DETAILED ACTION
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 .
Claim Objections
Claims 1-6 are objected to because of the following informalities: the claims has multiple lengthy run-on limitations, grammar and punctuations that needs to be corrected. The claims also have multiple antecedent basis because of amendments/terms that were removed and not corrected, which examiner have giving 112b rejections below. The claim(s) must be in one sentence form only, wherein an examiner can clearly understand each limitation. Examiner suggests to clearly revise all the claims. Appropriate correction is required.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 3-5 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claim 3 recites the limitation " the class label acquired by the label acquisition unit to the plurality of identification target points ". There is insufficient antecedent basis for this limitation in the claim. The “the label acquisition unit” was crossed out in the amendments and therefore, “the label acquisition unit” should be amended accordingly.
Claim 3 recites the limitation "…with respect to each of the plurality of identification target points into the class label assigning learned model". There is insufficient antecedent basis for this limitation in the claim. The “the class label assigning learned model” is referenced in the deleted limitation of claim 1, however, claim 3 is amended as an independent claim. therefore, “the class label assigning learned model” should be amended accordingly.
Claim 5 recites the limitation " reading the second feature quantity". There is insufficient antecedent basis for this limitation in the claim. The “the second feature quantity” is referenced in claim 4 , however, claim 5 is based on claim 3. therefore, “the second feature quantity” should be amended accordingly.
The term “a value according to a degree of possibility of the same class label” in claim 4 is a relative term which renders the claim indefinite. The term “a degree of possibility” is not defined by the claim, the specification does not provide a standard for ascertaining the requisite degree, and one of ordinary skill in the art would not be reasonably apprised of the scope of the invention.
All the dependent claims that dependence on claim 3 are rejected based on the dependency.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claim 3 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The analysis of the claims’ subject matter eligibility will follow the 2019 Revised Patent Subject Matter Eligibility Guidance, 84 Fed. Reg. 50-57 (January 7, 2019) (“2019 PEG”).
With respect to claim 3.
Claim 3 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1: Is the claim to a process, machine, manufacture, or composition of matter? Yes—claim 1 recites a device, which is a machine.
Step 2A, prong one: Does the claim recite an abstract idea, law of nature or natural phenomenon? Yes—the limitations identified below each, under its broadest reasonable interpretation, covers mental processes abstract idea grouping (concepts performed in the human mind (including an observation, evaluation, judgment, opinion)), see MPEP 2106.04(a)(2), subsection III and the 2019 PEG or a mathematical concept, but for the recitation of generic computer components:
“acquiring a plurality of identification target points by sampling a target point group that is a set of three-dimensional target points; (Mental processes- concept of observation and evaluation of labeling datasets or mathematical data selection operation based on sampling).
calculating relative coordinates of a neighboring point that is a target point set for the identification target point with respect to the identification target point, for each of the plurality of identification target points; (Mental processes- concept of observation and evaluation or mathematical computations based on calculations).
acquiring class labels of the plurality of identification target points and validity of the class labels of each of the plurality of identification target points with respect to the neighboring points…; (Mental processes- concept of observation and evaluation of labeling datasets).
and assigning the class label acquired by the label acquisition unit to the plurality of identification target points, assigns the class labels to the neighboring points for each of the plurality of identification target points when the validity of the class label is included in a range determined by a predetermined threshold value, and identifies the class labels of the identification target points and the neighboring points”: (Mental processes- concept of observation and evaluation of labeling datasets based on a criteria or a score or mathematical comparison based on comparing a value to a threshold).
Step 2A, prong two: Does the claim recite additional elements that integrate the judicial exception into a practical application? No—the judicial exception is not integrated into a practical application.
“acquiring a plurality of identification target points by;” involves the mere gathering of data, which is insignificant extra-solution activity. See MPEP § 2106.05(g).
“a processor configured to execute operations” and “acquiring class labels of the plurality of identification target points and validity of the class labels of each of the plurality of identification target points with respect to the neighboring points by inputting coordinates of the plurality of identification target points and the relative coordinates of the neighboring points with respect to each of the plurality of identification target points into the class label assigning learned model”: using the assigned learned model to implement an abstract idea id merely reciting the words “apply it” (or an equivalent) with the judicial exception, or merely including instructions to implement an abstract idea on a computer, or merely using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f).
