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 .
Response to Arguments
Regarding applicants arguments directed at the 35 USC 103 rejection:
Applicant argues on page 11-12 of the remarks that Mosallanezhad does not explicitly teach the adding of new embedding information:
Examiner respectfully disagrees that adding new information is a requirement. The claim does not require that there be “additional information”. Examiner notes that the document can be the host data acting as a container. Mosallanezhad teaches embedding information into a document (host data). The claims do not define what the container may be, what the host data may be, or that the information needs to be additionally. These limiting interpretation should be amended into the claims as reading them into the claims, absent the language, would be improper.
Furthermore the examiner respectfully notes what constitutes “additional information,” modifying current data to produce a new representation can be considered additional information.
Applicant further argues that the claims generate the container containing the embedded information, examiner respectfully notes the language of “generating” is not in the specification. The claims speak on an already existing container that had information embedded into it. These limiting interpretation should be amended into the claims as reading them into the claims, absent the language, would be improper.
Furthermore applicant argues on page 13 of the remarks that the host data carries both the original host content and additional embedded information. Examiner again respectfully disagrees and notes the claims do not speak on “original host content,” claim 1 defines the container as the host data, and therefore the document in M is interpreted to be the container which is by definition the host data. These limiting interpretation should be amended into the claims as reading them into the claims, absent the language, would be improper.
Examiner note: Applicant is respectfully requested to amend to the interpretation argued, the language will always be interpreted under BRI, anything narrower would require the examiner to read in elements that are absent from the claims.
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.
Claim 4-5 and 14-15 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.
Regarding claim 4, which depends from claim 3: Claim 3 recites the first task output is based on inputting “another container with hidden information” to produce a task output on “the container”, yet claim 4 recites the first task output is generated based on “the container with embedded information” examiner is unsure how the task can be output on the container and then the container with embedded information which again is defined as the container with hidden information. Examiner respectfully notes again the confusion between the container, the container with embedded information, and the another container with hidden information. It is all the same container. The container exists in different forms in different stages, for example prior to any steps AKA at stage 1 its merely a container, then stage two information is embedded into the container, the result of the embedding creates the same container but now has hidden information AKA stage 3. Yet the claims continue to claim them as being different, and is confusing because the only difference is the stage of when they are being acted on.
Claims 5, and 14-15 inherit the same rejection as claim 4 above in light of their dependency, or for reciting similar limitations.
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 factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claim(s) 1, 6-11, and 16-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Mosallanezhad et al. (US 20220374606 A1) hereinafter M in view of Liu et al (US 20190281024 A1)
Regarding claim 1, M teaches a method, comprising: ([0017] method)
performing, by a processor of an apparatus, ([0087-0091] system with hardware processors executing software stored in memory)
task-aware information hiding ([0017-0018] teachings anonymizing users private data and other textual information [0023] In this disclosure: (1) the problem of text anonymization is studied by learning a reinforced task-aware text anonymizer, (2) a data-utility task-aware checker is incorporated into the present system to ensure that the utility of textual embeddings is preserved with respect to a given task, and (3) experiments are conducted on real-world data to demonstrate the effectiveness of the system 100 in an important natural language processing task. See also detailed Task -Aware anonymizer in paragraphs 0035-0046) or partial task-aware information hiding (Examiner notes M teaches task-aware information hiding (see above) thereby meeting “task-aware information hiding or partial task-aware information hiding”)
using an information hiding (anonymizing; see above mapping) artificial intelligence (AI)/machine learning (ML) model (0027 deep learning model 104 see also 0051++ teaching deep learning) to embed information (manipulation to produce new text embedding representation) in a host data serving as a container; (0024-0025; teach the containers may be documents, with embedded textual information, more specifically extracting embedded data, and re-embedding anonymized host input data; ([0027] 2) the deep RL-based privacy and utility preserver 104. Then, the deep RL preserver 104 manipulates the embedded text representation by learning the optimal strategy so that both privacy and utility of the embedded representation are preserved. The deep RL preserver 104 includes two sub-components: 1) private attribute classifier component D.sub.P 142, and 2) task-aware utility classifier component D.sub.U 144. The private attribute classifier component 142 seeks to infer user private-attribute information based on their embedded text representation. The utility classifier component 144 incorporates the given manipulated embedded text representation for a given task T and investigates the usefulness of the latent representation for T.[0047] the RL privacy and utility preserver 104 which incorporates a private attribute classifier component 142 and manipulates the initial given text embedding representation 112 accordingly to fool the attacker. See also 0050)
M does not explicitly teach communicating, by the processor, with a network using the container containing the embedded information.
