DETAILED ACTION
Claim(s) 1 and 16 and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation:
Claim(s) 2,3,8 and 17,18,19 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation applied in claims 1,16,20 further in view of FROLOVA et al. (US 2021/0081754 A1):
Claim(s) 5 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III with in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation applied in claims 1,16,20 further in view of FROLOVA et al. (US 2021/0081754 A1) as applied in claims 2,3,8 and 17,18,19 further in view of Haacke (US 2021/0275676 A1):
Claim(s) 6 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation applied in claims 1,16,20 further in view of FROLOVA et al. (US 2021/0081754 A1) as applied in claims 2,3,8 and 17,18,19 further in view of MASAJIRO et al. (JP 2021-043603 A) with SEARCH machine translation:
Claim(s) 7 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation applied in claims 1,16,20 further in view of FROLOVA et al. (US 2021/0081754 A1) as applied in claims 2,3,8 and 17,18,19 further in view of MASAJIRO et al. (JP 2021-043603 A) with SEARCH machine translation as applied in claim 6 further in view of SUN et al. (CN 114675742 B) with SEARCH machine translation:
Claim(s) 9,10 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation applied in claims 1,16,20 further in view of FROLOVA et al. (US 2021/0081754 A1) as applied in claims 2,3,8 and 17,18,19 further in view of further in view of LIU (CN 105809146 A) with SEARCH machine translation:
Claim(s) 11 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation applied in claims 1,16,20 further in view of FROLOVA et al. (US 2021/0081754 A1) as applied in claims 2,3,8 and 17,18,19 further in view of SUN et al. (CN 113449691 A), referred to as SUN II, with SEARCH machine translation:
Claim(s) 12 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation applied in claims 1,16,20 further in view of FROLOVA et al. (US 2021/0081754 A1) as applied in claims 2,3,8 and 17,18,19 further in view of HE at al. (US 2022/0036059 A1):
Claim(s) 13 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation applied in claims 1,16,20 further in view of CAO et al. (CN 111670357 A) with SEARCH machine translation:
Claim(s) 14 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation applied in claims 1,16,20 further in view of CAO et al. (CN 111670357 A) with SEARCH machine translation as applied in claim 13 further in view of RIM et al. (US 2019/0221313 A1):
Claim(s) 15 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation applied in claims 1,16,20 further in view of CAO et al. (CN 111670357 A) with SEARCH machine translation as applied in claim 13 further in view of RIM et al. (US 2019/0221313 A1) as applied in claim 14 further in view of Clapper (US 2003/0107584 A1):
Claim(s) 21 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation applied in claims 1,16,20 further in view of CHENG (CN 108647089 A) with SEARCH machine translation:
Response to Amendment
The claim/specification amendment was received 2/11/2026. Claim 4 cancel. Claims 1,2,3,5,6,7,8,9,10,11,12,13,14,15 and 16,17,18,19,21 and 20 pending.
PNG
media_image1.png
1608
420
media_image1.png
Greyscale
Response to Arguments
Applicant's arguments filed 5/7/2026, pages 8-14 have been fully considered but they are not persuasive..
II. Rejections of Claims 1,16 and 20 Under 35 USC 103
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 12, 2nd para, last S:
“preventing1 overfitting by2 removing3 ‘important data’”)
are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
In contrast claim 1 says:
“the exclusion4 of the at least one first object reducing5 overfitting…wherein the at least one first object has at least one feature importance…greater than or equal to a threshold6”.
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 identically7 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 and 16 and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation:
PNG
media_image2.png
1614
669
media_image2.png
Greyscale
Re 1., YUAN teaches A method performed by an electronic device, the electronic device (in the sealed box to be opened) including memory and a processor having processing circuitry, the method comprising:
acquiring,8 by the electronic device (still in the sealed box),9 a first image (“frame identified as the first type”, pg. 11) corresponding to a first scene identifier10 (via “the target sub-scene11 identifier”, pg. 16, 1st txt blk);
by using a first artificial intelligence (AI) model (i.e., “an artificial intelligence
(Artificial Intelligence, AI) set in a virtual scene battle through training”, pg. 5: fig. 4: a battle) and the first image (“frame identified as the first type”, pg. 11), identifying, by the processor of the electronic device (still in the box), at least one first (“terminal”, pg. 9, 2nd txt blk) object of the first image contributing to (“type”, pg. 11, 3rd txt blk) classification (“and the type of the first video frame is output”, pg. 27, 5th txt blk) of the first image (“frame identified as the first type”, pg. 11) as the first scene identifier (via “the target sub-scene12 identifier”, pg. 16, 1st txt blk); and
generating (via “the first terminal Render”, pg. 13 last txt blk), by the processor of the electronic device (still in the box), training (“battle”, pg. 5, 4th txt blk) data (via identify texture data via “judge13…a target virtual object14 is15 displayed16 in the video frame”, pg. 12, wherein “displayed in the video frame” (comprising texture data) represents being or being equivalent to the target virtual object: fig. 4) for the first (battle-training) AI model, by performing first (“input” “stage of processing” via said “displayed in the video frame”) processing for17 excluding the at least one first object of the first image (“frame identified as the first type”, pg. 11, via fig. 4) , the exclusion of the at least one first object reducing overfitting for the first AI model for the at least one first object,
wherein the at least one first object has at least one feature18 importance192021 associated with the first scene identifier greater than or equal to a threshold feature importance22 (or likewise importances of distinct features {i.e., distinct feature importances} greater than a threshold via “the first type in the probability23 distribution24… being greater than or equal to the probability threshold”, pg. 23, last para, penult S), and
wherein the reduction of overfitting for the first AI model for the at least one first object increases a variety of objects used by the first Al model to identify the first scene identifier.
PNG
media_image3.png
935
871
media_image3.png
Greyscale
YUAN does not teach under the full scope (and thus not limited to an improper narrow subset of the full scope) of the broadest reasonable interpretation the difference25 a), b) of claim 1 of:
a) training (data)26…27
b)28 excluding (the at least one first object of the first image)29…30the exclusion…reducing overfitting…
{(feature) importance…(feature) importance…}31…
wherein the reduction of overfitting…increases a variety of objects used by (the first Al model).
Lissi teaches the difference32 a) of claim 1 of:
a) training (“for said AI model” [0014]) (data).
Since YUAN suggests other possibilities for training pg. 22, last txt blk to page 23, 1st txt blk:
In a possible implementation manner, the second server inputs the first video frame into the image recognition model. The second server performs feature extraction and classification on the first video frame through the image recognition model, and outputs the type of the first video frame.
Wherein, the image recognition model is trained based on a set of sample video frames. The sample video frame set includes positive sample videoframes and negative sample video frames. The positive sample videoframes are video frames showing target virtual objects, and the negative sample video frames are Video frames with target dummy not shown. The image recognition model trained by using the sample video frame set has the ability to judge whether a target virtual object is displayed in the videoframe. In some embodiments, the image recognition model is trained by the second server in advance, and the image recognition model can be directly used to determine the type of the video frame during the game.
one of skill in the art of training would of looked to the other possibilities (i.e., teachings) for training and thus make YUAN’s be as Lissi’s predictably recognizing the change “can be utilized dynamically during gameplay to assist in selecting automatically different camera angles that are the best for the specific type of content being experienced by a gamer.”, Lissi [0074], via “explicit…creative steps…or even routine steps”33 (TRAINING STEPS: a); b); c); d); e)):
TRAINING STEPS
a) create an artificial intelligence training program based on Lissi’s fig. 3’s 220: “AI Model Training Engine”:
PNG
media_image4.png
1219
919
media_image4.png
Greyscale
b) install/run the training engine program into YUAN’s server computer 150:
PNG
media_image5.png
1067
897
media_image5.png
Greyscale
c) inputting YUAN’s “positive” and “negative” images into the program, YUAN page 22, last para:
Wherein, the image recognition model is trained based on a set of sample video frames. The sample video frame set includes positive sample videoframes and negative sample video frames. The positive sample videoframes are video frames showing target virtual objects, and the negative sample video frames are Video frames with target dummy not shown. The image recognition model trained by using the sample video frame set has the ability to judge whether a target virtual object is displayed in the videoframe. In some embodiments, the image recognition model is trained by the second server in advance, and the image recognition model can be directly used to determine the type of the video frame during the game.
c1) make the positive negative images be as Lissi’s fig. 3:302a: 206: “Video”;
d) obtain the “Camera Angle AI Model” (Lissi’s fig. 3:350) from the program;
e) make the Camera Angle AI Model 350 be as YUAN’s image recognition model in YUAN’s teachings of the image recognition model.
YUAN of the combination of YUAN,Lissi does not teach the remaining difference (b) of claim 1:
b)34 excluding (the at least one first object of the first image)35…36 the exclusion…reducing overfitting…
{(feature) importance…(feature) importance…}37
wherein the reduction of overfitting…increases a variety of objects used by (the first Al model).
Sjolund teaches the remaining difference (b) of claim 1:
b)38 excluding (“features” [0130] last S) (the at least one first object of the first image)39…40 the (feature) exclusion…reducing overfitting (“to less overfitting” [0131]) …
{(feature) importance…(feature) importance…}41
wherein the (less) reduction of overfitting…increases a variety (“such that the classifier learned can be more generalized and lead to less overfitting” [0131]) of (“image” [0073] 4th S) objects (“features” [0073] 4th S) used by (a “separated optimizer” [0146], 5th S: fig. 14) (the first Al model).
Since YUAN of the combination of YUAN,Lissi suggests other possibilities for training, YUAN, pg. 22, last txt blk to page 23, 1st txt blk:
In a possible implementation manner, the second server inputs the first video frame into the image recognition model. The second server performs feature extraction and classification on the first video frame through the image recognition model, and outputs the type of the first video frame.
Wherein, the image recognition model is trained based on a set of sample video frames. The sample video frame set includes positive sample videoframes and negative sample video frames. The positive sample videoframes are video frames showing target virtual objects, and the negative sample video frames are Video frames with target dummy not shown. The image recognition model trained by using the sample video frame set has the ability to judge whether a target virtual object is displayed in the videoframe. In some embodiments, the image recognition model is trained by the second server in advance, and the image recognition model can be directly used to determine the type of the video frame during the game.
one of skill in the art of training would of looked to the other possibilities (i.e., teachings) for training and thus make YUAN’s of the combination of YUAN,Lissi be as Sjolund’s seeing in the change that “models provide improved results for a variety of imaging processing purposes such as reconstruction, segmentation, and other image processing aspects which may have missing or incomplete data or modeling.”, Sjolund [0009] last S, via additional explicit creative steps or even routine steps (IMAGE PRE-PROCESSING STEPS: a); b); c); c1) ; d):
IMAGE PRE-PROCESSING STEPS (phase 1):
a) create an image adjustment program based on Sjolund’s fig. 5:
PNG
media_image6.png
1330
904
media_image6.png
Greyscale
b) install/run the image adjustment program in YUAN’s of the combination of YUAN,Lissi server 150:
PNG
media_image5.png
1067
897
media_image5.png
Greyscale
c) inputting YUAN’s “positive” and “negative” images into the image adjustment program, YUAN page 22, last para:
Wherein, the image recognition model is trained based on a set of sample video frames. The sample video frame set includes positive sample videoframes and negative sample video frames. The positive sample videoframes are video frames showing target virtual objects, and the negative sample video frames are Video frames with target dummy not shown. The image recognition model trained by using the sample video frame set has the ability to judge whether a target virtual object is displayed in the videoframe. In some embodiments, the image recognition model is trained by the second server in advance, and the image recognition model can be directly used to determine the type of the video frame during the game.
c1) make the positive negative images be as Sjolund’s fig. 5:510: “OBTAINING TRAINING IMAGAING DATA”;
d) use the adjusted positive and negative training images for the TRAINING STEPS phase 2: a),b), c), d), e) below:
TRAINING STEPS (phase 2):
a) create an artificial intelligence training program based on Lissi’s fig. 3’s 220: “AI Model Training Engine”:
PNG
media_image4.png
1219
919
media_image4.png
Greyscale
b) install/run the training engine program into YUAN’s server computer 150:
PNG
media_image5.png
1067
897
media_image5.png
Greyscale
c) inputting YUAN’s “positive” and “negative” images (as adjusted) into the program, YUAN page 22, last para:
Wherein, the image recognition model is trained based on a set of sample video frames. The sample video frame set includes positive sample videoframes and negative sample video frames. The positive sample videoframes are video frames showing target virtual objects, and the negative sample video frames are Video frames with target dummy not shown. The image recognition model trained by using the sample video frame set has the ability to judge whether a target virtual object is displayed in the videoframe. In some embodiments, the image recognition model is trained by the second server in advance, and the image recognition model can be directly used to determine the type of the video frame during the game.
c1) make the positive negative images (as adjusted) be as Lissi’s fig. 3:302a: 206: “Video”;
d) obtain the “Camera Angle AI Model” (Lissi’s fig. 3:350) from the program;
e) make the Camera Angle AI Model 350 be as YUAN’s image recognition model in YUAN’s teachings of the image recognition model.