The generic computer components in these steps are recited at a high-level of generality (i.e., as a generic computer component performing a generic computer function) such that it amounts no more than mere instructions to apply the exception using a generic computer component. Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea. The claim is directed to an abstract idea.
Step 2B: Does the claim recite additional elements that amount to significantly more than the judicial exception? No—there are no additional limitations beyond the mental processes identified above. The limitation treated above, are directed to the well-understood, routine, and conventional activity of storing and retrieving information in memory. See MPEP § 2106.05(d)(II); Versata Dev. Group, Inc. v. SAP Am., Inc., 793 F.3d 1306, 1334, 115 USPQ2d 1681, 1701 (Fed. Cir. 2015). It also includes limitations that Merely reciting the words “apply it” (or an equivalent) with the judicial exception, or merely including instructions to implement an abstract idea on a computer, or merely using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f). The additional element is insignificant application, which is similar to examples of activities that the courts have found to be insignificant extra-solution activity, in accordance with MPEP 2106.05(g), Insignificant Extra-Solution Activity. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept.
Thus, considering the additional elements individually and in combination and the claims as a whole, the additional elements do not provide significantly more than the abstract idea. This claim is not patent eligible.
Claim Rejections - 35 USC § 103
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.
Claim(s) 1-6 and 10 is/are rejected under 35 U.S.C. 103 as being unpatentable over Qi et al. (“PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space”, 31st Conference on Neural Information Processing Systems (NIPS 2017)) in view of Jiang et al. (“MentorNet: Learning Data-Driven Curriculum for Very Deep Neural Networks on Corrupted Labels”, Proceedings of the 35th International Conference on Machine Learning, Stockholm, Sweden, PMLR 80, 2018).
Regarding claim 1.
Qi teaches a learning device comprising a processor configured to execute operations comprising: acquiring learning data (see page 3 section 3.2, “Given an unordered point set {x1,x2,...,xn}”, also see page 6, section 4, “We evaluate our network on classifying point clouds sampled from both 2D (MNIST) and 3D (ModleNet40) Euclidean spaces. MNIST images are converted to 2D point clouds of digit pixel locations. 3D point clouds are sampled from mesh surfaces from ModelNet40 shapes. In default we use 512 points for MNIST and 1024 points for ModelNet40. In last row (ours normal) in Table 2, we use face normals as additional point features, where we also use more points (N = 5000) to further boost performance. All point sets are normalized to be zero mean and within a unit ball. We use a three-level hierarchical network with three fully connected layers”, i.e. point sets from modelNet40 for example are learning data), wherein the learning data includes at least:
first data including coordinates of a learning identification target point sampled from a plurality of learning target points expressed as a set of three- dimensional target points for learning (see 3, section 3.2, “set of points is processed and abstracted to produce a new set with fewer elements. The set abstraction level is made of three key layers: Sampling layer, Grouping layer and PointNet layer. The Sampling layer selects a set of points from input points, which defines the centroids of local regions. Grouping layer then constructs local region sets by finding “neighboring” points around the centroids. PointNet layer uses a mini-PointNet to encode local region patterns into feature vectors.”, also see page 3, section 3.2, “Given input points {x1,x2,...,xn}, we use iterative farthest point sampling (FPS) to choose a subset of points {xi1 ,xi2 ,...,xim }, such that xij is the most distant point (in metric distance) from the set {xi1 ,xi2 ,...,xij 1 } with regard to the rest points.”),
second data including relative coordinates of a plurality of learning neighboring points associated with the learning identification target point relative to the learning identification target point (see 3, section 3.2, “set of points is processed and abstracted to produce a new set with fewer elements. The set abstraction level is made of three key layers: Sampling layer, Grouping layer and PointNet layer. The Sampling layer selects a set of points from input points, which defines the centroids of local regions. Grouping layer then constructs local region sets by finding “neighboring” points around the centroids. PointNet layer uses a mini-PointNet to encode local region patterns into feature vectors.”, also see 4, “The coordinates of points in a local region are firstly translated into a local frame relative to the centroid point:
