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 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.
Claims 1-2 and 5 are rejected under 35 U.S.C. 103 as being unpatentable over Burch US 20150332075 in view of Won US 20130070074 and further in view of Ikeda US 20230216872.
Regarding claim 1, Burch teaches An apparatus, comprising: a device configured to be positioned on a human body; a vibration sensor in the device; and at least one processor circuit in the device, the at least one processor circuit having a memory comprising instructions, that when executed by the processor circuit, (Burch US 20150332075 abstract; [0006]-[0008]; [0030]-[0033]; [0037]-[0040]; [0043]-[0045]; [0053]-[0054]; [0063]-[0067]; [0084]; [0086]; [0103]-[0112]; [0121]-[0125]; [0140]-[0142]; figures 1-11)
In certain embodiments, device 100 may include one or more bioacoustic sensors 170 configured to receive, capture, process, and interpret bioacoustic information. In some aspects, bioacoustic information may include acoustics (e.g., vibrations) in and on a living subject produced upon skin-to-skin contact (e.g., when a finger taps an arm, a palm, another finger, etc.), bodily movements (e.g., making a fist), or other body stimuli. In some embodiments, bioacoustic sensor 170 may comprise a single sensor or an array of sensors, as depicted in FIG. 1. For example, in one embodiment, bioacoustic sensor 170 may comprise an array of piezo films (e.g., MiniSense 100, other cantilever-type vibration sensors, etc.) designed to detect vibrations throughout a human body. Device 100 may include appropriate hardware and software components (e.g., circuitry, software instructions, etc.) for transferring signals and information to and from bioacoustics sensor 170 to conduct processes consistent with the disclosed embodiments. As described below, bioacoustics sensor 170 may assist in the detection of certain types of input events, such as gesture inputs (Burch par. 37). FIG. 2A depicts a block diagram of example components of a wearable processing device 100 consistent with the disclosed embodiments. In some embodiments, device 100 may include one or more processors 202 connected to a communications backbone 206 such as a bus, circuitry, wiring, or external communications network (e.g., any medium of digital data communication such as a LAN, MAN, WAN, cellular network, WiFi network, NFC link, Bluetooth, GSM network, PCS network, network 320 of FIG. 6, etc., and any associated protocols such as HTTP, TCP/IP, RFID, etc.). Any component of device 100 may communicate signals over backbone 206 to exchange information and/or data. For example, in one aspect, projector 110, scanner 120, depth camera 130, display 140, speaker 150, microphone 160, and/or bioacoustic sensor 170 may exchange information with each other, provide or receive information or signals to and from processor 202, store or retrieve information in memory, provide or receive information to and from external computing systems, and so on. In some embodiments, components not pictured in FIG. 2 may also communicate over backbone 206, such as an accelerometer, RF circuitry, GPS trackers, vibration motor, card readers, etc. For example, device 100 may include a GPS receiver (not shown) for receiving location and time information from satellites and may communicate such information to the other components of the device, such as processor 202 (Burch par. 43).
Burch does not explicitly teach causes the at least one processor circuit to at least: input a sample of vibration data from the vibration sensor into a trained convolutional neural network, the vibration data having been generated from a vibration event, the trained convolutional neural network outputting one of a plurality of predefined vibration event descriptors; and wherein the trained convolutional neural network is adapted based at least in part on a plurality of Siamese contrastive loss calculations, each Siamese contrastive loss calculation being generated from a corresponding pair of preexisting samples of vibration data from a pool of preexisting samples of vibration data.
Won teach causes the at least one processor circuit to at least: input a sample of vibration data from the vibration sensor into a trained convolutional neural network, the vibration data having been generated from a vibration event, the trained convolutional neural network outputting one of a plurality of predefined vibration event descriptors; (Won US 20130070074 abstract; paragraphs [0065]-[0072]; [0082]-[0083]; [0235]-[0236]; [0241]-[0251]; figures 1-27)
The method includes the steps of training the neural network with data generated using a "3" dimensional finite element method ("3D FEM"); and determining a transformation between 3D FEM data of a model of the tactile sensor and actual tactile sensor data (Won par. 66). The controller may be configured to calculate a characteristic of the object by: transforming data gathered from the tactile sensor to values for a model of the tactile sensor; and inputting the transformed data to a neural network to obtain as output estimated characteristics of the object, wherein the neural network was trained using data from the model (Won par. 72).