In an analogous art Liu teaches communicating, by the processor, with a network using the container containing the embedded information. ([0034] Information hiding, which is also known as data hiding, is an emerging technology field that combines multidisciplinary theories and technologies. Information hiding technology mainly refers to the embedding of information into digital host information such as text, digitized sound, images, video signals, and the like. The purpose of information hiding is not to limit the correct information storing, reading, and accessing, but to ensure that the hidden information does not attract the attention and concern of the monitor, thereby reducing the possibility of being attacked. The information hiding may include hidden technology, visual cryptography technology, and watermarking technology. The information hiding in this application may be the hidden technology. That is, the secret information is embedded in the information that looks ordinary to be sent to prevent the third party from detecting the secret information. The particular information processed by the information hiding method may be regarded as hidden information. [0038] It should be noted that the communication system provided by the present application may be, but is not limited to, a communication solution involving three client terminals. The above first client terminal 102 is a sender. The above second client terminal 104 is a receiver. The above third client terminal 106 is a third-party application. The above first client terminal 102 (the sender) may embed the communication information that needs to be kept secret into the carrier object in a manner of information hiding. )
It would have been obvious to one of ordinary skill in the art prior to the effective filing of the application to modify the teachings of [M] to include [sending a host container with hidden embedded data through a network] as is taught by [Liu].
The suggestion/motivation for doing so is to [improve secure transmission [0002-0004]].
Regarding claims 6, M in view of Liu teach the method of Claim 1, (Examiner respectfully notes claim 6 is dependent on claim 1 which recites additional limitations to further modify the alternative of performing, partial task-aware information hiding. From a claim interpretation standard, Claim 6 includes all the limitations of claim 1 and further modifies the alternative limitation of performing, partial task-aware information hiding. Accordingly, claim 6 is met by references M and Liu by disclosing limitation performing, task-aware information hiding, since the alternative limitation performing, partial task-aware information hiding is written in the alternative (and it is not required to be taught be M and Liu, since M and Liu already teach performing, task-aware information hiding).
Regarding claims 7, M in view of Liu teach the method of Claim 1, (Claim 7 is dependent on claim 1 which recites additional limitations to further modify the alternative of performing, partial task-aware information hiding. From a claim interpretation standard, Claim 7 includes all the limitations of claim 1 and further modifies the alternative limitation of performing, partial task-aware information hiding. Accordingly, claim 7 is met by references M and Liu by disclosing limitation performing, task-aware information hiding, since the alternative limitation performing, partial task-aware information hiding is written in the alternative (and it is not required to be taught be M and Liu, since M and Liu already teach performing, task-aware information hiding)).
Regarding claims 8, M in view of Liu teach the method of Claim 1, (Claim 8 is dependent on claim 1 which recites additional limitations to further modify the alternative of performing, partial task-aware information hiding. From a claim interpretation standard, Claim 8 includes all the limitations of claim 1 and further modifies the alternative limitation of performing, partial task-aware information hiding. Accordingly, claim 8 is met by references M and Liu by disclosing limitation performing, task-aware information hiding, since the alternative limitation performing, partial task-aware information hiding is written in the alternative (and it is not required to be taught be M and Liu, since M and Liu already teach performing, task-aware information hiding)).
Regarding claims 9, M in view of Liu teach the method of Claim 1, (Claim 9 is dependent on claim 1 which recites additional limitations to further modify the alternative of performing, partial task-aware information hiding. From a claim interpretation standard, Claim 9 includes all the limitations of claim 1 and further modifies the alternative limitation of performing, partial task-aware information hiding. Accordingly, claim 9 is met by references M and Liu by disclosing limitation performing, task-aware information hiding, since the alternative limitation performing, partial task-aware information hiding is written in the alternative (and it is not required to be taught be M and Liu, since M and Liu already teach performing, task-aware information hiding)).
Regarding claims 10, M in view of Liu teach the method of Claim 1, (“Claim 10 is dependent on claim 1 which recites additional limitations to further modify the alternative of performing, partial task-aware information hiding. From a claim interpretation standard, Claim 10 includes all the limitations of claim 1 and further modifies the alternative limitation of performing, partial task-aware information hiding. Accordingly, claim 10 is met by references M and Liu by disclosing limitation performing, task-aware information hiding, since the alternative limitation performing, partial task-aware information hiding is written in the alternative (and it is not required to be taught be M and Liu, since M and Liu already teach performing, task-aware information hiding)).