PNG
media_image1.png
30
542
media_image1.png
Greyscale
is the coordinate of the centroid.”),
third data including teacher data of a class label of the learning identification target point (see page 3, “Figure 2: Illustration of our hierarchical feature learning architecture and its application for set segmentation and classification using points”),
and fourth data including teacher data of see page 5, “SHREC15: 1200 shapes from 50 categories. Each category contains 24 shapes which are mostly organic ones with various poses such as horses, cats, etc. We use fivefold cross validation to acquire classification accuracy on this dataset.”, also see page 6, “We randomly drop points (see Fig. 4 left) during test time to validate our network’s robustness to non-uniform and sparse data. In Fig. 4 right, we see MSG+DP (multi-scale grouping with random input dropout during training) and MRG+DP (multi-resolution grouping with random input dropout during training) are very robust to sampling density variation.”, also see page 7, “scale point cloud analysis, we also evaluate on semantic scene labeling task. The goal is to predict semantic object label for points in indoor scans. [5] provides a baseline using fully convolutional neural network on voxelized scans.”);
and learning, based on the learning data, a class label assigning model, wherein the class label assigning model (see page 3, “Our hierarchical structure is composed by a number of set abstraction levels (Fig. 2). At each level, a set of points is processed and abstracted to produce a new set with fewer elements. The set abstraction level is made of three key layers: Sampling layer, Grouping layer and PointNet layer.”) includes:
a first model, wherein the first model receives the second data including the relative coordinates of the plurality of neighboring points relative to the learning identification target point and outputs conversion coordinates and a first feature quantity (see 4, “The coordinates of points in a local region are firstly translated into a local frame relative to the centroid point:
PNG
media_image1.png
30
542
media_image1.png
Greyscale
is the coordinate of the centroid.”, also see page 3, “The Sampling layer selects a set of points from input points, which defines the centroids of local regions. Grouping layer then constructs local region sets by finding “neighboring” points around the centroids. PointNet layer uses a mini-PointNet to encode local region patterns into feature vectors.”),
the conversion coordinates are obtained by converting the relative coordinates of the learning neighboring points, a second model, wherein the second model receives the first data including the coordinates of the learning identification target point and the first feature quantity as input and outputs a second feature quantity and the third data including the class label of the learning identification target point (see page 4, “The input to this layer is a point set of size N ⇥ (d + C) and the coordinates of a set of centroids of size N0 ⇥ d. The output are groups of point sets of size N0 ⇥ K ⇥ (d + C), where each group corresponds to a local region and K is the number of points in the neighborhood of centroid points. Note that K varies across groups but the succeeding PointNet layer is able to convert flexible number of points into a fixed length local region feature vector.”, wherein second feature and class label of target point is the “application for set segmentation and classification using points ” from figure 2 and “progressively abstract larger and larger local regions along the hierarchy” from page 3, section 3.2),
and a third model, see page 5, “We adopt a hierarchical propagation strategy with distance based interpolation and across level skip links (as shown in Fig. 2). In a feature propagation level, we propagate point features from
PNG
media_image2.png
26
614
media_image2.png
Greyscale
are point set size of input and output of set abstraction level l. We achieve feature propagation by interpolating feature values f of Nl points at coordinates of the Nl 1 points.”).
Qi do not specifically teach fourth data including teacher data of validity of the class label of the learning identification target point; and wherein the third model receives the second feature quantity and the conversion coordinates and outputs validity data indicating validity of respective class labels of the plurality of learning neighboring points.