Thus in view of the above YUAN of the combination of YUAN,Lissi,Sjolund makes obvious claim 1 under the broadest reasonable interpretation.
YUAN of the combination of YUAN,Lissi,Sjolund does not teach the “improper… narrow subset of claim scope”42 via application of the ipsissimis verbis test of the broadest reasonable interpretation of the last difference43 of claim 1 of:
{(feature) importance44…(feature) importance…}45.
PNG
media_image7.png
1320
1031
media_image7.png
Greyscale
HUANG teaches the “improper… narrow subset of claim scope” via application of the ipsissimis verbis test of the broadest reasonable interpretation of the last difference of claim 1 of:
feature importance…feature importance (or likewise “feature filter wherein the feature importance degree of each image feature by using feature visualization technology, recognition the feature importance is greater than the preset threshold value of the image feature is the target feature;”, pg. txt blk).
Since YUAN of the combination of YUAN,Lissi,Sjolund suggests other possibilities for training, YUAN, pg. 22, last txt blk to page 23, 1st txt blk:
In a possible implementation manner, the second server inputs the first video frame into the image recognition model. The second server performs feature extraction and classification on the first video frame through the image recognition model, and outputs the type of the first video frame.
Wherein, the image recognition model is trained based on a set of sample video frames. The sample video frame set includes positive sample videoframes and negative sample video frames. The positive sample videoframes are video frames showing target virtual objects, and the negative sample video frames are Video frames with target dummy not shown. The image recognition model trained by using the sample video frame set has the ability to judge whether a target virtual object is displayed in the videoframe. In some embodiments, the image recognition model is trained by the second server in advance, and the image recognition model can be directly used to determine the type of the video frame during the game.
one of skill in the art of training would of looked to the other possibilities (i.e., teachings) for training and thus make YUAN’s of the combination of YUAN,Lissi,Sjolund be as HUANG’s seeing in the change improved training via “targeted image expansion, which is good for improving the effect of image expansion” “to train the model with good effect”, HUANG, pg. 2, 2nd tx blk, page 6, last txt blk, via additional “explicit…creative steps”46 (IMAGE PRE-PROCESSING STEPS: a2); b); c2) and d) and TRAINING STEPS a1); a2); and a3):
IMAGE PRE-PROCESSING STEPS a,b,c,d (phase 1):
a1) create an image adjustment program based on Sjolund’s fig. 5:
PNG
media_image6.png
1330
904
media_image6.png
Greyscale
a2) create a feature importance image extension program of HUANG’s fig. 1:
PNG
media_image8.png
878
1023
media_image8.png
Greyscale
b) install/run Sjolund’s image adjustment program and HUANG’s feature importance image extension program in YUAN’s of the combination of YUAN,Lissi server 150:
PNG
media_image5.png
1067
897
media_image5.png
Greyscale
c1) inputting YUAN’s “positive” and “negative” images into the image adjustment program, YUAN page 22, last para:
Wherein, the image recognition model is trained based on a set of sample video frames. The sample video frame set includes positive sample videoframes and negative sample video frames. The positive sample videoframes are video frames showing target virtual objects, and the negative sample video frames are Video frames with target dummy not shown. The image recognition model trained by using the sample video frame set has the ability to judge whether a target virtual object is displayed in the videoframe. In some embodiments, the image recognition model is trained by the second server in advance, and the image recognition model can be directly used to determine the type of the video frame during the game.
c1.1) make the positive negative images be as Sjolund’s fig. 5:510: “OBTAINING TRAINING IMAGAING DATA”;
c2) inputting the adjusted positive and negative training images into HUANG’s feature importance image extension program; and
d) use the adjusted and extended positive and negative training images for the TRAINING STEPS phase: a),b), c), d), e) below:
TRAINING STEPS a,b,c,d,e (phase 2):
a) create an artificial intelligence training program based on Lissi’s fig. 3’s 220: “AI Model Training Engine”:
PNG
media_image4.png
1219
919
media_image4.png
Greyscale
a1) make Lissi’s fig. 3:305: “Video Feature Extractor” be as Sjolund’s “feature detection47” or “Image analysis” or Sjolund’s figure 12 (feature extraction in the hidden/latent layer “Z”) or fig. 13 (feature extraction in the hidden/latent layer “Z” plus segmentation) or figure 14 (feature extraction in the hidden/latent layer “Z” plus segmentation plus fusion: each encoder/decoder in figures 12,13,14 is detailed in figure 9 (encoder) and figure 10 (decoder)):
--[0073] Image analysis is a general term describing the extraction of useful information from images. It includes feature detection, image description, segmentation, recognition, classification and more. Objects are often detected based on their physical characteristics. Discontinuities in image brightness due to difference in depth, surface orientation and structure provide distinctive features. Shape, color and size are some common features that are used to describe objects in the image. In segmentation, thresholding is a common low-level technique that detects objects in accordance to the pixels intensities. Further, there are methods that consider spatial information such Normalized Cuts, Split-and-Merge, and Mean-Shift algorithm. Feature selection for these algorithms is therefore paramount.--:
PNG
media_image9.png
1652
1168
media_image9.png
Greyscale
PNG
media_image10.png
1292
895
media_image10.png
Greyscale
a2) make Lissi’s fig. 3:306: “Video Feature Classifiers” be as Sjolund’s “classification” or “Image analysis” via Sjolund’s [0073]:1st & 2nd Ss; [0157]:1st S:
[0073] Image analysis is a general term describing the extraction of useful information from images. It includes feature detection, image description, segmentation, recognition, classification and more. Objects are often detected based on their physical characteristics. Discontinuities in image brightness due to difference in depth, surface orientation and structure provide distinctive features. Shape, color and size are some common features that are used to describe objects in the image. In segmentation, thresholding is a common low-level technique that detects objects in accordance to the pixels intensities. Further, there are methods that consider spatial information such Normalized Cuts, Split-and-Merge, and Mean-Shift algorithm. Feature selection for these algorithms is therefore paramount.
[0157] The flowchart 1600 continues with operations that perform an image processing operation (e.g., to produce an inference, prediction, classification, etc.) base don a result produced with the trained neural network model (operation 1640). In an example, this image processing operation is used as part of at least one of segmentation, denoising, synthesis, classification, regression, or reconstruction. The flowchart 1600 concludes with specific optional operations to decode and reconstruct the encoded imaging data (e.g., with a decoder model) (operation 1650).
a3) create a neural network model training program for training either of Sjolund’s models in figures 12,13,14 (reproduced above) based on Sjolund’s flowchart of fig. 15:
PNG
media_image11.png
1270
853
media_image11.png
Greyscale
b) install/run the training neural network engine program into YUAN’s server computer 150:
PNG
media_image5.png
1067
897
media_image5.png
Greyscale
c) inputting YUAN’s pre-processed “positive” and “negative” images (as adjusted and extended) into the neural network training engine program, YUAN page 22, last para:
Wherein, the image recognition model is trained based on a set of sample video frames. The sample video frame set includes positive sample videoframes and negative sample video frames. The positive sample videoframes are video frames showing target virtual objects, and the negative sample video frames are Video frames with target dummy not shown. The image recognition model trained by using the sample video frame set has the ability to judge whether a target virtual object is displayed in the videoframe. In some embodiments, the image recognition model is trained by the second server in advance, and the image recognition model can be directly used to determine the type of the video frame during the game.
c1) make the pre-processed positive negative images (as adjusted and extended) be as Lissi’s fig. 3:302a: 206: “Video”;
d) obtain the “Camera Angle AI Model” (Lissi’s fig. 3:350) from the neural network engine program; and
e) make the Camera Angle AI Model 350 be as YUAN’s image recognition model in YUAN’s teachings of the image recognition model.
Thus claim 1 is rendered obvious under the improper narrow subset via application of the ipsissimis verbis test of claim scope of the broadest reasonable interpretation.
Claim 16 is rejected like claim 1:
16. (Currently Amended) An electronic (“computer”, YUAN, pg. 33, 5th txt blk) device48 comprising:
memory storing one or more computer programs including computer-executable instructions; and one or more processors,
wherein the computer-executable instructions, when executed by the one or more processors , cause the electronic device to:
acquire a first image corresponding to a first scene identifier,
by using a first artificial intelligence (AI) model and the first image, identify at least one first object of the first image contributing to classification of the first image as the first scene identifier, and
generate training data for the first AI model by performing first processing for excluding the at least one first object of the first image (via rejection of claim 1), the exclusion of the at least one first object reducing overfitting for the first AI model the at least one first object, wherein the at least one first object has at least one feature importance associated with the first scene identifier greater than or equal to a threshold feature importance, and
wherein the reduction of overfitting for the first AI model for the at least one first object increases a variety of objects used by the first AI model to identify the first scene identifier.
Claim 20 is rejected like claims 1 and 16:
20. (Currently Amended) One or more non-transitory computer-readable storage media49 (“including a computer program” YUAN, pg. 33, 4th txt blk) storing one or more computer programs including computer-executable instructions that, when executed by one or more processors of an electronic device, cause the electronic device to perform operations, the operations50 comprising:
acquiring, by the electronic device, a first image corresponding to a first scene identifier;
by using a first artificial intelligence (AI) model and the first image, identifying, by the electronic device, at least one first object of the first image contributing to classification of the first image as the first scene identifier; and
generating, by the electronic device, training data for the first AI model by performing first processing for excluding the at least one first object of the first image
, the exclusion of the at least one first object reducing overfitting for the first Al model for the at least one first object, wherein the at least one first object has at least one feature importance associated with the first scene identifier greater than or equal to a threshold feature importance, and
wherein the reduction of overfitting for the first AI model for the at least one first object increases a variety of objects used by the first AI model to identify the first scene identifier.
Claim(s) 2,3,8 and 17,18,19 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation applied in claims 1,16,20 further in view of FROLOVA et al. (US 2021/0081754 A1):
PNG
media_image12.png
1612
1109
media_image12.png
Greyscale
Re 2. (Previously Presented), YUAN of the combination (illustrated above) of YUAN,Lissi,Sjolund,HUANG teaches The method of claim 1, wherein the identifying of the at least one first object by using the first AI model (i.e., “an artificial intelligence (Artificial Intelligence, AI) set in a virtual scene battle through training”, pg. 5: fig. 4: a battle) and the first image (“frame identified as the first type”, pg. 11) comprises:
identifying an activation (feature) map (“of the first video frame”, pg. 13, 2nd txt blk or “of multiple image blocks”, pg. 13, 4th txt blk) corresponding51 to the first image (“frame identified as the first type”, pg. 11); and
based52 on the activation (feature) map (“of the first video frame”, pg. 13, 2nd txt blk or “of multiple image blocks”, pg. 13, 4th txt blk), identifying the at least one first object.
YUAN of the combination of YUAN,Lissi,Sjolund,HUANG does not teach the difference of claim 2:
“identifying…activation (map)53…54
based on the activation (map)…”.
FROLOVA teaches the difference of claim 2:
identifying…activation (“most correlated with the newly received image” [0030] 4th S: fig. 4: “COMPARE ACTIVATION MAP(S)”) (map)55…56
(“the referenced set 150A can be compared to sets associated with reference image(s) 170” [0081] 1st S: fig. 1) based on the activation (map)….