According to the cited passages and figures, examiner interprets tactile sensor as vibration sensor and tactile sensor data as vibration data.
Therefore, it would have been obviously to one of ordinary skill in the art before the effective filing date of the invention to apply the value from tactile sensor and tactile sensor data into the neural network taught by Won reference into the modified system of Burch reference in order to tracking a motion pattern of human body.
The combination of Burch and Won do not explicitly teach wherein the trained convolutional neural network is adapted based at least in part on a plurality of Siamese contrastive loss calculations, each Siamese contrastive loss calculation being generated from a corresponding pair of preexisting samples of vibration data from a pool of preexisting samples of vibration data.
Ikeda teach wherein the trained convolutional neural network is adapted based at least in part on a plurality of Siamese contrastive loss calculations, each Siamese contrastive loss calculation being generated from a corresponding pair of preexisting samples of vibration data from a pool of preexisting samples of vibration data. (Ikeda US 20230216872 abstract; paragraphs [0004]-[0005]; [0094]-[0098]; [0102]-[0114]; [0141]-[0144]; [0148]-[0149]; figures 1-13)
The learning unit 13 learns a conversion model with use of sample data by way of metric learning. In metric learning, a metric (e.g., a distance or a similarity degree) between pieces of data is learned. For example, a Siamese network, a triplet network, or the like is used in metric learning (Ikeda par. 102). FIG. 6 is a diagram for describing one example of metric learning. In the example of FIG. 6, the conversion model is learned using a loss function that makes use of a distance between low-dimensional vectors after conversion of feature vectors. For example, in the Siamese network, a contrastive loss function is used as the loss function. In the example of FIG. 6, the conversion model is learned so that the distance between a positive example pair is shortened and the distance between a negative example pair is increased (Ikeda par. 103). Note that Xi and Xj of FIG. 6 represent feature vectors of sample data. NN of FIG. 6 represents a neural network that converts feature vectors into low-dimensional vectors. Zi and Zj in FIG. 6 represent low-dimensional vectors. Also, Loss i,j represents a contrastive loss with respect to sample data (Ikeda par. 104).
According to the cited passages and figures, examiner interprets the sample data in figure 6 as the substitution for the vibration data.
Therefore, it would have been obviously to one of ordinary skill in the art before the effective filing date of the invention to apply Siamese network, a contrastive loss function taught by Ikeda reference into the modified system of Burch and Won reference in order to improve an accuracy of tracking distance between a positive example pair and a negative example pair (Ikeda par. 103).
Regarding claim 2, the combination of Burch, Won and Ikeda disclose The apparatus of claim 1, wherein the pool of preexisting samples of vibration data further comprises a first portion of the preexisting samples of vibration data being generated from a plurality of source domains, and a second portion of the preexisting samples of vibration data being generated from a target domain.
The learning unit 13 learns a conversion model with use of sample data by way of metric learning. In metric learning, a metric (e.g., a distance or a similarity degree) between pieces of data is learned. For example, a Siamese network, a triplet network, or the like is used in metric learning (Ikeda par. 102). FIG. 6 is a diagram for describing one example of metric learning. In the example of FIG. 6, the conversion model is learned using a loss function that makes use of a distance between low-dimensional vectors after conversion of feature vectors. For example, in the Siamese network, a contrastive loss function is used as the loss function. In the example of FIG. 6, the conversion model is learned so that the distance between a positive example pair is shortened and the distance between a negative example pair is increased (Ikeda par. 103). Note that Xi and Xj of FIG. 6 represent feature vectors of sample data. NN of FIG. 6 represents a neural network that converts feature vectors into low-dimensional vectors. Zi and Zj in FIG. 6 represent low-dimensional vectors. Also, Loss i,j represents a contrastive loss with respect to sample data (Ikeda par. 104).
According to the cited passages and figures, examiner interprets the sample data in figure 6 as the substitution for the vibration data.
Regarding claim 5, the combination of Burch, Won and Ikeda disclose The apparatus of claim 1, wherein the device is a smartwatch.
FIG. 4A depicts an example illustration of a worn device 100 on a user's arm consistent with the disclosed embodiments. In some aspects, a user 302 may wear device 100 on his or her person. Device 100 may be designed for comfortable use on a user's arm, shoulder, leg, wrist (e.g., as a watch), or any other body part (e.g., via connector 180) (Burch par. 65).