Jiang teaches fourth data including teacher data of validity of the class label of the learning identification target point; and wherein the third model receives the second feature quantity and the conversion coordinates and outputs validity data indicating validity of respective class labels of the plurality of learning neighboring points (see page 3 section 3.1, “In this paper, we assign binary labels to v∗ i , where v∗ i = 1 iff yi is a correct label. As v∗ i is binary, Θ is learned by minimizing the cross-entropy loss between v∗ i and g(zi;Θ). Intuitively, this process is similar to a mock test for the teacher (MentorNet) to learn to update her teaching strategy (curriculum). The student (StudentNet) provides features φ(·,·,w) for the mock test using the latest model w. The teacher can learn an updated curriculum from the data to better supervise the latest student model. The learned curriculum is jointly determined by the teacher and student together.”, also page 6, section 5, “Fig. 3 plots the training and test error on the clean validation data, under a representative setting: resnet-101 on CIFAR 100 of 40% noise, where the x-axis denotes the training iteration. The y-axis is the validation error on the clean validation in Fig. 3(a) and the mini-batch training error on corrupted labels in Fig. 3(b).”).
Both Qi and Jiang pertain to the problem of neural network radar systems, thus being analogous. It would have been obvious to one skilled in the art before the effective filing date of the claimed invention to combine Qi and Jiang to teach the above limitations. The motivation for doing so would be “we propose a novel technique of learning another neural network, called MentorNet, to supervise the training of the base deep networks, namely, StudentNet. During training, MentorNet provides a curriculum (sample weight ing scheme) for StudentNet to focus on the sample the label of which is probably correct. Un like the existing curriculum that is usually predefined by human experts, MentorNet learns a data-driven curriculum dynamically with Student Net. Experimental results demonstrate that our approach can significantly improve the generalization performance of deep networks trained on corrupted training data. Notably, to the best of our knowledge, we achieve the best-published result on WebVision, a large benchmark containing 2.2 million images of real-world noisy labels.” (see Jiang Abstract).
Regarding claim 2.
Qi and Jiang teaches the learning device according to claim 1,
Jiang further teaches wherein: the learning further comprises learning the class label by either minimizing or maximizing a function and generating a learnt class label assigning model (see page 3, section 3.1, MentorNet can be learned to 1) approximate existing curriculums or 2) discover new curriculums from data
PNG
media_image3.png
238
596
media_image3.png
Greyscale
) using the learning data corresponding to each of a plurality of the learning identification target points (see page 3 section 3.1,
PNG
media_image4.png
354
586
media_image4.png
Greyscale
i.e. i=1…n are plurality of training samples),
wherein the function is based at least on:
PNG
media_image5.png
508
602
media_image5.png
Greyscale
),
wherein the class label is associated with the learning identification target point output from the class label assigning model during learning or before learning (see page 3, section 3.1, “It takes the input of a mini-batch of samples, and outputs their corresponding sample weights. The feature zi = φ(xi,yi,w) includes the loss, loss difference to the moving average, label and epoch percentage. pt maintains an exponential moving average on the p-th percentile of the loss in each mini-batch. For a sample, its loss and loss difference − pt over the last few epochs can be encoded by a bidirectional LSTM network to capture the prediction variance”),
and the teacher data represents a correct answer value of the class label of the learning identification target point, and a deviation between the validity of the class label and the teacher data (see page 4, section 3.2, “This paper tackles the problem of overcoming corrupted la bels. It is interesting to analyze why the learned curriculum can improve the generalization performance. It turns out that StudentNet, when jointly learned with MentorNet, may optimize an underlying robust objective and the objective is also related to the robust M-estimator… Proposition 1. Suppose (x,y) denotes a training sample and its corrupted label. For simplicity, let the MentorNet input φ(x,y,w) = be the loss computed by the StudentNet model parameter w. The MentorNet gm( ;Θ) = v, where v is the sample weight. If gm decreases with respect to l, then there exists an underlying robust objective F”),
wherein the class label is associated with the learning neighboring point output from the class label assigning model during learning or before learning and the teacher data represents a correct answer value of the validity of the class label of the learning neighboring point (see page 3 section 3.1, “In this paper, we assign binary labels to v∗ i , where v∗ i = 1 iff yi is a correct label. As v∗ i is binary, Θ is learned by minimizing the cross-entropy loss between v∗ i and g(zi;Θ). Intuitively, this process is similar to a mock test for the teacher (MentorNet) to learn to update her teaching strategy (curriculum). The student (StudentNet) provides features φ(·,·,w) for the mock test using the latest model w. The teacher can learn an updated curriculum from the data to better supervise the latest student model. The learned curriculum is jointly determined by the teacher and student together.”, also page 6, section 5, “Fig. 3 plots the training and test error on the clean validation data, under a representative setting: resnet-101 on CIFAR 100 of 40% noise, where the x-axis denotes the training iteration. The y-axis is the validation error on the clean validation in Fig. 3(a) and the mini-batch training error on corrupted labels in Fig. 3(b).”).
