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 .
indication.
Claim Rejections – 35 U.S.C. §103
2. 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.
3. The factual inquiries set forth in Graham v John Deere Co., 383 U.S. 1, 148 USPQ
459 (1966), that are applied for establishing a background for determining obviousness
under 35 U.S.C. 103 are summarized as follows:
a. Determining the scope and contents of the prior art
b. Ascertaining the differences between the prior art and the claims at issue
c. Resolving the level of ordinary skill in the pertinent art
d. Considering objective evidence present in the application indicating
obviousness or nonobviousness
Claims 1, 3, 4, 7, 9, 10 and 13 are rejected under 35 U.S.C. 103 as being unpatentable over Cheruvu et al. (United States Patent Publication Number 2021/0390447), in view of Wei et al. (United States Patent Publication Number
2017/0017407), hereinafter referred to as Wei.
Regarding claim 1 Cheruvu teaches a system (Fig. 1, (100) system [0005], [0022]) for collaborative generative artificial intelligence content identification and verification, comprising: a plurality of computing devices (Fig. 1, (110) content generation platform [0022], (130) content consumer platform [0022]) each comprising at least a processor, (Fig. 1, (111) CPU. (131) CPU [0023]) a memory, (Fig. 1, (118) storage memory device, (138) storage memory device [0035]) and a network interface; (a network interface 670 [0077]) wherein a plurality of programming instructions (The instruction sets 614 [0088], [0107]) stored in one or more of the memories (may be loaded, stored, or otherwise retained in system memory 640, in whole or in part, [0088]) and operating on one or more of the processors (one or more processor cores to process instructions which, when executed [0025]) (during execution by the processor cores 618 and/or graphics processor circuitry 612. [0088]) of the plurality of computing devices (Fig. 1, (110) content generation platform [0022], (130) content consumer platform [0022])causes the plurality of computing devices to: (Fig. 1, (110) content generation platform [0022], (130) content consumer platform [0022]) build an artificial-intelligence-generated content registry by: (Fig. 1, (120) Shared A.I. Registry [0022][0033][0036][0050] and [0064]) receiving generated content from a generating service, (provides for consumption of content generated by a content generation platform, such as content generated by processing of the ML data via ML model 116 at content generation platform 110. [0028])wherein the generated content from the generating service is created using a generative artificial intelligence system; ( In one implementation, the AI-generated output may refer to the content generated from processing the ML data via an inference stage of the ML model 116 at the content
generation platform 110. [0029]) assigning a service identifier (The shared AI registry 120 maintains the record of the ML models' GUIDs indexed by model IDs 125.
[0036]) to the generated content, (implementations of the disclosure include the GUID as part of a digital signature that is sent with the content to the content consumer platform 130. [0031])wherein the service identifier (model IDs 125 [0036]) is associated with the generating service; (In implementations of the disclosure, the content generation platform utilizes a global unique identifier (GUID) to authenticate and validate the output of the ML model 116 by content generator 112. The GUID authenticates an identity of the ML model 116. [0031]) (ML models' GUIDs indexed by model IDs 125. [0036]) to assign a segment identifier (digital signature [0034]) based on the content (based on [0033]) of the respective data segment; (multi-modal content [0030]) Linking the segment identifier with the service identifier; (GUID as part of the digital signature [0031]; GUIDs indexed by model IDs 125 [0036]) and adding the segment identifier (digital signature [0034]) to a content group, (record of the ML models' GUIDs indexed by model IDs 125 [0036]) such as “content group” wherein the content group (record of the ML models' GUIDs indexed by model IDs 125 [0036]) such as “content group” comprises a plurality of segment identifiers (digital signature with GUIDs … the digital signature includes a fusion of the generated content with model identification (including the GUID) [0031]) for the generated content; (generated content [0031]) and storing the content group (record of the ML models' GUIDs indexed by model IDs 125 [0036]) such as “content group” in a database; (the shared AI registry 260 is the same as shared AI registry 120 described with respect to FIG. 1 and maintains a data store 265 of ML models' GUIDs indexed by model IDs, [0036]) such as “database” and verify (Fig. 5, (530) Verify the extracted GUID against data obtained from a shared registry, the data obtained from the shared registry comprising identifying information of the ML model including the GUID [0072]) whether subsequent content (Fig. 5, (510) content generated by a machine learning (ML) model and a digital signature corresponding to the content [0072]) was generated by the generating service by: (Fig. 5, (510 ML Model [0072]) receiving subsequent content from a generating service purported to be the generating service; (Fig. 5, (510) Receive, by a processor of a content consumer platform, content generated by a machine learning (ML) model and a digital signature corresponding to the content [0072]) comparing the hashes of the selected plurality of data segments with hashes stored in the content group in the database; (Fig. 5, (530) Verify the extracted GUID against data obtained from a shared registry, the data obtained from the shared registry comprising identifying information of the ML model including the GUID [0072]) and producing a verification score (Fig. 5, (540) the processor determine whether the extracted GUID is successfully verified, “YES”, “NO” [0073]) such as “verification score” based on the comparison of the hashes indicating a likelihood that the subsequent content was generated by the same generating service as the generated content (Fig. 5 (540) Extracted GUID successfully verified? []) (550) if “YES” Provide the content for consumption at the content consumer platform and indicating that the content is generated by the ML model having verified authenticity [0074], (560) If “NO” Refuse the content for consumption at the content consumer platform and indicate that the ML model generating the content is not verified for authenticity [0074])