According to the cited passages and figures, examiner interprets wearable device 100 as the smartwatch.
Claim 3 is rejected under 35 U.S.C. 103 as being unpatentable over Burch US 20150332075, in view of Won US 20130070074, in view of Ikeda US 20230216872 and further in view of Thorne et al. US 20170364558.
Regarding claim 3, the combination of Burch, Won and Ikeda teach all the limitations in the claim 2.
The combination of Burch, Won and Ikeda do not explicitly teach The apparatus of claim 2, wherein the first portion of the preexisting samples of vibration data generated from the plurality of source domains comprises at least 90 percent of a total number of preexisting samples in the pool.
Thorne et al. teach The apparatus of claim 2, wherein the first portion of the preexisting samples of vibration data generated from the plurality of source domains comprises at least 90 percent of a total number of preexisting samples in the pool. (Thorne et al. US 20170364558 paragraph [0035];)
In an exemplary embodiment, an API allows a user to search via an SQL query submitted to the system. The user can also use the API to define the data sample rate that they would like to use, which may be any amount up to and including 100%. The sample rate selected by a user dictates the number of table partitions that the system will search for a particular query. It may be desired by a user to only query 20% of the available data in order to obtain a quicker response than a query of all the data will provide. However, if a user believes that a low rate will present an anomaly, or if a user otherwise desires a larger set of data to search, the user may select a higher percentage (such as 90% or 100%) of the data to query. Selection of a higher rate may cause a longer delay in result time. It will be appreciated by those of ordinary skill in the art that users may have various reasons for selecting different data sampling rates (Thorne et al. par. 35).
Therefore, it would have been obviously to one of ordinary skill in the art before the effective filing date of the invention to substitute a method of selecting 90% of the data taught by Thorne et al. reference into the modified system of Burch, Won and Ikeda reference and the result of the substitution would be predictable.
Claim 4 is rejected under 35 U.S.C. 103 as being unpatentable over Burch US 20150332075, in view of Won US 20130070074, in view of Ikeda US 20230216872 and further in view of Pamu et al. US 20130301430.
Regarding claim 4, the combination of Burch, Won and Ikeda teach all the limitations in the claim 2.
The combination of Burch, Won and Ikeda do not explicitly teach The apparatus of claim 2, wherein the second portion of the preexisting samples of vibration data generated from the target domain comprises at least 10 percent of a total number of preexisting samples in the pool.
Pamu et al. teach The apparatus of claim 2, wherein the second portion of the preexisting samples of vibration data generated from the target domain comprises at least 10 percent of a total number of preexisting samples in the pool. (Pamu et al. US 20130301430 paragraph [0035];)
The amount of data that is used as the sample data may vary from situation to situation based on a variety of factors. For example, in one embodiment, the amount of sample data is a particular fraction of the data to be sent (e.g. 10% of data 202 may be selected as the sample data). In situations where the amount of data to be sent is not known ahead of time, the size of the sample data may be a predetermined size, such as four PDUs (Pamu et al. par. 35).
Therefore, it would have been obviously to one of ordinary skill in the art before the effective filing date of the invention to substitute a method of selecting 10% of the data sample taught by Pamu et al. reference into the modified system of Burch, Won and Ikeda reference and the result of the substitution would be predictable.
Claim 17 is rejected under 35 U.S.C. 103 as being unpatentable over Won US 20130070074 in view of Ikeda US 20230216872.