Since YUAN of the combination of YUAN,Lissi,Sjolund, HUANG suggests other possibilities for training, YUAN, pg. 22, last txt blk to page 23, 1st txt blk:
In a possible implementation manner, the second server inputs the first video frame into the image recognition model. The second server performs feature extraction and classification on the first video frame through the image recognition model, and outputs the type of the first video frame.
Wherein, the image recognition model is trained based on a set of sample video frames. The sample video frame set includes positive sample videoframes and negative sample video frames. The positive sample videoframes are video frames showing target virtual objects, and the negative sample video frames are Video frames with target dummy not shown. The image recognition model trained by using the sample video frame set has the ability to judge whether a target virtual object is displayed in the videoframe. In some embodiments, the image recognition model is trained by the second server in advance, and the image recognition model can be directly used to determine the type of the video frame during the game.
one of skill in the art of training would of looked to the other possibilities (i.e., teachings) for training and thus make YUAN’s of the combination of YUAN,Lissi,Sjolund, HUANG be as FROLOVA’s predictably recognizing the change “can enable the system to learn and improve from data based on its statistical characteristics rather on predefined rules of human experts”, FROLOVA [0014], via additional explicit creative steps (TRAINING STEPS: a1.1) & a1.2):
IMAGE PRE-PROCESSING STEPS a,b,c,d (phase 1):
a1) create an image adjustment program based on Sjolund’s fig. 5:
PNG
media_image6.png
1330
904
media_image6.png
Greyscale
a2) create a feature importance image extension program of HUANG’s fig. 1:
PNG
media_image8.png
878
1023
media_image8.png
Greyscale
b) install/run Sjolund’s image adjustment program and HUANG’s feature importance image extension program in YUAN’s of the combination of YUAN,Lissi server 150:
PNG
media_image5.png
1067
897
media_image5.png
Greyscale
c1) inputting YUAN’s “positive” and “negative” images into the image adjustment program, YUAN page 22, last para:
Wherein, the image recognition model is trained based on a set of sample video frames. The sample video frame set includes positive sample videoframes and negative sample video frames. The positive sample videoframes are video frames showing target virtual objects, and the negative sample video frames are Video frames with target dummy not shown. The image recognition model trained by using the sample video frame set has the ability to judge whether a target virtual object is displayed in the videoframe. In some embodiments, the image recognition model is trained by the second server in advance, and the image recognition model can be directly used to determine the type of the video frame during the game.
c1.1) make the positive negative images be as Sjolund’s fig. 5:510: “OBTAINING TRAINING IMAGAING DATA”;
c2) inputting the adjusted positive and negative training images into HUANG’s feature importance image extension program; and
d) use the adjusted and extended positive and negative training images for the TRAINING STEPS phase: a),b), c), d), e) below:
TRAINING STEPS a,b,c,d,e (phase 2):
a) create an artificial intelligence training program based on Lissi’s fig. 3’s 220: “AI Model Training Engine”:
PNG
media_image4.png
1219
919
media_image4.png
Greyscale
a1) make Lissi’s fig. 3:305: “Video Feature Extractor” be as Sjolund’s “feature detection57” or “Image analysis” or Sjolund’s figure 12 (feature extraction in the hidden/latent layer “Z”) or fig. 13 (feature extraction in the hidden/latent layer “Z” plus segmentation) or figure 14 (feature extraction in the hidden/latent layer “Z” plus segmentation plus fusion: each encoder/decoder in figures 12,13,14 is detailed in figure 9 (encoder) and figure 10 (decoder)):
--[0073] Image analysis is a general term describing the extraction of useful information from images. It includes feature detection, image description, segmentation, recognition, classification and more. Objects are often detected based on their physical characteristics. Discontinuities in image brightness due to difference in depth, surface orientation and structure provide distinctive features. Shape, color and size are some common features that are used to describe objects in the image. In segmentation, thresholding is a common low-level technique that detects objects in accordance to the pixels intensities. Further, there are methods that consider spatial information such Normalized Cuts, Split-and-Merge, and Mean-Shift algorithm. Feature selection for these algorithms is therefore paramount.--:
PNG
media_image9.png
1652
1168
media_image9.png
Greyscale
PNG
media_image10.png
1292
895
media_image10.png
Greyscale
a1.1) make Sjolund’s “activation layer” (i.e., “LeakyRelu”, Sjolund: [0139] 3rd S, in Sjolund’s fig.9: “BLOCK 1”, reproduce above) in each encoder be as FROLOVA’s fig. 1:142A: “LAYERS” connecting to a set of activation maps, reproduced below:
[0139] FIGS. 9 and 10 illustrate data processing operations performed by neural networks of a respective encoder and decoder, employed in an exemplary encoding process for generating and using a latent representation of an imaging input. The encoder, shown in FIG. 9, includes a configuration that is similar to the U-Net but shallower, with only two downsample steps (compared to U-Net's four downsample steps). The output of the encoder shown in FIG. 9 is the latent representation with 16 channels. Instead of standard ReLU, LeakyRelu may be used as the activation layer in the encoders in order to prevent the optimization from getting trapped in local minima by preserving negative gradients. The input and output of the encoder were of same dimension in width and height. The number of channels of the encoder output may be, for example, 16, which produces produced good results while keeping the model small. In a similar fashion as an implementation of VAE, each model may be encoded by individual encoder before latent representation fusion.
PNG
media_image13.png
799
1129
media_image13.png
Greyscale
a1.2) output FROLOVA’s activation map from Sjolund’s fig. 9: “BLOCK 1”: “Leaky ReLU” to Sjolund’s fig. 9: “CONV 3 x 3 x 32” for the rest of the encoding operations of Sjound’s fig. 9;
a2) make Lissi’s fig. 3:306: “Video Feature Classifiers” be as Sjolund’s “classification” or “Image analysis” via Sjolund’s [0073]:1st & 2nd Ss; [0157]:1st S:
[0073] Image analysis is a general term describing the extraction of useful information from images. It includes feature detection, image description, segmentation, recognition, classification and more. Objects are often detected based on their physical characteristics. Discontinuities in image brightness due to difference in depth, surface orientation and structure provide distinctive features. Shape, color and size are some common features that are used to describe objects in the image. In segmentation, thresholding is a common low-level technique that detects objects in accordance to the pixels intensities. Further, there are methods that consider spatial information such Normalized Cuts, Split-and-Merge, and Mean-Shift algorithm. Feature selection for these algorithms is therefore paramount.
[0157] The flowchart 1600 continues with operations that perform an image processing operation (e.g., to produce an inference, prediction, classification, etc.) base don a result produced with the trained neural network model (operation 1640). In an example, this image processing operation is used as part of at least one of segmentation, denoising, synthesis, classification, regression, or reconstruction. The flowchart 1600 concludes with specific optional operations to decode and reconstruct the encoded imaging data (e.g., with a decoder model) (operation 1650).
a3) create a neural network model training program for training either of Sjolund’s models in figures 12,13,14 (reproduced above) based on Sjolund’s flowchart of fig. 15:
PNG
media_image11.png
1270
853
media_image11.png
Greyscale
b) install/run the training neural network engine program into YUAN’s server computer 150:
PNG
media_image5.png
1067
897
media_image5.png
Greyscale
c) inputting YUAN’s pre-processed “positive” and “negative” images (as adjusted and extended) into the neural network training engine program, YUAN page 22, last para:
Wherein, the image recognition model is trained based on a set of sample video frames. The sample video frame set includes positive sample videoframes and negative sample video frames. The positive sample videoframes are video frames showing target virtual objects, and the negative sample video frames are Video frames with target dummy not shown. The image recognition model trained by using the sample video frame set has the ability to judge whether a target virtual object is displayed in the videoframe. In some embodiments, the image recognition model is trained by the second server in advance, and the image recognition model can be directly used to determine the type of the video frame during the game.
c1) make the pre-processed positive negative images (as adjusted and extended) be as Lissi’s fig. 3:302a: 206: “Video”;
d) obtain the “Camera Angle AI Model” (Lissi’s fig. 3:350) from the neural network engine program; and
e) make the Camera Angle AI Model 350 be as YUAN’s image recognition model in YUAN’s teachings of the image recognition model.
Re 3. (Previously Presented) , YUAN of the combination (illustrated above) of YUAN,Lissi,Sjolund,HUANG,FROLOVA teaches The method of claim 2, wherein the identifying of the at least one first object, based on the activation map (which “can enable the system to learn and improve from data based on its statistical characteristics rather on predefined rules of human experts”, FROLOVA [0014]), comprises:
identifying (“as a candidate for modification in the CNN”, FROLOVA [0053] penult S) at least one area (via reflected “activation maps that correspond58 to various…regions…of the image” FROLOVA [0011] 7th S), in (via “reflected” “data”, FROLOVA [0093] 3rd S) which (“in which” is referring any of: (1) claim 3’s “the activation map”; (2) claim 3’s “at least one image area”; and (3) claim 1’s “the at least one first object”) a feature importance59 (or likewise said YUAN’s probability distribution) satisfies a designated first condition (or likewise said YUAN’s probability threshold), in (via “reflected” “data”, FROLOVA [0093] 2nd S) the activation map (“most correlated with the newly received image” FROLOVA: [0030] 4th S: fig. 4: “COMPARE ACTIVATION MAP(S)”); and
identifying the at least one first object, which corresponds to the at least one area (via “activation maps… reflecting …regions” [0011] 7th S) in (via “reflected” “data”, FROLOVA [0093] 3rd S) the activation map (“most correlated with the newly received image” FROLOVA: [0030] 4th S: fig. 4: “COMPARE ACTIVATION MAP(S)”), of the first image (“frame identified as the first type”, pg. 11).
Re 8. (Previously Presented), YUAN of the combination (illustrated above) of YUAN,Lissi,Sjolund,HUANG, FROLOVA teaches The method of claim 2, wherein the identifying of the at least one first object further comprises:
identifying at least one area (via “activation maps…correspond to various … regions” FROLOVA [0011] 7th S), in which (i.e., “activation maps…correspond to various … regions” FROLOVA [0011] 7th S) a feature importance60 (i.e., “similarity”61 [0053] 2nd S of a combination of two similar features/importances/characteristics/significances) satisfies a designated first condition (via “One or more criteria (e.g., a threshold)”. FROLOVA [0053] 2nd S), in the activation map (“most correlated with the newly received image” FROLOVA [0030] 4th S: fig. 4: “COMPARE ACTIVATION MAP(S)”);
determining that the at least one area (via “activation maps…correspond to various…regions” FROLOVA [0011] 7th S) of the activation map (“most correlated with the newly received image” FROLOVA [0030] 4th S: fig. 4: “COMPARE ACTIVATION MAP(S)”) satisfies at least one second condition (via “One or more criteria (e.g., a threshold)”. FROLOVA [0053] 2nd S); and
identifying the at least one first object of the first image (“frame identified as the first type”, YUAN pg. 11), which (i.e., “detecting”62 “a” “first” “area”, YUAN pg. 11, 3rd txt blk) corresponds to (via the above illustrated combination) the at least one area (via “activation maps…correspond to various…regions” FROLOVA [0011] 7th S) based63 on
the identifying of the at least one area (via “activation maps…correspond to various…regions”, FROLOVA [0011] 7th S, “most correlated with the newly received image” FROLOVA [0030] 4th S: fig. 4: “COMPARE ACTIVATION MAP(S)”) and
the determining that the at least one area (via “activation maps… correspond to various…regions” FROLOVA [0011] 7th S) of the activation map (“most correlated with the newly received image” FROLOVA [0030] 4th S: fig. 4: “COMPARE ACTIVATION MAP(S)”) satisfies the at least one second condition (via “One or more criteria (e.g., a threshold)”. FROLOVA [0053] 2nd S).
Claim 17 is rejected like claim 2:
17. (Previously Presented) The electronic device of claim 16, wherein the one or more computer programs further comprise computer-executable instructions to, as at least a part of the identifying of the at least one first object by using the first AI model and the first image:
identify an activation map corresponding to the first image, and
based on the activation map, identify the at least one first object (via the rejection of claim 2:
Claim 18 is rejected like claim 3:
Re 18. (Previously Presented), YUAN of the combination of YUAN,Lissi,Sjolund, HUANG, FROLOVA teaches The electronic device of claim 17, wherein the one or more computer programs further comprise computer-executable instructions to, as at least a part of the identifying of the at least one first object, based on the activation map:
identify at least one area, in which64 a feature importance satisfies a designated first condition, in the activation map, and
identify the at least one first object, which corresponds to the at least one area in the activation map, of the first image.