Regarding claims 11, and 16-20, the claims inherit the same rejections as claim 1, and 6-10 above for reciting similar limitations in the form of an apparatus claim, M teaches a system. (0090-0092)
Claim(s) 2 and 12 is/are rejected under 35 U.S.C. 103 as being unpatentable over Mosallanezhad et al. (US 20220374606 A1) hereinafter M in view of Liu et al (US 20190281024 A1) in view of Yadav et al. (US 20210406589 A1).
Regarding claim 2, M in view of Liu teach the method of Claim 1, and is disclosed above, M in view of Liu not explicitly teach wherein the performing of the task-aware information hiding comprises hiding a part of the information that does not affect an AI/ML task working on the container.
In an analogous art Yadav teaches wherein the performing of the task-aware information hiding comprises hiding a part of the information (obscuring portions of the data) that does not affect an AI/ML task (models such as ML classifiers working on the image) working on the container(image).(0024: The device 110/system(s) 120 receives (132) input image data representing the image. The device 110/system(s) 120 selects (134) a first component configured to obscure the input image data for a particular task, where the first component is configured using at least second image data associated with an output label corresponding to the particular task. The first component may be referred to as a masker component herein (e.g., 210, 310). As described below in detail, the masker component may be configured to mask image data for a particular task, so that portions irrelevant for the task are obscured. For example, for hair-based classification, the first component may select irrelevant pixels as those that do not represent hair in the image, and obscure these pixels to generate masked image data. [0030] Various machine learning techniques may be used to train and operate such models. Models may be trained and operated according to various machine learning techniques. Such techniques may include, for example, neural networks (such as deep learning neural networks, convolutional neural networks, and/or recurrent neural networks), inference engines, trained classifiers, etc. Examples of trained classifiers include Support Vector Machines (SVMs), neural networks, probabilistic graphs, decision trees, AdaBoost (short for “Adaptive Boosting”) combined with decision trees, and random forests. [0027] The device 110/system(s) 120 processes (140) the masked image data using a second component that is configured to perform the task corresponding to the first component used to generate the masked image data. For example, the second component may be configured to classify image data based on the color of the person's hair.)
It would have been obvious to one of ordinary skill in the art prior to the effective filing of the application to modify the teachings of [M in view of Liu] to include [hiding information that is not relevant to the task being performed on the data] as is taught by [Yadav].
The suggestion/motivation for doing so is to [improve task performance [0001]].
Regarding claims 12 the claims inherit the same rejections as claim 2 above for reciting similar limitations in the form of an apparatus claim, M teaches a system. (0090-0092)
Claim(s) 3 and 13 is/are rejected under 35 U.S.C. 103 as being unpatentable over Mosallanezhad et al. (US 20220374606 A1) hereinafter M in view of Liu et al (US 20190281024 A1) in view of Yadav et al. (US 20210406589 A1) in view of Ando et al. (US 20240289615 A1).
Regarding claim 3, M teaches the method of Claim 1, and is disclosed above, M does not explicitly teach but Liu further teaches wherein the performing of the task-aware information hiding comprises: inputting the container (C) and the information (I) into the information hiding AI/ML model to produce another container (digital host information such as text, digitized sound, images, video signals, and the like) with hidden information (secret information)(C’’) and recovered information (digital host information) (I’’); ([0034] Information hiding, which is also known as data hiding, is an emerging technology field that combines multidisciplinary theories and technologies. Information hiding technology mainly refers to the embedding of information into digital host information such as text, digitized sound, images, video signals, and the like. The purpose of information hiding is not to limit the correct information storing, reading, and accessing, but to ensure that the hidden information does not attract the attention and concern of the monitor, thereby reducing the possibility of being attacked. The information hiding may include hidden technology, visual cryptography technology, and watermarking technology. The information hiding in this application may be the hidden technology. That is, the secret information is embedded in the information that looks ordinary to be sent to prevent the third party from detecting the secret information. The particular information processed by the information hiding method may be regarded as hidden information.)
It would have been obvious to one of ordinary skill in the art prior to the effective filing of the application to modify the teachings of [M] to include [generate a host container with hidden data through a network] as is taught by [Liu].