Regarding claim 17, Won teaches A method, comprising: inputting a sample of vibration data from a vibration sensor into a trained convolutional neural network, the vibration data having been generated from a touch interaction, the trained convolutional neural network outputting one of a plurality of predefined touch interaction positions; (Won US 20130070074 abstract; paragraphs [0065]-[0072]; [0082]-[0083]; [0127]; [0188]; [0235]-[0236]; [0241]-[0251]; [0287]; figures 1-27)
The method includes the steps of training the neural network with data generated using a "3" dimensional finite element method ("3D FEM"); and determining a transformation between 3D FEM data of a model of the tactile sensor and actual tactile sensor data (Won par. 66). The controller may be configured to calculate a characteristic of the object by: transforming data gathered from the tactile sensor to values for a model of the tactile sensor; and inputting the transformed data to a neural network to obtain as output estimated characteristics of the object, wherein the neural network was trained using data from the model (Won par. 72). For the following discussion see FIGS. 1 and 2. The embodiment of the sensor 100 used in the additional experiments and the mouse experiments is as follows. The sensor 100 comprises an optical waveguide 110, a digital imager 130, light sources 120, and plano-convex lens 135. The optical waveguide 110 was composed of PDMS with three layers 110.1, 110.2, 110.3. The elastic moduli of each PDMS layer 110.1, 110.2, 110.3, was set to the moduli values of epidermis (elastic coefficient 1.4.times.10.sup.5 Pa), dermis (8.0.times.10.sup.4 Pa) and subcutanea (3.4.times.10.sup.4 Pa), respectively, of a human finger to realize sensitivity to the level of the human touch sensation. The light sensor 130 was a mono-cooled complementary camera with 8.4 .mu.m (H).times.9.8 .mu.m (V) individual pixel size. The maximum lens resolution was 768 (H).times.492 (V) with an angle of view of 60.degree.. The light sensor 130 was placed below an optical waveguide 110. A heat-resistant borosilicate glass plate 115 was placed as a substrate between the light sensor 130 and the optical waveguide 110 to sustain it without losing resolution. The glass plate 115 emulates the bone or the nail in the human finger. The internal light source 120 is a micro-LED 120 with a diameter of 1.8 mm. There were four LEDs 120 used on four sides of the waveguide 110 to provide illumination. The direction and incident angle of light sources 120 were calibrated with the cone of an acceptance angle for the total internal reflection in a waveguide. The plano-convex lenses 135 was used for the light coupling between the waveguide 110 and light sources 120 (Won par. 287).
According to the cited passages and figures, examiner interprets tactile sensor as vibration sensor and tactile sensor data as vibration data.
Won does not explicitly teaches wherein the trained convolutional neural network is retrained periodically based at least in part a plurality of Siamese contrastive loss calculations, each Siamese contrastive loss calculation being generated from a corresponding pair of samples of vibration data from a pool of samples of vibration data generated by individual ones of a plurality of source domains and a target domain.
Ikeda teaches wherein the trained convolutional neural network is retrained periodically based at least in part a plurality of Siamese contrastive loss calculations, each Siamese contrastive loss calculation being generated from a corresponding pair of samples of vibration data from a pool of samples of vibration data generated by individual ones of a plurality of source domains and a target domain. (Ikeda US 20230216872 abstract; paragraphs [0004]-[0005]; [0094]-[0098]; [0102]-[0114]; [0141]-[0144]; [0148]-[0149]; figures 1-13)
The learning unit 13 learns a conversion model with use of sample data by way of metric learning. In metric learning, a metric (e.g., a distance or a similarity degree) between pieces of data is learned. For example, a Siamese network, a triplet network, or the like is used in metric learning (Ikeda par. 102). FIG. 6 is a diagram for describing one example of metric learning. In the example of FIG. 6, the conversion model is learned using a loss function that makes use of a distance between low-dimensional vectors after conversion of feature vectors. For example, in the Siamese network, a contrastive loss function is used as the loss function. In the example of FIG. 6, the conversion model is learned so that the distance between a positive example pair is shortened and the distance between a negative example pair is increased (Ikeda par. 103). Note that Xi and Xj of FIG. 6 represent feature vectors of sample data. NN of FIG. 6 represents a neural network that converts feature vectors into low-dimensional vectors. Zi and Zj in FIG. 6 represent low-dimensional vectors. Also, Loss i,j represents a contrastive loss with respect to sample data (Ikeda par. 104).
According to the cited passages and figures, examiner interprets the sample data in figure 6 as the substitution for the vibration data.
Therefore, it would have been obviously to one of ordinary skill in the art before the effective filing date of the invention to apply Siamese network, a contrastive loss function taught by Ikeda reference into the modified system of Won reference in order to improve an accuracy of tracking distance between a positive example pair and a negative example pair (Ikeda par. 103).
Claim 18 is rejected under 35 U.S.C. 103 as being unpatentable over Won US 20130070074 in view of Ikeda US 20230216872 and further in view of Burch US 20150332075.
Regarding claim 18, the combination of Won and Ikeda teach all the limitations in the claim 17.
The combination of Won and Ikeda do not explicitly teach The method of claim 17, wherein the input samples of vibration data and the preexisting samples of vibration data undergo noise reduction prior to input into the trained convolutional neural network.