The motivation utilized in the combination of claim 1, super, applies equally as well to claim 2.
Regarding claim 3.
Qi teaches a identification device comprising a processor configured to execute operations comprising: acquiring a plurality of identification target points by sampling a target point group that is a set of three-dimensional target points (see page 3 section 3.2, “Given an unordered point set {x1,x2,...,xn}”, also see page 6, section 4, “We evaluate our network on classifying point clouds sampled from both 2D (MNIST) and 3D (ModleNet40) Euclidean spaces. MNIST images are converted to 2D point clouds of digit pixel locations. 3D point clouds are sampled from mesh surfaces from ModelNet40 shapes. In default we use 512 points for MNIST and 1024 points for ModelNet40. In last row (ours normal) in Table 2, we use face normals as additional point features, where we also use more points (N = 5000) to further boost performance. All point sets are normalized to be zero mean and within a unit ball. We use a three-level hierarchical network with three fully connected layers”, i.e. point sets from modelNet40 for example are learning data, also see 3, section 3.2, “set of points is processed and abstracted to produce a new set with fewer elements. The set abstraction level is made of three key layers: Sampling layer, Grouping layer and PointNet layer. The Sampling layer selects a set of points from input points, which defines the centroids of local regions. Grouping layer then constructs local region sets by finding “neighboring” points around the centroids. PointNet layer uses a mini-PointNet to encode local region patterns into feature vectors.”, also see page 3, section 3.2, “Given input points {x1,x2,...,xn}, we use iterative farthest point sampling (FPS) to choose a subset of points {xi1 ,xi2 ,...,xim }, such that xij is the most distant point (in metric distance) from the set {xi1 ,xi2 ,...,xij 1 } with regard to the rest points.”);
calculating relative coordinates of a neighboring point that is a target point set for the identification target point with respect to the identification target point, for each of the plurality of identification target points (see 3, section 3.2, “set of points is processed and abstracted to produce a new set with fewer elements. The set abstraction level is made of three key layers: Sampling layer, Grouping layer and PointNet layer. The Sampling layer selects a set of points from input points, which defines the centroids of local regions. Grouping layer then constructs local region sets by finding “neighboring” points around the centroids. PointNet layer uses a mini-PointNet to encode local region patterns into feature vectors.”, also see 4, “The coordinates of points in a local region are firstly translated into a local frame relative to the centroid point:
PNG
media_image1.png
30
542
media_image1.png
Greyscale
is the coordinate of the centroid.”);
acquiring class labels of the plurality of identification target points and see page 3, “Figure 2: Illustration of our hierarchical feature learning architecture and its application for set segmentation and classification using points”, see page 4, “The input to this layer is a point set of size N ⇥ (d + C) and the coordinates of a set of centroids of size N0 ⇥ d. The output are groups of point sets of size N0 ⇥ K ⇥ (d + C), where each group corresponds to a local region and K is the number of points in the neighborhood of centroid points. Note that K varies across groups but the succeeding PointNet layer is able to convert flexible number of points into a fixed length local region feature vector.”, wherein second feature and class label of target point is the “application for set segmentation and classification using points ” from figure 2 and “progressively abstract larger and larger local regions along the hierarchy” from page 3, section 3.2);
and assigning the class label acquired by the label acquisition unit to the plurality of identification target points, see page 5, “We adopt a hierarchical propagation strategy with distance based interpolation and across level skip links (as shown in Fig. 2). In a feature propagation level, we propagate point features from
PNG
media_image2.png
26
614
media_image2.png
Greyscale
are point set size of input and output of set abstraction level l. We achieve feature propagation by interpolating feature values f of Nl points at coordinates of the Nl 1 points.”).