Claim 19 is rejected like claim 3:
Re 19. (Original), YUAN of the combination of YUAN,Lissi,Sjolund,HUANG, FROLOVA teaches The electronic device of claim 18, wherein the one or more computer programs further comprise computer-executable instructions to,
as at least a part of the identifying of the at least one area (“as a candidate for modification in the CNN”, FROLOVA [0053] penult S), in (via “reflected” “data”, FROLOVA [0093] 3rd S) which (i.e., any the activation map or image region or the first object) the feature importance satisfies the designated first condition, in the activation map,
identify the at least one area (“as a candidate for modification in the CNN”, FROLOVA [0053] penult S) in which65 the feature importance is equal to or greater than a designated threshold feature importance.
Claim(s) 5 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III with in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation applied in claims 1,16,20 further in view of FROLOVA et al. (US 2021/0081754 A1) as applied in claims 2,3,8 and 17,18,19 further in view of Haacke (US 2021/0275676 A1):
PNG
media_image14.png
1612
1123
media_image14.png
Greyscale
Re 5. (Currently Amended), YUAN of the combination (illustrated above) of YUAN,Lissi,Sjolund,HUANG,FROLOVA teaches under the broadest reasonable interpretation66 The method of claim 3 , wherein the identifying of the at least one area (“as a candidate for modification in the CNN”, FROLOVA [0053] penult S), in (via “reflected” “data”, FROLOVA [0093] 3rd S) which {the claimed “in which” is referring to any67 of:
(1) claim 3’s “the activation map”;
(2) claim 3’s “at least one image area”;
(3) claim 1’s “the at least one first object”}68
the feature importance (or likewise said YUAN’s said probability distribution under the broadest reasonable interpretation and thus not under the improper narrow subset of claim scope of the broadest reasonable interpretation) is equal to or greater than the
blank-processing (via “Such a corrected set can be provided for processing”, FROLOVA [0030] 7th S, of activation maps) remaining areas (via reflected “activation maps that correspond69 to various…regions…of the image” FROLOVA [0011] 7th S), in (via “reflected” “data”, FROLOVA [0093] 3rd S) which (either the activation map or image area or first object) the70 feature importance (“to improve the model prediction precision”, LIU, page 9, last txt blk) is less than the 71 (or likewise “In response to the probability being less than the probability threshold… the second server determines the first video frame as the second type, that is, the first video frame does not display the target virtual object”, YUAN, page 23, 3rd txt blk, last sentence to page 23, last para, last S that continues on page 24, 1st txt blk, wherein “the probability” is referring to “in the probability distribution”, YUAN, page 23, last para, penult S, and “the probability threshold” is referring to “the probability threshold” in page 23, last para, penult S), in the activation map (via reflected “activation maps that correspond72 to various…regions…of the image” FROLOVA [0011] 7th S);
configuring (“to use the template image”, YUAN pg. 27, 4th txt blk) at least one contour (comprised by a “template73 image”, YUAN pg. 11, 3rd txt blk) for the blank-processed (via “Such a corrected set can be provided for processing”, FROLOVA [0030] 7th S, of activation maps) remaining areas (via reflected “activation maps that correspond74 to various…regions…of the image” FROLOVA [0011] 7th S) in (via “reflected” “data”, FROLOVA [0093] 3rd S) the activation map (“most correlated with the newly received image” FROLOVA: [0030] 4th S: fig. 4: “COMPARE ACTIVATION MAP(S)”); and
(“The second server acquires the similarity75 between the template image of the target virtual object and multiple regions on the first video frame”, YUAN pg. 12, 2nd txt blk) based on the at least one contour (comprised by a “template76 image”, YUAN pg. 11, 3rd txt blk), identifying the at least one area (i.e., “detecting”77 “a” “first” “area”, YUAN pg. 11, 3rd txt blk) of the activation map (“most correlated with the newly received image” FROLOVA: [0030] 4th S: fig. 4: “COMPARE ACTIVATION MAP(S)”).
YUAN of the combination of YUAN,Lissi,HUANG,FROLOVA does not teach under the broadest reasonable interpretation the difference of claim 5 of:
blank-(processing)78…79remaining (areas)…
blank-(processed) remaining (areas).
Haacke teaches the remaining difference of claim 5:
(“left” [0072] 5th S) blank-(“normalized” [0075]: fig. 1:109: “Generate Output”) (processing)80…81 (“left” [0072] 5th S) remaining (“blank” [0072] 5th S) (areas)…
blank-(normalized: via fig. 1: 101: “Receive Time Resolved MR Data”) (processed) (“left” [0072] 5th S) remaining (“blank” [0072] 5th S: represented as fig. 1:109: “Generate Output”) (areas).
Since YUAN of the combination of YUAN,Lissi,HUANG, FROLOVA suggests selecting “color value similarity or the gray value similarity” via page 20, last para continued on page 21, 1st para, 3rd S:
For example, the second server uses the template image of the target virtual object to slide on the first video frame, and obtains the similarity between the template image of the target virtual object and multiple regions on the first video frame, and the multiple regions That is, the area covered when the template image of the target virtual object slides on the first video frame. In some embodiments, when the second server determines the similarity between the template image of the target virtual object and multiple regions, it can be realized by using the color value similarity or the gray value similarity, which is not discussed in this embodiment of the present application. Do limited. In response to the existence of an area among the plurality of areas whose similarity with the template image of the target virtual object is greater than or equal to a similarity threshold, the second server determines that the area matches the template image of the target virtual object, and sends the first video The frame is determined to be of the first type, that is, the first video frame displays the target virtual object. In response to the similarity between the plurality of regions and the template image of the target virtual object being less than a similarity threshold, the second server determines that there is no region matching the template image of the target virtual object in the first video frame, and the second server A video frame is determined to be of the second type, that is, the first video frame does not display the target virtual object.
and since FROLOVA suggests selecting “similarity metric” [0052], 2nd S:
[0052] Having identified a set within repository 160 as being most correlated to the set generated with respect to the received input, a degree or measure of similarity between respective activation maps from such sets can be computed. For example, having identified set 150B as being most closely correlated to set 150A, a Pearson correlation coefficient (PCC) (or any other such similarity metric) can be computed with respect to the respective activation maps from such sets. In certain implementations, such a metric can reflect a value between −1 and 1 (with zero reflecting no correlation, 1 reflecting a perfect correlation, and −1 reflecting negative correlation).
, one of skill in the art of image similarity would of looked to other teachings of similarity for selecting a similarity and thus can make YUAN’s of the combination of YUAN,Lissi, HUANG, FROLOVA be as Haacke’s predictably recognizing the change “enhances82 the SNR83 in the output images”, Haacke [0103] 2nd S, “to provide excellent structural details both in the original images and in the TSMs”, Haacke [0094] 1st S, via additional creative explicit or routine steps (TRAINING STEPS: a1.11); a1.12); a1.13)):
IMAGE PRE-PROCESSING STEPS a,b,c,d (phase 1):
a1) create an image adjustment program based on Sjolund’s fig. 5:
PNG
media_image6.png
1330
904
media_image6.png
Greyscale
a2) create a feature importance image extension program of HUANG’s fig. 1:
PNG
media_image8.png
878
1023
media_image8.png
Greyscale
b) install/run Sjolund’s image adjustment program and HUANG’s feature importance image extension program in YUAN’s of the combination of YUAN,Lissi server 150:
PNG
media_image5.png
1067
897
media_image5.png
Greyscale
c1) inputting YUAN’s “positive” and “negative” images into the image adjustment program, YUAN page 22, last para:
Wherein, the image recognition model is trained based on a set of sample video frames. The sample video frame set includes positive sample videoframes and negative sample video frames. The positive sample videoframes are video frames showing target virtual objects, and the negative sample video frames are Video frames with target dummy not shown. The image recognition model trained by using the sample video frame set has the ability to judge whether a target virtual object is displayed in the videoframe. In some embodiments, the image recognition model is trained by the second server in advance, and the image recognition model can be directly used to determine the type of the video frame during the game.
c1.1) make the positive negative images be as Sjolund’s fig. 5:510: “OBTAINING TRAINING IMAGAING DATA”;
c2) inputting the adjusted positive and negative training images into HUANG’s feature importance image extension program; and
d) use the adjusted and extended positive and negative training images for the TRAINING STEPS phase: a),b), c), d), e) below:
TRAINING STEPS a,b,c,d,e (phase 2):
a) create an artificial intelligence training program based on Lissi’s fig. 3’s 220: “AI Model Training Engine”:
PNG
media_image4.png
1219
919
media_image4.png
Greyscale
a1) make Lissi’s fig. 3:305: “Video Feature Extractor” be as Sjolund’s “feature detection84” or “Image analysis” or Sjolund’s figure 12 (feature extraction in the hidden/latent layer “Z”) or fig. 13 (feature extraction in the hidden/latent layer “Z” plus segmentation) or figure 14 (feature extraction in the hidden/latent layer “Z” plus segmentation plus fusion: each encoder/decoder in figures 12,13,14 is detailed in figure 9 (encoder) and figure 10 (decoder)):
--[0073] Image analysis is a general term describing the extraction of useful information from images. It includes feature detection, image description, segmentation, recognition, classification and more. Objects are often detected based on their physical characteristics. Discontinuities in image brightness due to difference in depth, surface orientation and structure provide distinctive features. Shape, color and size are some common features that are used to describe objects in the image. In segmentation, thresholding is a common low-level technique that detects objects in accordance to the pixels intensities. Further, there are methods that consider spatial information such Normalized Cuts, Split-and-Merge, and Mean-Shift algorithm. Feature selection for these algorithms is therefore paramount.--:
PNG
media_image9.png
1652
1168
media_image9.png
Greyscale
PNG
media_image10.png
1292
895
media_image10.png
Greyscale
a1.1) make Sjolund’s “activation layer” (i.e., “LeakyRelu”, Sjolund: [0139] 3rd S, in Sjolund’s fig.9: “BLOCK 1”, reproduce above) in each encoder be as FROLOVA’s fig. 1:142A: “LAYERS” connecting to a set of activation maps, reproduced below:
[0139] FIGS. 9 and 10 illustrate data processing operations performed by neural networks of a respective encoder and decoder, employed in an exemplary encoding process for generating and using a latent representation of an imaging input. The encoder, shown in FIG. 9, includes a configuration that is similar to the U-Net but shallower, with only two downsample steps (compared to U-Net's four downsample steps). The output of the encoder shown in FIG. 9 is the latent representation with 16 channels. Instead of standard ReLU, LeakyRelu may be used as the activation layer in the encoders in order to prevent the optimization from getting trapped in local minima by preserving negative gradients. The input and output of the encoder were of same dimension in width and height. The number of channels of the encoder output may be, for example, 16, which produces produced good results while keeping the model small. In a similar fashion as an implementation of VAE, each model may be encoded by individual encoder before latent representation fusion.
PNG
media_image13.png
799
1129
media_image13.png
Greyscale
a1.11) make FROLOVA’s activation maps sets fig. 1:150A,150B be as Haacke’s “region” [0072] 2nd S;
a1.12) set FROLOVA’s “similarity”-“criteria (e.g., a threshold)”, [0053] 2nd S, as Haacke’s “threshold level”, Haacke [0072] 2nd S;
a.1.13 threshold/blank the activation map sets based on the set “threshold level” of Haacke;
a1.2) output FROLOVA’s activation map from Sjolund’s fig. 9: “BLOCK 1”: “Leaky ReLU” to Sjolund’s fig. 9: “CONV 3 x 3 x 32” for the rest of the encoding operations of Sjound’s fig. 9;
a2) make Lissi’s fig. 3:306: “Video Feature Classifiers” be as Sjolund’s “classification” or “Image analysis” via Sjolund’s [0073]:1st & 2nd Ss; [0157]:1st S:
[0073] Image analysis is a general term describing the extraction of useful information from images. It includes feature detection, image description, segmentation, recognition, classification and more. Objects are often detected based on their physical characteristics. Discontinuities in image brightness due to difference in depth, surface orientation and structure provide distinctive features. Shape, color and size are some common features that are used to describe objects in the image. In segmentation, thresholding is a common low-level technique that detects objects in accordance to the pixels intensities. Further, there are methods that consider spatial information such Normalized Cuts, Split-and-Merge, and Mean-Shift algorithm. Feature selection for these algorithms is therefore paramount.