Burch teaches The method of claim 17, wherein the input samples of vibration data and the preexisting samples of vibration data undergo noise reduction prior to input into the trained convolutional neural network.
In some embodiments, device 100 may refine the results of the edge and/or corner detection using the depth coordinates of each pixel. For example, device 100 may require two pixels forming an edge to be within a threshold distance of each other (e.g., to differentiate the object from the background or other distance objects). Moreover, device 100 may suppress pixels flagged as corners having a smooth depth gradient above a threshold (e.g., because the pixel is not truly a corner). Device 100 may employ other types of object detection algorithms suitable for such purposes (e.g., Gabor filters, noise reduction, Sobel operators, image gradients, depth gradients, etc.), and the discussion of certain algorithms herein is for illustrative purposes only. For example, device 100 may compute derivatives of a current depth field to generate a derivative depth maps or derivative intensity/color maps. These derivatives may be computed numerically by determining the change in the depth or intensity between two points (e.g., adjacent pixels), and dividing that change by the distance between them. In some aspects, device 100 may use these gradient or derivative maps to conduct further object recognition (e.g., recognizing a user's finger as discussed below), perform further image refinement (e.g., edge detection), and so on (Burch par. 86).
According to the cite passages and figures, examiner interprets the system using the noise reduction before input into the neural network.
Therefore, it would have been obviously to one of ordinary skill in the art before the effective filing date of the invention to apply noise reduction taught by Burch reference into the modified system of Won and Ikeda reference to improve an accuracy of tracking data within a surrounding environment.
Claim 19 is rejected under 35 U.S.C. 103 as being unpatentable over Won US 20130070074 in view of Ikeda US 20230216872 and further in view of Han US 20190332940.
Regarding claim 19, the combination of Won and Ikeda teach all the limitations in the claim 17.
The combination of Won and Ikeda do not explicitly teach The method of claim 17, wherein data related to a domain of the sample of vibration data is removed prior to input into the trained convolutional neural network.
Han teaches The method of claim 17, wherein data related to a domain of the sample of vibration data is removed prior to input into the trained convolutional neural network. (Han US 20190332940 abstract; paragraphs [0043]-[0044]; [0058]- [0065]; [0080]; figures 1-5;)
In some embodiments, time information can be involved with neural networks, such as a recurrent neural network (RNN) and a Long Short Term Memory network (LSTM), and the removed portion of the training data can be related to time domain. It is appreciated that these neural networks can process sequences of data. Therefore, a stimula of the neural network may not only come from a new input data from time T but also from historical information from time T−1. Therefore, before the training data is input to the neural network, first training data associated with time moment T1 can be removed out while second training data associated with time moment T2 can be provided to the neural network for training (Han par. 44).
Therefore, it would have been obviously to one of ordinary skill in the art before the effective filing date of the invention to substitute the method of removing the data before input to the neural network taught by Han reference into the modified system of Won and Ikeda reference and the result of the substitution would be predictable.
Claim 20 is rejected under 35 U.S.C. 103 as being unpatentable over Won US 20130070074 in view of Ikeda US 20230216872 and further in view of Shen et al. US 20220327035.
Regarding claim 20, the combination of Won and Ikeda teach all the limitations in the claim 17.
The combination of Won and Ikeda do not explicitly teach The method of claim 17, wherein the samples of vibration data generated by the target domain comprises unlabeled domain data generated by use of a device that includes the vibration sensor.
Shen et al. teach The method of claim 17, wherein the samples of vibration data generated by the target domain comprises unlabeled domain data generated by use of a device that includes the vibration sensor. (Shen et al. US 20220327035 abstract; paragraphs [0005]-[0015]; [0055]-[0060]; [0064]-[0067; figures 1-7)
Step 2: performing an FFT on samples in a source domain and samples in a target domain, and feeding labeled samples in the source domain and unlabeled samples in the target domain at the same time into a deep intra-class adaptation convolutional neural network model with initialized parameters in a training stage (Shen et al. par. 7)
Therefore, it would have been obviously to one of ordinary skill in the art before the effective filing date of the invention to substitute the method of removing the data before input to the neural network taught by Shen et al. reference into the modified system of Won and Ikeda reference and the result of the substitution would be predictable.
Allowable Subject Matter
Claims 6-11 objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims.