Qi do not specifically teach validity of the class labels of each of the plurality of identification target points with respect to the neighboring points by inputting coordinates of the plurality of identification target points; and assigns the class labels to the neighboring points for each of the plurality of identification target points when the validity of the class label is included in a range determined by a predetermined threshold value.
Jiang teaches validity of the class labels of each of the plurality of identification target points with respect to the neighboring points by inputting coordinates of the plurality of identification target points; and assigns the class labels to the neighboring points for each of the plurality of identification target points when the validity of the class label is included in a range determined by a predetermined threshold value (see page 3 section 3.1, “In this paper, we assign binary labels to v∗ i , where v∗ i = 1 iff yi is a correct label. As v∗ i is binary, Θ is learned by minimizing the cross-entropy loss between v∗ i and g(zi;Θ). Intuitively, this process is similar to a mock test for the teacher (MentorNet) to learn to update her teaching strategy (curriculum). The student (StudentNet) provides features φ(·,·,w) for the mock test using the latest model w. The teacher can learn an updated curriculum from the data to better supervise the latest student model. The learned curriculum is jointly determined by the teacher and student together.”, also page 6, section 5, “Fig. 3 plots the training and test error on the clean validation data, under a representative setting: resnet-101 on CIFAR 100 of 40% noise, where the x-axis denotes the training iteration. The y-axis is the validation error on the clean validation in Fig. 3(a) and the mini-batch training error on corrupted labels in Fig. 3(b).”).
Both Qi and Jiang pertain to the problem of neural network radar systems, thus being analogous. It would have been obvious to one skilled in the art before the effective filing date of the claimed invention to combine Qi and Jiang to teach the above limitations. The motivation for doing so would be “we propose a novel technique of learning another neural network, called MentorNet, to supervise the training of the base deep networks, namely, StudentNet. During training, MentorNet provides a curriculum (sample weight ing scheme) for StudentNet to focus on the sample the label of which is probably correct. Un like the existing curriculum that is usually predefined by human experts, MentorNet learns a data-driven curriculum dynamically with Student Net. Experimental results demonstrate that our approach can significantly improve the generalization performance of deep networks trained on corrupted training data. Notably, to the best of our knowledge, we achieve the best-published result on WebVision, a large benchmark containing 2.2 million images of real-world noisy labels.” (see Jiang Abstract).
Regarding claim 4.
Qi and Jiang teaches the identification device according to claim 3,
Qi further teaches wherein the class label assigning learned model includes a learned first model (see 3, section 3.2, “set of points is processed and abstracted to produce a new set with fewer elements. The set abstraction level is made of three key layers: Sampling layer, Grouping layer and PointNet layer, also see 4, “The coordinates of points in a local region are firstly translated into a local frame relative to the centroid point:
PNG
media_image1.png
30
542
media_image1.png
Greyscale
is the coordinate of the centroid.”, also see page 3, “The Sampling layer selects a set of points from input points, which defines the centroids of local regions. Grouping layer then constructs local region sets by finding “neighboring” points around the centroids. PointNet layer uses a mini-PointNet to encode local region patterns into feature vectors.”), a learned second model (see 3, section 3.2, “set of points is processed and abstracted to produce a new set with fewer elements. The set abstraction level is made of three key layers: Sampling layer, Grouping layer and PointNet layer, also see page 4, “The input to this layer is a point set of size N ⇥ (d + C) and the coordinates of a set of centroids of size N0 ⇥ d. The output are groups of point sets of size N0 ⇥ K ⇥ (d + C), where each group corresponds to a local region and K is the number of points in the neighborhood of centroid points. Note that K varies across groups but the succeeding PointNet layer is able to convert flexible number of points into a fixed length local region feature vector.”, wherein second feature and class label of target point is the “application for set segmentation and classification using points ” from figure 2 and “progressively abstract larger and larger local regions along the hierarchy” from page 3, section 3.2), and a learned third model, the learned third model, based on conversion coordinates obtained by converting the relative coordinates of the neighboring point output from the learned first model and a second feature quantity output from the learned second model (see 3, section 3.2, “set of points is processed and abstracted to produce a new set with fewer elements. The set abstraction level is made of three key layers: Sampling layer, Grouping layer and PointNet layer, also see page 5, “We adopt a hierarchical propagation strategy with distance based interpolation and across level skip links (as shown in Fig. 2). In a feature propagation level, we propagate point features from
