. (US 20200142421 A1) teaches balancing steering angle data by manually upsampling data and training an autonomous vehicle to perform control commands (steering, accelerating, braking).
Wang et al. (“Improve Aggressive Driver Recognition Using Collision Surrogate Measurement and Imbalanced Class Boosting”) teaches using trajectory data to classify drivers as aggressive or normal drivers, and then performing data balancing to increase samples of the aggressive driver minority class.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to PEDRO J MORALES whose telephone number is (571)272-6106. The examiner can normally be reached 8:30 AM - 6:00 PM.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, MIRANDA M HUANG can be reached at (571)270-7092. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/PEDRO J MORALES/Examiner, Art Unit 2124
/MIRANDA M HUANG/Supervisory Patent Examiner, Art Unit 2124
Prosecution Insights