The suggestion/motivation for doing so is to [improve secure transmission [0002-0004]].
M in view of Liu does not explicitly teach inputting the another container with hidden information into an AI/ML task on the container to produce a first task output as a task output on the container;
inputting the recovered information into an AI/ML task on the information to produce a second task output as a task output on the information;
comparing the first task output with a target output of task on the container to calculate a first loss;
comparing the second task output with a target output of task on the information to calculate a second loss;
and combining the first loss and the second loss to produce a joint loss.
In an analogous art Yadav teaches inputting the another container with hidden information (image with masked data) into an AI/ML task (second component classifier) on the container to produce a first task output (classification) as a task output on the container; ((0024: The device 110/system(s) 120 receives (132) input image data representing the image. The device 110/system(s) 120 selects (134) a first component configured to obscure the input image data for a particular task, where the first component is configured using at least second image data associated with an output label corresponding to the particular task. The first component may be referred to as a masker component herein (e.g., 210, 310). As described below in detail, the masker component may be configured to mask image data for a particular task, so that portions irrelevant for the task are obscured. For example, for hair-based classification, the first component may select irrelevant pixels as those that do not represent hair in the image, and obscure these pixels to generate masked image data. [0030] Various machine learning techniques may be used to train and operate such models. Models may be trained and operated according to various machine learning techniques. Such techniques may include, for example, neural networks (such as deep learning neural networks, convolutional neural networks, and/or recurrent neural networks), inference engines, trained classifiers, etc. Examples of trained classifiers include Support Vector Machines (SVMs), neural networks, probabilistic graphs, decision trees, AdaBoost (short for “Adaptive Boosting”) combined with decision trees, and random forests. [0027] The device 110/system(s) 120 processes (140) the masked image data using a second component that is configured to perform the task corresponding to the first component used to generate the masked image data. For example, the second component may be configured to classify image data based on the color of the person's hair.)
inputting the recovered information (Fig 6: input image 610) into an AI/ML task on the information to produce a second task (Fig 6: second task is the adversarial masker 210 different from method 2 output 630) output as a task output (output 650) on the information; (0078: The example images 650 are generated by the masker component 210 processing the input images 610, where the masker component 210 is trained in an adversarial manner as described above in connection with FIG. 2. The masked images 640, 650 generated using the masker component 310 and 210 may result in better hair-based classification outcomes, while reducing the amount of data/pixels used by the classification task component. The block based method may block out more pixels related to the person's face compared to the masker components 210, 310, however, the masked images 630 may result in poor hair-based classification outcomes because relevant pixels are also blocked. The Gaussian blur based method may result in comparative hair-based classification outcomes as the masker component 310/210, but the masked images 620 reveal more pixels than necessary for the classification task.)
comparing the first task output with a target output of task on the container (0033; an input may be sent to the neural network and compared with the associated output to determine how the neural network's performance compares to a target performance.)
comparing the second task output with a target output of task on the information (0033; an input may be sent to the neural network and compared with the associated output to determine how the neural network's performance compares to a target performance.)
It would have been obvious to one of ordinary skill in the art prior to the effective filing of the application to modify the teachings of [M in view of Liu] to include [applying different ML/AI tasks of the data] as is taught by [Yadav].
The suggestion/motivation for doing so is to [improve task performance [0001]].
M in view of Liu in view of Yadav do not explicitly teach the underlined comparing the first task output with a target output of task on the container to calculate a first loss;
comparing the second task output with a target output of task on the information to calculate a second loss;
and combining the first loss and the second loss to produce a joint loss.
In an analogous art Ando teaches comparing the first task output (each output) with a target output (correct answer) of task on the container to calculate a first loss; (calculating a loss based on the comparison) ([0007] compare the plurality of output data with a plurality of pieces of correct answer information associated with the plurality of training data, to calculate a loss value for each of the plurality of output data; select, among the plurality of output data, relevant output data, the loss value for which meets a predetermined reference, and irrelevant output data, the loss value for which does not meet the predetermined reference; and create processed correct answer information by processing the correct answer information compared with the relevant output data, compare the relevant output data with the processed correct answer information, to output a processed loss value, and update the neural network by using the processed loss value, or create processed training data by processing the training data associated with the relevant output data, input the processed training data into the neural network, to cause the neural network to output processed output data obtained as a result of classifying the processed training data, compare the processed output data with the correct answer information associated with the relevant output data to output a processed loss value, and update the neural network by using the processed loss value. [0039] In contrast, in the present embodiment, the training loss outputted from the training loss calculation unit 3 is supplied to the correct answer information processing unit 4 (also referred to as a teaching data correction unit). The correct answer information processing unit 4 is configured to compare the output data with a plurality of pieces of correct answer information associated with the training data, to thereby calculate the loss value (training loss) for each of the output data. Then, the correct answer information processing unit 4 is configured to select, among the output data, relevant output data, the loss value for which meets a predetermined reference, and irrelevant output data, the loss value for which does not meet the predetermined reference.)