PNG
media_image2.png
26
614
media_image2.png
Greyscale
are point set size of input and output of set abstraction level l. We achieve feature propagation by interpolating feature values f of Nl points at coordinates of the Nl 1 points.”),
Jiang teaches outputs validity of the class labels for the neighboring points for each of the plurality of identification target points according to a function, and wherein the function outputs a value according to a degree of possibility of the same class label being assigned to the identification target point and the neighboring point (see page 3 section 3.1, “In this paper, we assign binary labels to v∗ i , where v∗ i = 1 iff yi is a correct label. As v∗ i is binary, Θ is learned by minimizing the cross-entropy loss between v∗ i and g(zi;Θ). Intuitively, this process is similar to a mock test for the teacher (MentorNet) to learn to update her teaching strategy (curriculum). The student (StudentNet) provides features φ(·,·,w) for the mock test using the latest model w. The teacher can learn an updated curriculum from the data to better supervise the latest student model. The learned curriculum is jointly determined by the teacher and student together.”, also page 6, section 5, “Fig. 3 plots the training and test error on the clean validation data, under a representative setting: resnet-101 on CIFAR 100 of 40% noise, where the x-axis denotes the training iteration. The y-axis is the validation error on the clean validation in Fig. 3(a) and the mini-batch training error on corrupted labels in Fig. 3(b).”).
The motivation utilized in the combination of claim 3, super, applies equally as well to claim 4.
Regarding claim 5.
Qi and Jiang teaches the identification device according to claim 3,
Qi further teaches wherein the acquiring class labels further comprises: inputting the relative coordinates with respect to the identification target points of the target points for each of the plurality of identification target points into the learned first model among the class label assigning learned models inputting the relative coordinates with respect to the identification target points of the target points for each of the plurality of identification target points into the learned first model among the class label assigning learned models (see 3, section 3.2, “set of points is processed and abstracted to produce a new set with fewer elements. The set abstraction level is made of three key layers: Sampling layer, Grouping layer and PointNet layer. The Sampling layer selects a set of points from input points, which defines the centroids of local regions. Grouping layer then constructs local region sets by finding “neighboring” points around the centroids. PointNet layer uses a mini-PointNet to encode local region patterns into feature vectors.”, also see 4, “The coordinates of points in a local region are firstly translated into a local frame relative to the centroid point:
PNG
media_image1.png
30
542
media_image1.png
Greyscale
is the coordinate of the centroid.”)
reading the second feature quantity from an information storage unit that stores the second feature quantity (see page 4, “The input to this layer is a point set of size N ⇥ (d + C) and the coordinates of a set of centroids of size N0 ⇥ d. The output are groups of point sets of size N0 ⇥ K ⇥ (d + C), where each group corresponds to a local region and K is the number of points in the neighborhood of centroid points. Note that K varies across groups but the succeeding PointNet layer is able to convert flexible number of points into a fixed length local region feature vector.”, wherein second feature and class label of target point is the “application for set segmentation and classification using points ” from figure 2 and “progressively abstract larger and larger local regions along the hierarchy” from page 3, section 3.2) and the class label output from the learned second model when the coordinates of the identification target point and the relative coordinates of the neighboring point with respect to the identification target point for each of the plurality of identification target points are input into the class label assigning learned model reading the second feature quantity from an information storage unit that stores the second feature quantity and the class label output from the learned second model when the coordinates of the identification target point and the relative coordinates of the neighboring point with respect to the identification target point for each of the plurality of identification target points are input into the class label assigning learned model (see page 5, “We adopt a hierarchical propagation strategy with distance based interpolation and across level skip links (as shown in Fig. 2). In a feature propagation level, we propagate point features from
PNG
media_image2.png
26
614
media_image2.png
Greyscale
are point set size of input and output of set abstraction level l. We achieve feature propagation by interpolating feature values f of Nl points at coordinates of the Nl 1 points.”)