[0157] The flowchart 1600 continues with operations that perform an image processing operation (e.g., to produce an inference, prediction, classification, etc.) base don a result produced with the trained neural network model (operation 1640). In an example, this image processing operation is used as part of at least one of segmentation, denoising, synthesis, classification, regression, or reconstruction. The flowchart 1600 concludes with specific optional operations to decode and reconstruct the encoded imaging data (e.g., with a decoder model) (operation 1650).
a3) create a neural network model training program for training either of Sjolund’s models in figures 12,13,14 (reproduced above) based on Sjolund’s flowchart of fig. 15:
PNG
media_image11.png
1270
853
media_image11.png
Greyscale
b) install/run the training neural network engine program into YUAN’s server computer 150:
PNG
media_image5.png
1067
897
media_image5.png
Greyscale
c) inputting YUAN’s pre-processed “positive” and “negative” images (as adjusted and extended) into the neural network training engine program, YUAN page 22, last para:
Wherein, the image recognition model is trained based on a set of sample video frames. The sample video frame set includes positive sample videoframes and negative sample video frames. The positive sample videoframes are video frames showing target virtual objects, and the negative sample video frames are Video frames with target dummy not shown. The image recognition model trained by using the sample video frame set has the ability to judge whether a target virtual object is displayed in the videoframe. In some embodiments, the image recognition model is trained by the second server in advance, and the image recognition model can be directly used to determine the type of the video frame during the game.
c1) make the pre-processed positive negative images (as adjusted and extended) be as Lissi’s fig. 3:302a: 206: “Video”;
d) obtain the “Camera Angle AI Model” (Lissi’s fig. 3:350) from the neural network engine program; and
e) make the Camera Angle AI Model 350 be as YUAN’s image recognition model in YUAN’s teachings of the image recognition model.
Claim(s) 6 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation applied in claims 1,16,20 further in view of FROLOVA et al. (US 2021/0081754 A1) as applied in claims 2,3,8 and 17,18,19 further in view of MASAJIRO et al. (JP 2021-043603 A) with SEARCH machine translation:
PNG
media_image15.png
1612
1123
media_image15.png
Greyscale
Re 6. (Original), YUAN of the combination (illustrated above) of YUAN,Lissi,Sjolund,HUANG,FROLOVA teaches The method of claim 2, further comprising:
based on an output layer (illustrated above) of the AI model (i.e., “an artificial intelligence (Artificial Intelligence, AI) set in a virtual scene battle through training”, pg. 5: fig. 4: a battle) and at least one feature map (“of the first video frame”, YUAN, pg. 13, 2nd txt blk) of the AI model (i.e., “an artificial intelligence (Artificial Intelligence, AI) set in a virtual scene battle through training”, pg. 5: fig. 4: a battle: a first video frame), identifying at least one contribution degree; and
based85 on the at least one contribution degree and the at least one feature map (“of the first video frame”, YUAN, pg. 13, 2nd txt blk), identifying (via detecting/ determining etc..) multiple feature importances86 of the activation map (which “can enable the system to learn and improve from data based on its statistical characteristics rather on predefined rules of human experts”, FROLOVA [0014]).
YUAN of the combination (illustrated above) of YUAN,Lissi,Sjolund,FROLOVA does not teach the difference of claim 2:
“at least one contribution degree; and
based87 on the at least one contribution degree and…
multiple feature importances88”.
MASAJIRO teaches the difference of claim 6:
at least one contribution degree (“calculated for each grid in the image”, pg. 3, 4th txt blk); and
based89 on the at least one contribution degree and (“based on the feature map”, pg. 3, 4th txt blk) …(identifying) multiple feature importances90 (“higher than the predetermined value as the useful area”, pg. 12, 1st txt blk: fig. 2:AR11: t-shirt:
PNG
media_image16.png
725
1086
media_image16.png
Greyscale
Since YUAN of the combination (illustrated above) of YUAN,Lissi,Sjolund, HUANG, FROLOVA teaches recognition, one of skill in the art of recognition can make YUAN’s of the combination (illustrated above) of YUAN,Lissi,Sjolund,HUANG, FROLOVA be as MASAJIRO’s predictably recognizing the change “improving the recognition rate”, MASAJIRO, pg. 11, 9th txt blk.
Claim(s) 7 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation applied in claims 1,16,20 further in view of FROLOVA et al. (US 2021/0081754 A1) as applied in claims 2,3,8 and 17,18,19 further in view of MASAJIRO et al. (JP 2021-043603 A) with SEARCH machine translation as applied in claim 6 further in view of SUN et al. (CN 114675742 B) with SEARCH machine translation:
PNG
media_image17.png
1611
1235
media_image17.png
Greyscale
Re 7. (Previously Presented), YUAN of the combination (illustrated above) of YUAN,Lissi,FROLOVA, MASAJIRO teaches The method of claim 6, wherein the identifying of the at least one contribution degree (“calculated for each grid in the image”, MASAJIRO pg. 3, 4th txt blk) is performed based
1 9yc on Equation 1, where Equation 1 is ag == Yi dy, i kK
wherein aᶜ is the contribution degree, C is a class of the output laver of the AI model. k is an index of the at least one feature map. Z is a product of a row and a column of a matrix of the at least one feature map, i is an i-th element in the matrix, i is a i-th element in the matrix, V° is an output layer, and Akii is at least one feature map
wherein the identifying of the multiple feature importances of the activation map is identified based on Equation 2, where Equation 2 is L° = ReLU Y, a€A*, and
wherein Lc is an activation map, c is a class of the output layer of the AI model, k is an index of the at least one feature map, z is a product of a row and a column of a matrix of the at least one feature map, 1 is an i-th element in the matrix, and j is a j-th element in the matrix.
YUAN of the combination (illustrated above) of YUAN, Lissi, FROLOVA, MASAJIRO does not teach the difference of claim 7:
equations 1 and 2.
SUN teaches equations 1 and 2 at [0206][0213] and corresponding machine translation pg 7, text block:
PNG
media_image18.png
1431
1050
media_image18.png
Greyscale
Since MASAJIRO of the combination (illustrated above) of YUAN,Lissi, Sjolund,HUANG,FROLOVA, MASAJIRO teaches a contribution degree, one of skill in the art of contribution degrees can make MASAJIRO’s of the combination (illustrated above) of YUAN,Lissi,Sjolund,HUANG, FROLOVA, MASAJIRO be as SUN’s predictably recognizing the change “can effectively distinguish the related and non-related pixels and extract the characteristic of the classification result with higher contribution degree”, SUN, pg. 7, text block.
Claim(s) 9,10 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation applied in claims 1,16,20 further in view of FROLOVA et al. (US 2021/0081754 A1) as applied in claims 2,3,8 and 17,18,19 further in view of further in view of LIU (CN 105809146 A) with SEARCH machine translation:
PNG
media_image19.png
1611
1235
media_image19.png
Greyscale
Re 9. (Original), FROLOVA of the combination (illustrated above) of YUAN, Lissi, Sjolund,HUANG,FROLOVA teaches The method of claim 8, wherein the determining that the at least one area (via “activation maps…correspond to various … regions” FROLOVA [0011] 7th S) of the activation map (“most correlated with the newly received image” [0030] 4th S: fig. 4: “COMPARE ACTIVATION MAP(S)”) satisfies at least one second condition (via “One or more criteria (e.g., a threshold)”. FROLOVA [0053] 2nd S) comprises:
determining that a size of the at least one area (via “activation maps… correspond to various … regions” FROLOVA [0011] 7th S) of the activation map (“most correlated with the newly received image” [0030] 4th S: fig. 4: “COMPARE ACTIVATION MAP(S)”) is equal to or larger than a designated threshold (via “One or more criteria (e.g., a threshold)”. FROLOVA [0053] 2nd S) size.
FROLOVA of the combination (illustrated above) of YUAN,Lissi,Sjolund,HUANG, FROLOVA does not teach the difference of claim 9 of:
“a size…a…size”.
LIU teaches the difference of claim 9:
a size (or “activation parameter91…is 0.9”, pg. 10, 1st txt blk)…
a (designated threshold) size (“is 0.8”, pg. 10).
Since FROLOVA of the combination (illustrated above) of YUAN,Lissi, FROLOVA teaches a threshold, one of skill in the art of thresholds can make FROLOVA’s of the combination (illustrated above) of YUAN,Lissi,Sjolund,HUANG, FROLOVA be as LIU’s predictably recognizing the change “to improve the applicability of the scene recognition”, LIU, pg. 12, 5th txt blk
Claim 10 is rejected like claim 9:
Re 10. (Original), YUAN,Lissi,Sjolund,HUANG, FROLOVA, LIU teaches The method of claim 8, wherein the determining that the at least one area of the activation map satisfies at least one second condition comprises:
determining that a number of the at least one area of the activation map is equal to or more than a designated threshold number.
Claim(s) 11 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation applied in claims 1,16,20 further in view of FROLOVA et al. (US 2021/0081754 A1) as applied in claims 2,3,8 and 17,18,19 further in view of SUN et al. (CN 113449691 A), referred to as SUN II, with SEARCH machine translation:
PNG
media_image20.png
1611
1235
media_image20.png
Greyscale
Re 11. (Original), FROLOVA of the combination (illustrated above) of YUAN,Lissi, FROLOVA The method of claim 8, wherein the determining that the at least one area (via “activation maps…correspond to various … regions” FROLOVA [0011] 7th S) of the activation map (“most correlated with the newly received image” [0030] 4th S: fig. 4: “COMPARE ACTIVATION MAP(S)”) satisfies at least one second condition (via “One or more criteria (e.g., a threshold)”. FROLOVA [0053] 2nd S) comprises:
determining that an overlapping degree between the at least one area (via “activation maps…correspond to various … regions” FROLOVA [0011] 7th S) of the activation map (“most correlated with the newly received image” [0030] 4th S: fig. 4: “COMPARE ACTIVATION MAP(S)”) is equal to or less than a threshold (via “One or more criteria (e.g., a threshold)”, FROLOVA [0053] 2nd S) overlapping degree.
FROLOVA of the combination (illustrated above) of YUAN,Lissi,Sjolund,HUANG, FROLOVA does not teach the difference of claim 11 of:
“an overlapping degree…
overlapping degree”.
SUN II teaches the difference of claim 11:
an overlapping degree (or “the overlapping degree”, pg. 7, 12th txt blk)…
overlapping degree (“threshold value”, pg. 7,12th txt blk).
Since FROLOVA of the combination of YUAN,Lissi,Sjolund,HUANG,FROLOVA teaches a threshold, one of skill in the art of thresholds can make FROLOVA’s of the combination (illustrated above) of YUAN,Lissi,Sjolund,HUANG, FROLOVA be as SUN II’s predictably recognizing the change “improving the network model under the premise of not increasing the calculation cost, inheriting the speed advantage of the YOLOv5algorithm. using non-local attention characteristic, not limited to local receptive field the target detection process, but using the global information, enhancing the characteristic fusion ability, considering the real time and effectively improving the accuracy of the identification network.”, SUN II, pg. 8, 9th txt blk.