comparing the second task output with a target output of task on the information to calculate a second loss; (calculating a loss based on the comparison) ([0007] compare the plurality of output data with a plurality of pieces of correct answer information associated with the plurality of training data, to calculate a loss value for each of the plurality of output data; select, among the plurality of output data, relevant output data, the loss value for which meets a predetermined reference, and irrelevant output data, the loss value for which does not meet the predetermined reference; and create processed correct answer information by processing the correct answer information compared with the relevant output data, compare the relevant output data with the processed correct answer information, to output a processed loss value, and update the neural network by using the processed loss value, or create processed training data by processing the training data associated with the relevant output data, input the processed training data into the neural network, to cause the neural network to output processed output data obtained as a result of classifying the processed training data, compare the processed output data with the correct answer information associated with the relevant output data to output a processed loss value, and update the neural network by using the processed loss value. [0039] In contrast, in the present embodiment, the training loss outputted from the training loss calculation unit 3 is supplied to the correct answer information processing unit 4 (also referred to as a teaching data correction unit). The correct answer information processing unit 4 is configured to compare the output data with a plurality of pieces of correct answer information associated with the training data, to thereby calculate the loss value (training loss) for each of the output data. Then, the correct answer information processing unit 4 is configured to select, among the output data, relevant output data, the loss value for which meets a predetermined reference, and irrelevant output data, the loss value for which does not meet the predetermined reference.)
and combining the first loss and the second loss to produce a joint loss. ([0061] FIG. 7 shows the training loss obtained by the training loss recalculation. In the example shown in FIG. 7, the correct answer information for the image part P4a of the image P4 has been changed to “unknown”, the training loss value varies to the relatively small value (0.7). The training loss recalculation unit 5 outputs the calculated training loss values to the neural network 2. As shown in FIG. 8, the update unit 14 of the neural network 2 updates the parameters of the neural network 2 based on the inputted training loss values, by using the SGD method, for example (S9). [0068] he image processing unit 9 as a teaching data correction unit compares each of the training losses (loss values) acquired from the training loss calculation unit 3 with a predetermined threshold, to determine whether the training loss exceeds the predetermined threshold. In other words, the image processing unit 9 selects, among the output data, relevant output data, the loss value for which meets the predetermined reference, and irrelevant output data, the loss value for which does not meet the predetermined reference (the training loss is equal to or smaller than the predetermined threshold). Note that the image processing unit 9 may select, among the output data, the output data within a predetermined number in an order starting from the one, the loss value for which is the largest, as the relevant output data.)
It would have been obvious to one of ordinary skill in the art prior to the effective filing of the application to modify the teachings of [M in view of Liu in view of Yadav] to include [comparing outputs of ML/AI models and generating loss values] as is taught by [Ando].
The suggestion/motivation for doing so is to [improve AI processing of image data [0002-0006].
Regarding claims 13 the claims inherit the same rejections as claim 3 above for reciting similar limitations in the form of an apparatus claim, M teaches a system. (0090-0092)
Conclusion
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ABDERRAHMEN H CHOUAT whose telephone number is (571)431-0695. The examiner can normally be reached on Mon-Fri from 9AM to 5PM.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Christopher Parry, can be reached at telephone number 571-272-8328. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from Patent Center. Status information for published applications may be obtained from Patent Center. Status information for unpublished applications is available through Patent Center to authorized users only. Should you have questions about access to the USPTO patent electronic filing system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free).
Examiner interviews are available via a variety of formats. See MPEP § 713.01. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) Form at https://www.uspto.gov/InterviewPractice.
Abderrahmen Chouat
Examiner
Art Unit 2451
/Chris Parry/Supervisory Patent Examiner, Art Unit 2451
Prosecution Insights