Cheruvu does not fully disclose deconstructing the generated content into a plurality of data segments; for each data segment: using a hashing algorithm deconstructing the subsequent content into a plurality of data segments; hashing the selected plurality of data segments; selecting a scope of sampling for the subsequent content; randomly selecting a plurality of data segments of the subsequent content from within the scope;
Wei teaches deconstructing the generated content (data object [0024]) into a plurality of data segments; (Fig. 2, (23) Divide the data object by using a fixed length [0048]) (24) Perform sampling and compression on data of each block [0049]), (25) Aggregate consecutive blocks with a same compression ratio characteristic into a data segment according to the chunking policy mapping table [0055]) for each data segment: using a hashing algorithm deconstructing the subsequent content into a plurality of data segments; (Fig. 2 (26) Split each data segment into chunking subsequences of a specific expected length according to the chunking policy mapping table [0051] and (27) Splice chunking subsequences of neighboring data segments [0052]) hashing the selected plurality of data segments; (Fig. 2, (28) Calculate a hash fingerprint of each chunk [0052]) selecting a scope of sampling (A data object is divided into seven blocks: blocks 1 to 7. By performing sampling on each block … blocks that belong to the same compression ratio range are aggregated into one data segment, [0029]) for the subsequent content; (Fig. 2, (24) Perform sampling and compression on data of each block [0049]) randomly selecting a plurality of data segments of the subsequent content(the blocks are aggregated into a data segment according to a sample compression ratio of eachblock, [0022]) from within the scope; (A data object is divided into seven blocks: blocks 1 to 7. By performing sampling on each block … blocks that belong to the same compression ratio range are aggregated into one data segment, [0029])
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Cheruvu to incorporate the teachings of Wei whereby deconstructing the generated content into a plurality of data segments; for each data segment: using a hashing algorithm deconstructing the subsequent content into a plurality of data segments; hashing the selected plurality of data segments; selecting a scope of sampling for the subsequent content; randomly selecting a plurality of data segments of the subsequent content from within the scope. By doing so more than one segmentation manner can be used. Wei [0044]
Claims 7 and 13 correspond to claim 1 and are rejected accordingly
Regarding claim 3 Cheruvu in view of Wei teaches the system of claim 1,
Cheruvu as modified further teaches wherein the generated content is multimedia content (The generated content may be multi-modal content, including,
but not limited to, one or more of plaintext, image(s), video(s), audio, and/or any other form of content [0028])
Claim 9 corresponds to claim 3 and is rejected accordingly
Regarding claim 4 Cheruvu in view of Wei teaches the system of claim 1,
Cheruvu as modified does not fully disclose wherein the hashing algorithm is a perceptual hashing algorithm.
Wei teaches wherein the hashing algorithm (a rapid hash algorithm [0057]) is a perceptual hashing algorithm (calculate a hash (Hash) value of a data chunk as a fingerprint. [0042], [0057])
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Cheruvu to incorporate the teachings of Wei wherein the hashing algorithm is a perceptual hashing algorithm. By doing so by using the fingerprint, whether each of the data chunks is stored in a storage device is determined. Wei [0041]
Claim 10 corresponds to claim 4 and is rejected accordingly
Conclusion
4. The prior art made of record and not relied upon is considered pertinent to
applicant's disclosure.
Mirowski et al., (United States Patent Publication Number 2007/0014435) teaches “FIG. l(B) illustrates a workflow that embodies a computer-based methodology for the automatic generation and the validation of training images and multipoint geostatistical simulations based thereon in accordance with the present invention. [0040]”
5. Any inquiry concerning this communication or earlier communications from the
examiner should be directed to Kweku Halm whose telephone number is (469) 295-
9144. The examiner can normally be reached on 7:30AM - 5:30PM Mon - Thur. If
attempts to reach the examiner by telephone are unsuccessful, the examiner's
supervisor, Sanjiv Shah can be reached on (571) 272-4098. 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 the
Patent Application Information Retrieval (PAIR) system. Status information for published
applications may be obtained from either Private PAIR or Public PAIR. Status information
for unpublished applications is available through Private PAIR only. For more
information about the PAIR system, see http://pair-direct.uspto.gov. Should you have
questions on access to the Private PAIR system, contact the Electronic Business Center
(EBC) at 866-217-9197 (toll-free).
/KWEKU WILLIAM HALM/Examiner, Art Unit 2166
/SANJIV SHAH/Supervisory Patent Examiner, Art Unit 2166