Claim(s) 12 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation applied in claims 1,16,20 further in view of FROLOVA et al. (US 2021/0081754 A1) as applied in claims 2,3,8 and 17,18,19 further in view of HE at al. (US 2022/0036059 A1):
PNG
media_image21.png
1611
1235
media_image21.png
Greyscale
Re 12. (Previously Presented), FROLOVA of the combination (illustrated above) of YUAN,Lissi, Sjolund,HUANG,FROLOVA teaches The method of claim 2, wherein the identifying of the at least one first object further comprises:
identifying (“as a candidate for modification in the CNN”, FROLOVA [0053] penult S) at least one area (via reflected “activation maps that correspond92 to various… regions…of the image” FROLOVA [0011] 7th S), in (via “reflected” “data”, FROLOVA [0093] 3rd S) which (either the activation map or image region) a feature (“map”, pg. 13, 2nd txt blk) a feature importance93 satisfies a designated first (via “One or more criteria (e.g., a threshold)”, FROLOVA [0053] 2nd S) condition (“that reflects a satisfactory result”, FROLOVA [0095] 2nd S), in (via “reflected” “data”, FROLOVA [0093] 2nd S) the activation map (“most correlated with the newly received image” FROLOVA: [0030] 4th S: fig. 4: “COMPARE ACTIVATION MAP(S)”);
determining (“whether”94 FROLOVA [0053] 2nd S) that the at least one area (via “activation maps that correspond95 to various… regions…of the image” FROLOVA [0011] 7th S) of the activation map (“most correlated with the newly received image” FROLOVA: [0030] 4th S: fig. 4: “COMPARE ACTIVATION MAP(S)”) does not satisfy (not satisfy being implied or understood) at least one second condition (via “One or more criteria (e.g., a threshold)”, FROLOVA [0053] 2nd S); and
adjusting96 (“as a candidate for modification97 in the CNN”, FROLOVA [0053] penult S)
(A) a shape,
(B) a size, and/or
(C) a position
of at least a part of the at least one area (via reflected “activation maps that correspond98 to various… regions…of the image” FROLOVA [0011] 7th S),
based99 on
the identifying of the at least one area (“as a candidate for modification in the CNN”, FROLOVA [0053] penult S) and
the determining (“whether”100 FROLOVA [0053] 2nd S) that the at least one area (“as a candidate for modification in the CNN”, FROLOVA [0053] penult S) of the activation map (“most correlated with the newly received image” FROLOVA: [0030] 4th S: fig. 4: “COMPARE ACTIVATION MAP(S)”) does not satisfy (not satisfy being implied or understood via said “whether”101 FROLOVA [0053] 2nd S) the at least one second condition (via “One or more criteria (e.g., a threshold)”, FROLOVA [0053] 2nd S).
FROLOVA of the combination of YUAN,Lissi,Sjolund,HUANG, FROLOVA does not teach the difference of claim 12 the of Markush element [A,B and/or C]:
(A) a shape,
(B) a size, and/or
(C) a position
HE teaches the difference of claim 12 of the Markush element:
(A) a shape,
(B) (“adjusting”) a size (“of the first reference area through the processed second reference area to obtain the focus area of the human body attribute” [0092], fig. 2B: “Focus area C”: human body:
PNG
media_image22.png
518
857
media_image22.png
Greyscale
) , and/or
(C) a position102
Since FROLOVA of the combination of YUAN,Lissi,Sjolund,HUANG, FROLOVA teaches “recognition” (FROLOVA [0027]), one of skill in the art of recognition can make FROLOVA’s of the combination (illustrated above) of YUAN,Lissi,Sjolund,HUANG, FROLOVA be as HE’s predictably recognizing the change “can better focus on the area that it needs to focus on, thereby improving the accuracy of the human body attribute recognition”, HE [0159] lat S.
Claim(s) 13 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation applied in claims 1,16,20 further in view of CAO et al. (CN 111670357 A) with SEARCH machine translation:
PNG
media_image23.png
1611
1235
media_image23.png
Greyscale
Re 13. (Previously Presented), YUAN of the combination (illustrated above) of YUAN,Lissi teaches The method of claim 1, wherein the identifying of the at least one first object by using the first AI model (i.e., “an artificial intelligence (Artificial Intelligence, AI) set in a virtual scene battle through training”, pg. 5: fig. 4: a battle) and the first image (“frame identified as the first type”, pg. 11) comprises:
pre-processing the first image (“frame identified as the first type”, pg. 11); and
identifying the at least one first object, by using103 the pre-processed first image (“frame identified as the first type”, pg. 11) and the first AI model (i.e., “an artificial intelligence (Artificial Intelligence, AI) set in a virtual scene battle through training”, pg. 5: fig. 4: a battle).
YUAN of the combination (illustrated above) of YUAN,Lissi does not teach the difference of claim 13:
pre-processing…pre-processed.
CAO teaches the difference of claim 13:
pre-processing (via “pre-…processing data”, pg. 28, 5th txt blk)
(“The learning data selector can 1310-3 the data needed for learning from the”) pre-processed (“data”, pg. 28, last txt blk).
Since YUAN of the combination (illustrated above) of YUAN,Lissi teaches recognition, one of skill in the art of recognition can make YUAN’s of the combination (illustrated above) of YUAN,Lissi be as CAO’s predictably recognizing the change “becomes more and more intelligent system. the more the AI system is used, the recognition rate of the AI system can be improved more, and the AI system can more accurately understand the user preference, and therefore, the existing rule-based intelligent system is gradually replaced by the AI system based on the deep learning”, CAO, pg. 2, 7th txt blk.
Claim(s) 14 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation applied in claims 1,16,20 further in view of CAO et al. (CN 111670357 A) with SEARCH machine translation as applied in claim 13 further in view of RIM et al. (US 2019/0221313 A1):
PNG
media_image24.png
1610
1242
media_image24.png
Greyscale
Re 14. (Original), YUAN of the combination (illustrated above) of YUAN,Lissi,Sjolund, HUANG, CAO teaches The method of claim 13, wherein the pre-processing (via “pre-…processing data”, CAO pg. 28, 5th txt blk) of the first image (“frame identified as the first type”, pg. 11) comprises:
converting a size of the first image (“frame identified as the first type”, pg. 11); and
performing blurring on the first image (“frame identified as the first type”, pg. 11).
YUAN of the combination (illustrated above) of YUAN,Lissi,Sjolund,HUANG, CAO does not teach the difference of claim 14:
“converting a size of…
performing blurring on”.
RIM teaches the difference of claim 14:
converting a size of (“the image to an appropriate size” [0121])…
performing (resulting in an “applied”104 “Gaussian blur filter” RIM [0133]) blurring on (or “to105 an image” [0133]).
Since CAO of the combination (illustrated above) of YUAN,Lissi,Sjolund, HUANG, CAO teaches pre-processing, one of skill in the art of pre-processing can make CAO’s of the combination (illustrated above) of YUAN,Lissi,Sjolund,HUANG,CAO be as RIM’s predictably recognizing the change “improving the efficiency and performance of training”, RIM [0130].
Claim(s) 15 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation applied in claims 1,16,20 further in view of CAO et al. (CN 111670357 A) with SEARCH machine translation as applied in claim 13 further in view of RIM et al. (US 2019/0221313 A1) as applied in claim 14 further in view of Clapper (US 2003/0107584 A1):
PNG
media_image25.png
1611
1235
media_image25.png
Greyscale
Re 15. (Original), YUAN of the combination (illustrated above) of YUAN,Lissi,Sjolund,HUANG,CAO,RIM teaches The method of claim 14, wherein the performing blurring (resulting in an “applied”106 “Gaussian blur filter” RIM [0133]) on107108 the first image (“frame identified as the first type”, YUAN pg. 11) comprises:
determining a blurring degree, based on attributes of at least one (game-configuration) text (via “configuration interface109”, YUAN pg. 17, 5th txt blk: fig. 5:501-505, reproduced below, “displayed as type A-style B-red” YUAN, pg. 18 & “text data”, CAO, pg. 28, 6th txt blk) included110111 (via “including the target virtual object”, YUAN, pg. 21, 2nd txt blk) in the first image (“frame identified as the first type”, YUAN pg. 11); and
performing the blurring (resulting in an “applied”112 “Gaussian blur filter” RIM [0133]) based on the determined blurring degree.
PNG
media_image26.png
358
906
media_image26.png
Greyscale
YUAN of the combination of YUAN,Lissi,Sjolund,HUANG, CAO, RIM does not teach the difference of claim 15 of:
“determining a blurring degree, based113 on attributes…114
(performing blurring)115 based116 on the determined blurring degree.”
Clapper teaches the difference of claim 15:
(definitely) determining (via “adjust the degree of information” [0036] last S, blurring) a blurring degree (via fig. 4:402: “APPLY BLUR TO ALL OR SELECTED GRAPHIC DATA?”), based on attributes (“depending in part upon the value of attributes” [0031]) …
(performing blurring)117 (fig. 3:202) based on the (definitely) determined (information) blurring degree.
Since YUAN of the combination (illustrated above) of YUAN,Lissi,Sjolund, HUANG,CAO,RIM teaches “security” (YUAN, pg. 6, 2nd txt blk), one of skill in the art of security can make YUAN’s of the combination of YUAN,Lissi, Sjolund,HUANG, CAO, RIM be as Clapper’s “security” [0060] predictably recognizing the change “relates generally to the field of data processing and, more particularly, to improved systems and methods for providing secure viewing of information on a display.” Clapper [0001].
Claim(s) 21 is/are rejected under 35 U.S.C. 103 as being unpatentable over YUAN (WO 2023/029900 A1) with SEARCH machine translation III in view of Lissi (US 2022/0172426 A1) and Sjolund et al. (US 2019/0332900 A1) and HUANG (CN 114359645 A) with SEARCH machine translation applied in claims 1,16,20 further in view of CHENG (CN 108647089 A) with SEARCH machine translation:
PNG
media_image27.png
1611
1235
media_image27.png
Greyscale
Re 21. (Previously Presented), YUAN of the combination of YUAN,Lissi,Sjolund,HUANG teaches The electronic device of claim 16, wherein the one or more computer programs further comprise computer-executable instructions to:
apply a first (“terminal 110”, pg. 6, 1st txt blk) policy based on the first scene identifier,
wherein the first policy controls (via “control information”, pg. 6, 3rd txt blk) (“rendering”, pg. 17, 3rd txt blk) performance118 of the electronic device based on an index determined using the first scene identifier, the first scene identifier improving accuracy (“which can improve the accuracy of the template matching”, pg. 12, 3rd txt blk) of policy determination.
YUAN of the combination of YUAN,Lissi,Sjolund,HUANG does not teach the difference119 of claim 21 of:
a) a (first) policy …performance120…
b) an index determined using (the first scene identifier).
CHENG teaches the difference of claim 21:
a) a (first) policy121 (comprised by “programs”122, pg. 6, 7th txt blk: figures 4,5: re-created below: “strategy123 module”: policy module)…(“adjust124 the”) performance125 (“of the associated system resource based on the resource allocation strategy”, pg. 9,ll. 7th txt blk)…
b) an (“application operation index”, pg. 13, 7th txt blk) index determined (via fig. 10:1005) using126 (specially via “the preset application127 scene identifier corresponding to the application scene”, pg. 12, 6th txt blk: fig. 10: step 1003) (the first scene identifier (via figures 4,5:
PNG
media_image28.png
1641
1150
media_image28.png
Greyscale
PNG
media_image29.png
1207
1132
media_image29.png
Greyscale
Since YUAN of the combination of YUAN,Lissi,Sjolund,HUANG teaches a computer application, one of skill in the art of computer-apps can make YUAN’s of the combination of YUAN,Lissi,Sjolund,HUANG be as CHENG’s seeing in the change “As shown in FIG. 2, in order to improve the running quality of the third party application program, to the data communication between the third-party application program and operating system, so that the operating system can obtain the third party application program the current scene information so as to do targeted system resource adaptation based on the current scene, at the same time, the third-party application program can real-time obtain the operating state of the operating system, thereby performing program optimization based on the running state.” CHENG, page 5, 4th txt blk:
PNG
media_image30.png
1016
1025
media_image30.png
Greyscale
PNG
media_image31.png
1513
897
media_image31.png
Greyscale
Conclusion
The prior art “nearest to the subject matter defined in the claims” (MPEP 707.05) made of record and not relied upon is considered pertinent to applicant's disclosure.
The following table lists several references that are relevant to the subject matter claimed and disclosed in this Application. The references are not relied on by the Examiner, but are provided to assist the Applicant in responding to this Office action.
Citation
Relevance
Lin et al. (Lithological Classification by Hyperspectral Images Based on a Two-Layer XGBoost Model, Combined with a Greedy Algorithm)
Lin teaches “overfitting” and a “feature importance threshold”:
1. Introduction: page 3, 1st para, penult S:
“Using a greedy algorithm to apply the idea of ensemble learning in feature selection, the ideal feature selection strategy can be generated to address problems faced by traditional tree-based feature selection methods, such as overfitting functions, the masking of enumerated features, and high computational complexity.”; and
3.2 Feature Selection Results, page 13, 5th S:
“The cumulative feature importance threshold K was set at 80% to select the optimal number of importance features.”
as the closest to the claimed “threshold feature importance” and “reduction of overfitting” of claim 1.