Claims 6-16 are allowed.
The following is an examiner’s statement of reasons for allowance:
Regarding claim 6, Burch US 20150332075, Won US 20130070074, Ikeda US 20230216872, Kocienda et al. US 20230280821, Karri et al. US 20230089701, Deselaers et al. US 20180188938, Lee et al. US 20170319134, Woo et al. US 20160127624, Brehmer US 20150242094, Kosaki US 20140068476 and Horowitz et al. US 9785247 are the closest art. They are teaching every limitation of claim 6 except for a limitation cited “The apparatus of claim 1, further comprising: generating by way of a domain discriminator a plurality of instances of a contrast loss based on the pair of preexisting samples of vibration data; inverting individual ones of the instances of the contrast loss; and retraining the convolutional neural network based upon the inverted instances of the contrast loss.”.
After update search, there are none of the prior arts of record singularly or combination, teaches or fairly suggest the features present in the claim 6 “The apparatus of claim 1, further comprising: generating by way of a domain discriminator a plurality of instances of a contrast loss based on the pair of preexisting samples of vibration data; inverting individual ones of the instances of the contrast loss; and retraining the convolutional neural network based upon the inverted instances of the contrast loss.”.
Prior arts of record fail to disclose “The apparatus of claim 1, further comprising: generating by way of a domain discriminator a plurality of instances of a contrast loss based on the pair of preexisting samples of vibration data; inverting individual ones of the instances of the contrast loss; and retraining the convolutional neural network based upon the inverted instances of the contrast loss.”. However, upon consideration of the claim invention, there is no reason to combine the applied references to arrive in the context of the claim invention.
Claims 7-11 depend on and further limit of independent claim 6, therefore claims 7-11 are considered allowable for the same reason.
Regarding claim 12, Burch US 20150332075, Won US 20130070074, Ikeda US 20230216872, Kocienda et al. US 20230280821, Karri et al. US 20230089701, Deselaers et al. US 20180188938, Lee et al. US 20170319134, Woo et al. US 20160127624, Brehmer US 20150242094, Kosaki US 20140068476 and Horowitz et al. US 9785247 are the closest art. They are teaching every limitation of claim 12 except for a limitation cited “retrain the trained convolutional neural network in relation to the plurality of predefined touch interaction positions based at least in part on a difference in loss value of the sample of vibration data compared to a preexisting sample of vibration data, wherein the retraining of the trained convolutional neural network includes adapting the trained convolutional neural network using a plurality of Siamese contrastive loss calculations, where individual ones of the Siamese contrastive loss calculations are generated from a corresponding pair of preexisting samples of vibration data from a pool of preexisting samples of vibration data.”.
After update search, there are none of the prior arts of record singularly or combination, teaches or fairly suggest the features present in the claim 12 “retrain the trained convolutional neural network in relation to the plurality of predefined touch interaction positions based at least in part on a difference in loss value of the sample of vibration data compared to a preexisting sample of vibration data, wherein the retraining of the trained convolutional neural network includes adapting the trained convolutional neural network using a plurality of Siamese contrastive loss calculations, where individual ones of the Siamese contrastive loss calculations are generated from a corresponding pair of preexisting samples of vibration data from a pool of preexisting samples of vibration data.”.
Prior arts of record fail to disclose “retrain the trained convolutional neural network in relation to the plurality of predefined touch interaction positions based at least in part on a difference in loss value of the sample of vibration data compared to a preexisting sample of vibration data, wherein the retraining of the trained convolutional neural network includes adapting the trained convolutional neural network using a plurality of Siamese contrastive loss calculations, where individual ones of the Siamese contrastive loss calculations are generated from a corresponding pair of preexisting samples of vibration data from a pool of preexisting samples of vibration data.”. However, upon consideration of the claim invention, there is no reason to combine the applied references to arrive in the context of the claim invention.
Claims 13-16 depend on and further limit of independent claim 12, therefore claims 13-16 are considered allowable for the same reason.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to THANG D TRAN whose telephone number is (408)918-7546. The examiner can normally be reached Monday - Friday 8:00 am - 5:30 pm (pacific time).
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, Brian A Zimmerman can be reached at 571-272-3059. 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.
/THANG D TRAN/Examiner, Art Unit 2686
/BRIAN A ZIMMERMAN/Supervisory Patent Examiner, Art Unit 2686
Prosecution Insights