Jiang teaches acquiring validity of a class label of the target point by inputting the read second feature quantity and the conversion coordinates into the learned third model among the class label assigning learned models (see page 3 section 3.1, “In this paper, we assign binary labels to v∗ i , where v∗ i = 1 iff yi is a correct label. As v∗ i is binary, Θ is learned by minimizing the cross-entropy loss between v∗ i and g(zi;Θ). Intuitively, this process is similar to a mock test for the teacher (MentorNet) to learn to update her teaching strategy (curriculum). The student (StudentNet) provides features φ(·,·,w) for the mock test using the latest model w. The teacher can learn an updated curriculum from the data to better supervise the latest student model. The learned curriculum is jointly determined by the teacher and student together.”, also page 6, section 5, “Fig. 3 plots the training and test error on the clean validation data, under a representative setting: resnet-101 on CIFAR 100 of 40% noise, where the x-axis denotes the training iteration. The y-axis is the validation error on the clean validation in Fig. 3(a) and the mini-batch training error on corrupted labels in Fig. 3(b).”),
and wherein the assigning the class label further comprises: referencing the class label, and assigning the class label of the identification target point, of which the validity of the class label is included in the range determined by the predetermined threshold value, to the target point to identify the class label of the target point (see page 3 section 3.1, “In this paper, we assign binary labels to v∗ i , where v∗ i = 1 iff yi is a correct label. As v∗ i is binary, Θ is learned by minimizing the cross-entropy loss between v∗ i and g(zi;Θ). Intuitively, this process is similar to a mock test for the teacher (MentorNet) to learn to update her teaching strategy (curriculum). The student (StudentNet) provides features φ(·,·,w) for the mock test using the latest model w. The teacher can learn an updated curriculum from the data to better supervise the latest student model. The learned curriculum is jointly determined by the teacher and student together.”, also page 6, section 5, “Fig. 3 plots the training and test error on the clean validation data, under a representative setting: resnet-101 on CIFAR 100 of 40% noise, where the x-axis denotes the training iteration. The y-axis is the validation error on the clean validation in Fig. 3(a) and the mini-batch training error on corrupted labels in Fig. 3(b).”).
The motivation utilized in the combination of claim 3, super, applies equally as well to claim 5.
Claims 6 and 10 recites a method to perform the device recited in claims 1-2. Therefore the rejection of claims 1-2 above applies equally here.
Related prior arts:
Qi et al. (“PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation”, IEEE 2018) teaches a novel type of neural network that directly consumes point clouds, which well respects the permutation invariance of points in the input. Our network, named PointNet, pro vides a unified architecture for applications ranging from object classification, part segmentation, to scene semantic parsing. Though simple, PointNet is highly efficient and effective. Empirically, it shows strong performance on par or even better than state of the art. Theoretically, we provide analysis towards understanding of what the network has learnt and why the network is robust with respect to input perturbation and corruption.
Saruta et al. (US 20170039417 A1) teaches an image recognition apparatus includes detecting, setting, acquiring, selecting, and specifying. At least one part of an identification target is selected from an identification target image. An inquiry region is set based on the detected part. A feature amount of the set inquiry region is acquired. At least one instance image corresponding to the identification target image is selected based on the acquired feature amount. A specific region of the identification target from the identification target image is specified based on the selected instance image.
FUKUI et al. (US 20170243077 A1) teaches unifying unit unifies images of identification target regions cut out from, a learning image. The memory stores a learning model. The storing unit stores identification target images converted into images of different image sizes. The setting unit sets a position and a size of a candidate region which is likely to include an identification target object of an identification target image. The selecting unit selects an identification target image of an image size with which the size of the cut-out candidate region is closest to the fixed size. The extracting unit extracts the information. The determining unit determines a target object included in the image of the candidate region.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to IMAD M KASSIM whose telephone number is (571)272-2958. The examiner can normally be reached 10:30AM-5:30PM, M-F (E.S.T.).
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, Michael J. Huntley can be reached at (303) 297 - 4307. 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.
/IMAD KASSIM/ Primary Examiner, Art Unit 2129
Prosecution Insights