Huts et al. (US 2023/0067026 A1)
Huts teaches a “predictive value” or in other words a “feature importance” and “less than a threshold value” and “prevent overfitting” in [0163][0189]:
[0163] In some embodiments, feature engineering operations are performed based on the predictive value (e.g., feature importance) of the features. For example, feature engineering may include pruning “less important” features from the dataset. In this context, a feature may be classified as “less important” if the predictive value (e.g., feature importance) of the feature is less than a threshold value, if the feature has one of the M lowest predictive values among the features in the dataset, if the feature does not have one of the N highest predictive values among the features in the dataset, etc. As another example, feature engineering may include creating derived features from “more important” features in the dataset. In this context, a feature may be classified as “more important” if the predictive value of the feature is greater than a threshold value, if the feature has one of the N highest predictive values among the features in the dataset, if the feature does not have one of the M lowest predictive values among the features in the dataset, etc.
[0189] In some instances, for example, a significant number of training images may be needed to obtain good modeling results and prevent overfitting; however, obtaining an adequate supply of training images can be difficult. For example, the images may be costly or difficult to obtain in the field, or may be costly to annotate. In general, image augmentation is a process of creating new artificial training examples by introducing slight modifications to existing examples, as shown in FIG. 8A.
as the closest to the claimed “threshold feature importance” and “reduction of overfitting” of claim 1.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DENNIS ROSARIO whose telephone number is (571)272-7397. The examiner can normally be reached Monday-Friday, 9AM-5PM EST.
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, Henok Shiferaw can be reached at 571-272-4637. 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.
/DENNIS ROSARIO/Examiner, Art Unit 2676
/MATTHEW C BELLA/Supervisory Patent Examiner, Art Unit 2667
1 preventing: present participle of prevent (overfitting), wherein prevent (overfitting) is defined: to keep (overfitting) from occurring; avert; hinder. (Dictionary.com)
2 by: in consequence, as a result, or on the basis of (removing important data).. (Dictionary.com)
3 removing: present participle of remove., wherein remove (important data) is defined: to move (important data) from a place or position; take (important data) away or off. (Dictionary.com)
4 exclusion: an act or instance of excluding, wherein exclude (the at least one first object) is defined: to shut or keep out (the at least one first object); prevent the entrance of (the at least one first object…wherein the at least one first object has at least one feature importance…greater than or equal to a threshold) (Dictionary.com)
5 reducing: present participle of reduce, wherein reduce (overfitting) is defined: to bring down (overfitting) to a smaller extent, size, amount, number, etc. (Dictionary.com)
6 BROAD CLAIM LANGUAGE: threshold: a level or point at which something would happen, would cease to happen, or would take effect, become true, etc. (Dictionary.co
7 MPEP 2131 Anticipation — Application of 35 U.S.C. 102 [R-08.2017], 2nd to last S: The elements must be arranged as required by the claim, but this is not an ipsissimis verbis test, i.e., identity of terminology is not required. In re Bond, 910 F.2d 831, 15 USPQ2d 1566 (Fed. Cir. 1990).
8 Non-limiting claim limitation: comma: the punctuation mark(,) indicating a slight pause in the spoken sentence and used where there is a listing of items or to separate a nonrestrictive clause or phrase (“by the electronic device”) from a main clause (Dictionary.com)
9 Non-limiting claim limitation: comma: the punctuation mark(,) indicating a slight pause in the spoken sentence and used where there is a listing of items or to separate a nonrestrictive clause or phrase (“by the electronic device”) from a main clause (Dictionary.com)
10 identifier: a book, computer application, or other source that one can use to identify something, such as a bird, plant, etc., by matching it with a sound, photo, or description in an inventory, wherein description in defined: sort; kind; variety, wherein sort is defined: character, quality, or nature, wherein character is defined: the aggregate of features and traits that form the individual nature of some person or thing (Dictionary.com)
11 scene: an area or sphere of activity, current interest, etc.. (Dictionary.com)
12 scene: an area or sphere of activity, current interest, etc.. (Dictionary.com)
13 judge: to infer, think, or hold as an opinion; conclude about or assess, wherein infer is defined: to derive by reasoning; conclude or judge from premises or evidence, wherein derive is defined: to trace from a source or origin (comprised by “original pixel value”, pg. 14, 4th txt blk), wherein trace is defined: to ascertain by investigation; find out; discover, wherein find out (verb phrase) is defined: to uncover the true nature, identity, or intentions of (someone), wherein identity is defined: the qualities, beliefs, etc., that distinguish or identify a person or thing. (Dictionary.com)
14 object: Digital Technology. any item that can be individually selected or manipulated, as a picture, data file, or piece of text. (Dictionary.com)
15 Is: 3rd person singular present indicative of be ,wherein be is defined: (used as a copula to connect the subject {“a target virtual object”} with its predicate adjective, or predicate nominative (“displayed in the video frame”), in order to describe, identify, or amplify the subject {“a target virtual object”}), wherein identify is defined: to make, represent to be, or regard or treat as the same or identical, wherein represent is defined: to be the equivalent of; correspond to (Dictionary.com)
16 display: Digital Technology., to output (“texture”, pg. 17, 1st text blk, data) on a screen, wherein data is defined: (used with a singular verb), a body of facts; information, wherein information is defined: (“texture”) data at any stage of processing (input, output, storage, transmission, etc.) (Dictionary.com)
17 for: intended to belong to, or be used in connection with (Dictionary.com)
18 “feature” is interpreted as an adjective
19 “importance” is interpreted as a noun
20 Regarding “feature importance”: MPEP 2131 Anticipation — Application of 35 U.S.C. 102 [R-08.2017], 2nd para, 2nd to last S: The elements must be arranged as required by the claim, but this is not an ipsissimis verbis test, i.e., identity of terminology is not required. In re Bond, 910 F.2d 831, 15 USPQ2d 1566 (Fed. Cir. 1990).
21 CLAIM SCOPE via Applicant’s disclosure, last page, last para:
[00159] While the disclosure has been shown and described with reference to
various embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the disclosure as defined by the appended claims and their equivalents.
22 The “wherein the at least one first object has at least one feature importance associated with the first scene identifier greater than or equal to a threshold feature importance” clause does have “the limiting effect of the language in” and thus is “a limitation in” claim 1 “where the clause gave ‘meaning and purpose to the manipulative steps’ “, in view of claim 1’s “acquiring” step followed by the “identifying” step via MPEP 2111.04 I. "ADAPTED TO," "ADAPTED FOR," "WHEREIN," and "WHEREBY"
Claim scope is not limited by claim language that suggests or makes optional but does not require steps to be performed, or by claim language that does not limit a claim to a particular structure. However, examples of claim language, although not exhaustive, that may raise a question as to the limiting effect of the language in a claim are:
(A) "adapted to" or "adapted for" clauses;
(B) "wherein" clauses {--wherein the at least one first object has at least one feature importance associated with the first scene identifier greater than or equal to a threshold feature importance--} ; and
(C) "whereby" clauses.
The determination of whether each of these clauses is a limitation in a claim depends on the specific facts of the case. See, e.g., Griffin v. Bertina, 285 F.3d 1029, 1034, 62 USPQ2d 1431 (Fed. Cir. 2002) (finding that a "wherein" clause limited a process claim where the clause gave "meaning and purpose to the manipulative steps").
23 probability (i.e., a distinguishing feature importance: statistics a measure or estimate of the degree of confidence one may have in the occurrence of an event, measured on a scale from zero (impossibility) to one (certainty). It may be defined as the proportion of favourable outcomes to the total number of possibilities if these are indifferent ( mathematical probability ), or the proportion observed in a sample ( empirical probability ), or the limit of this as the sample size tends to infinity ( relative frequency ), or by more subjective criteria ( subjective probability ), wherein proportion is defined: the significance of a thing or event that an objective view reveals, wherein significance is defined: consequence or importance , wherein favorable is defined: characterized by approval or support; positive, wherein characterize is defined: to mark or distinguish as a characteristic; be a characteristic of, wherein characteristic is defined: a distinguishing feature or quality. (Dictionary.com)
24 distribution (i.e., a distinct feature importance: statistics the set of possible values of a random variable, or points in a sample space, considered in terms of new theoretical or observed frequency, wherein value is defined: the desirability of a thing, often in respect of some property such as usefulness or exchangeability; worth, merit, or importance, wherein random variable is defined: rv. statistics a quantity that may take any of a range of values, either continuous or discrete, which cannot be predicted with certainty but only described probabilistically, wherein quantity is defined: the aspect or property of anything that can be measured, weighed, counted, etc , wherein aspect is defined: a distinct feature or element in a problem, situation, etc; facet (Dictionary.com: BRITISH)
25 THE CLAIMED INVENTION AS A WHOLE regarding: “excluding”:
The problem is via applicant’s disclosure:
[0003] An electronic device may adjust the performance (which may be, but is not limited to, for example, a central processing unit (CPU) clock within the electronic device) of the electronic device, based on a current state (which may be, but is not limited to, for example, frames per second (FPS) and/or temperature for the output of a display). The electronic device may select a policy for a performance control based on the current state, and may control performance based on the selected policy. For example, in an overheat state, the electronic device may select a policy for reducing a CPU clock, and may control the performance of the electronic device, based on the selected policy. When a specific application (e.g., a game application) is executed, the electronic device may identify whether an index required by the application is satisfied. If the index required by the application is not satisfied, the electronic device may control performance by changing a policy. In order to accurately determine whether the index required for the specific application is satisfied, more accurate monitoring of the current state may be required.
The solution is:
[0090] In operation 605, the trainer may perform first processing for changing a pixel value of at least a part of a first area of the first image620, for example, an area including the visual objects626 and 627, thereby identifying training data630 for the first Al model. The change of the pixel value here may include, for example, black- processing of changing the pixel value to a value corresponding to black, but those skilled in the art may understand that there is no limitation on a pixel value and/or a pixel value pattern after the change. The training data630 for the first Al model may include, for example, black-processed areas636 and 637. As the first processing (e.g., black-processing) is performed on the first area, for example, the area including the visual objects626 and 627, the training data630 for the first Al model may be provided, but the black-processing is merely an example, and there is no limitation on a processing scheme. For example, the first Al model630 may identify, as the first scene identifier, a scene identifier corresponding to the first image620, based on the visual objects626 and 627 of the first image620. There is a possibility that an image, which includes no visual object capable of increasing a possibility of classification as a scene identifier, is not classified as the corresponding scene identifier. For example, there may be a possibility that the first Al model630 is trained to classify a scene identifier, which corresponds to an image that does not include the visual objects626 and 627, as a scene identifier other than the first scene identifier. The trainer may generate the training data630 excluding the visual objects626 and 627 which have made a relatively large contribution to classification as the first scene identifier. The trainer may train the first Al model630 by using the training data630. Accordingly, the first Al model630 may train the first Al model630 so that an image, which includes objects (e.g., at least some of the objects621,objects 621,622,objects 621, 622,623,objects 621, 622, 623,624, and 625) other than the visual objects626 and 627 having made a relatively large contribution to classification as the first scene identifier, is also classified as the first scene identifier. Accordingly, the first Al model630 may
be trained based on various visual objects, and may not be over-fitted for some visual objects.
26 (italics) represent claim limitations already taught
27 ellipses (…) represent claim limitations already taught
28 this difference reasonably maps to applicant’s solution to the CPU overheating: an indication of non-obviousness: however, the comma-enclosed claimed “, electronic device,” is not a limitation under the broadest reasonable interpretation of method claim 1.
29 (italics) represent claim limitations already taught
30 ellipses (…) represent claim limitations already taught
31 {curly brackets} represents (1) being a limitation already taught under the broadest reasonable interpretation or (2) a difference that is “improperly limited to a narrow subset of claim scope” of the broadest reasonable interpretation in view of applicant’s disclosure, such as last page, last paragraph regarding scope of the dislosure: MPEP 2143.03 All Claim Limitations Must Be Considered [R-01.2024], 1st para: --Examiners must consider all claim limitations when determining patentability of an invention over the prior art. In re Gulack, 703 F.2d 1381, 1385, 217 USPQ 401, 403-04 (Fed. Cir. 1983). The subject matter of a properly construed claim is defined by the terms that limit the scope of the claim when given their broadest reasonable interpretation. In Axonics, Inc. v. Medtronic, Inc., 73 F.4th 950, 958-59, 2023 USPQ2d 795 (Fed. Cir. 2023), the court found the claims were improperly narrowed based on a preferred embodiment to sacral anatomy or sacral neuromodulation, whereas the patent claims made no reference to sacral anatomy or sacral neuromodulation. Thus, the relevant prior art was improperly limited to a narrow subset (i.e., the exact word itself “feature importance” word-for-word via application of the ipsissimis verbis test) of claim scope. See also MPEP § 2111 et seq. It is the subject matter of the properly construed claim that must be examined. The determination of whether particular language is a limitation in a claim depends on the specific facts of the case. See, e.g., Griffin v. Bertina, 285 F.3d 1029, 1034, 62 USPQ2d 1431 (Fed. Cir. 2002).--
32 MPEP 2141.02 Differences Between Prior Art and Claimed Invention [R-01.2024]
I. THE CLAIMED INVENTION AS A WHOLE MUST BE CONSIDERED
II. DISTILLING THE INVENTION DOWN TO A "GIST" OR "THRUST" OF AN INVENTION DISREGARDS "AS A WHOLE" REQUIREMENT, last S:
Panduit Corp. v. Dennison Mfg. Co., 810 F.2d 1561, 1 USPQ2d 1593 (Fed. Cir. 1987), cert. denied, 481 U.S. 1052 (1987) (district court improperly distilled claims down to a one word solution to a problem)
III. DISCOVERING SOURCE/CAUSE OF A PROBLEM IS PART OF "AS A WHOLE" INQUIRY
"[A] patentable invention may lie in the discovery of the source of a problem even though the remedy may be obvious once the source of the problem is identified. This is part of the ‘subject matter as a whole’ which should always be considered in determining the obviousness of an invention under 35 U.S.C. § 103." In re Sponnoble, 405 F.2d 578, 585, 160 USPQ 237, 243 (CCPA 1969). However, "discovery of the cause of a problem . . does not always result in a patentable invention. . . . [A] different situation exists where the solution is obvious from prior art which contains the same solution for a similar problem." In re Wiseman, 596 F.2d 1019, 1022, 201 USPQ 658, 661 (CCPA 1979) (emphasis in original).
In In re Sponnoble, the claim was directed to a plural compartment mixing vial wherein a center seal plug was placed between two compartments for temporarily isolating a liquid-containing compartment from a solids-containing compartment. The claim differed from the prior art in the selection of butyl rubber with a silicone coating as the plug material instead of natural rubber. The prior art recognized that leakage from the liquid to the solids compartment was a problem, and considered the problem to be a result of moisture passing around the center plug because of microscopic fissures inherently present in molded or blown glass. The court found the inventor discovered the cause of moisture transmission was through the center plug, and there was no teaching in the prior art which would suggest the necessity of selecting applicant's plug material which was more impervious to liquids than the natural rubber plug of the prior art.
In In re Wiseman, 596 F.2d at 1022, 201 USPQ at 661, claims directed to grooved carbon disc brakes wherein the grooves were provided to vent steam or vapor during a braking action to minimize fading of the brakes were rejected as obvious over a reference showing carbon disc brakes without grooves in combination with a reference showing grooves in noncarbon disc brakes for the purpose of cooling the faces of the braking members and eliminating dust, thereby reducing fading of the brakes. The court affirmed the rejection, holding that even if the inventor discovered the cause of a problem, the solution would have been obvious from the prior art which contained the same solution (inserting grooves in disc brakes) for a similar problem.
33 MPEP 2143 Examples of Basic Requirements of a Prima Facie Case of Obviousness [R-01.2024], 3rd para: “"[T]he analysis that "should be made explicit" refers not to the teachings in the prior art of a motivation to combine, but to the court’s analysis. . . . Under the flexible inquiry set forth by the Supreme Court, the district court therefore erred by failing to take account of ‘the inferences and creative steps,’ or even routine steps, that an inventor would employ and by failing to find a motivation to combine related pieces from the prior art." Ball Aerosol, 555 F.3d at 993, 89 USPQ2d at 1877.
34 this difference reasonably maps to applicant’s solution to the CPU overheating: an indication of non-obviousness
35 (italics) represent claim limitations already taught
36 ellipses (…) represent claim limitations already taught
37 {curly brackets} represent a difference under the improper narrow subset (e,g,, application of the ipsissimis verbis test that is not being applied at this portion in the rejection of claim 1) of claim scope of the broadest reasonable interpretation
38 this difference reasonably maps to applicant’s solution to the CPU overheating: an indication of non-obviousness
39 (italics) represent claim limitations already taught
40 ellipses (…) represent claim limitations already taught
41 {curly brackets} represent a difference under the improper narrow subset of claim scope under the broadest reasonable interpretation: ipsissimis verbis test is still not being applied in this portion of the rejection of claim 1.
42 MPEP 2143.03 All Claim Limitations Must Be Considered [R-01.2024], 2nd para:
Examiners must consider all claim limitations when determining patentability of an invention over the prior art. In re Gulack, 703 F.2d 1381, 1385, 217 USPQ 401, 403-04 (Fed. Cir. 1983). The subject matter of a properly construed claim is defined by the terms that limit the scope of the claim when given their broadest reasonable interpretation. In Axonics, Inc. v. Medtronic, Inc., 73 F.4th 950, 958-59, 2023 USPQ2d 795 (Fed. Cir. 2023), the court found the claims were improperly narrowed based on a preferred embodiment to sacral anatomy or sacral neuromodulation, whereas the patent claims made no reference to sacral anatomy or sacral neuromodulation. Thus, the relevant prior art was improperly limited to a narrow subset of claim scope. See also MPEP § 2111 et seq. It is the subject matter of the properly construed claim that must be examined. The determination of whether particular language is a limitation in a claim depends on the specific facts of the case. See, e.g., Griffin v. Bertina, 285 F.3d 1029, 1034, 62 USPQ2d 1431 (Fed. Cir. 2002).
43 THE CLAIMED INVENTION AS A WHOLE regarding “feature importance” (fig. 8A:807):
The disclosed problem is multi-faceted (1) “an overheat state” [0003] (fig. 9:903: “OVERHEAT STATE?”) (2) “an index required…is not satisfied” [0003] (CLAIM 21: “index”) (3) “When…the fps of 59.83…is obtained …the electronic device 101 may…unnecessarily increase a CPU clock and/or a GPU clock” [0073] (4) “classification error” [00105] via applicant’s disclosure:
The disclosed solution [00105] is “server 108 may perform additional training” and “additional adjustment to the training data” [00100] (add a more triangular wing to a spaceship or add more thruster engine flare shapes to the spaceship) (e.g., figs. 5-8: Is there allowable subject matter in figures 5-8 as the solution {i.e., identifying/recognizing the correct action (60 fps) battle scene (see applicant’s original file drawing fig. 6C:zoom-in to 641: reproduced below)/the correct no-action (30 fps) interface scene: applicant’s fig. 10B to use the correct corresponding performance policy} to the overheating problem at 59.83 fps + 44.53 fps=104.36 fps [use the over-heat performance policy to reduce 104.36 fps to 60 fps based on correctly identifying/ recognizing the battle-scene (flare-shapes and triangular wings) 641]?).
44 MPEP 2131 Anticipation — Application of 35 U.S.C. 102 [R-08.2017], 2nd para, 2nd to last S: “The elements must be arranged as required by the claim, but this is not an ipsissimis verbis test, i.e., identity of terminology is not required. In re Bond, 910 F.2d 831, 15 USPQ2d 1566 (Fed. Cir. 1990).”, wherein identity is defined: the state or fact of being the same one as described, wherein same is defined: identical with what is about to be or has just been mentioned, wherein the synonym for ipsissimis verbis is ”word for word” ADVERB (Dictionary.com/ Thesaurus.com)
45 {curly brackets} represent a difference under the improper narrow subset of claim scope of the broadest reasonable interpretation
46 MPEP 2143 Examples of Basic Requirements of a Prima Facie Case of Obviousness [R-01.2024], 3rd para: "[T]he analysis that "should be made explicit" refers not to the teachings in the prior art of a motivation to combine, but to the court’s analysis. . . . Under the flexible inquiry set forth by the Supreme Court, the district court therefore erred by failing to take account of ‘the inferences and creative steps,’ or even routine steps, that an inventor would employ and by failing to find a motivation to combine related pieces from the prior art." Ball Aerosol, 555 F.3d at 993, 89 USPQ2d at 1877
47 detection: the act or process of extracting information, esp at audio or video frequencies, from an electromagnetic wave See also demodulation (Dictionary.com)
48 THE CLAIM INVENTION AS A WHOLE regarding “electronic device”:
The over-hearting CPU problem is discussed in the rejection of claim 1. All references (YUAN,Lissi,Sjoland) teach a similar CPU problem regarding the GPU that aids the overburdened CPU, wherein GPU is defined: Computers. graphics processing unit: a secondary processor usually dedicated to performing the calculations necessary for producing computer graphics, lessening the burden on the main processor. (Dictionary.com). Thus it would have been obvious to combine as the rejection of claim 1.
49 Applicant’s disclosure:
[0009] In accordance with another embodiment aspect of the disclosure, one or more non-transitory computer-readable storage media storing one or more computer programs including computer-executable instructions that, when executed by one or more processors of an electronic device, cause the electronic device to perform operations are provided. The operations include identifying, by the electronic device, a first image corresponding to a first scene identifier. The operations further include, by using a first Al model and the first image, identifying, by the electronic device, at least one first area contributing to classification of the first image as the first scene identifier, and identify, by the electronic device, training data for the first Al model by performing first processing for changing a pixel value of at least a part of the at least one first area of the first image.
[00152] According to an embodiment, one or more non-transitory computer-
readable storage media storing one or more computer programs including computer- executable instructions that, when executed by one or more processors 120 of an electronic device 101 or 108, cause the electronic device 101 or 108 to perform operations may be provided. The operations may include identifying, by the electronic device, a first image corresponding to a first scene identifier. Furthermore, by using a first artificial intelligence (AI) model and the first image, identifying, by the electronic device, at least one first area contributing to classification of the first image as the first scene identifier, and identifying training data for the first Al model by performing first processing for changing a pixel value of at least a part of the at least one first area of the first image.
50 Regarding another improvement under 35 USC 101: Applicant’s disclosure has in addition to improving AI training operations another type of operation (e.g., at paragraph [00105]: “In operation 903, the electronic device 101 may identify, based on the state information, whether a current state of the electronic device 101 is an overheat state.” This device operation, that is not claimed, is also improved via the disclosure (e.g., said [00105]: “In operation 911, the electronic device 101 may identify whether performance is improved compared to before applying of the policy.”)
51 Present participle
52 Past participle
53 (Italics) represent claim limitations already taught above
54 Ellipses (…) represent claim limitations already taught above
55 (Italics) represent claim limitations already taught above
56 Ellipses (…) represent claim limitations already taught above
57 detection: the act or process of extracting information, esp at audio or video frequencies, from an electromagnetic wave See also demodulation (Dictionary.com)
58 correspond: to be similar or analogous; be equivalent in function, position, amount, etc. (Dictionary.com)
59 “feature importance”: (1) appears redundant or (2) suggests that there are contextually two features or (3) may be a term of art.
60 “feature importance”: (1) appears redundant or (2) suggests that there are contextually two features or (3) may be a term of art.
61 similarity: an aspect, trait, or feature like or resembling another or another's, wherein feature is defined: a prominent or conspicuous part or characteristic, wherein prominent is defined: leading, important, or well-known, wherein important is defined: of much or great significance or consequence, wherein significance is defined: importance; consequence, wherein characteristic is defined: a distinguishing feature or quality, wherein feature is defined: a prominent or conspicuous part or characteristic, wherein prominent is defined: leading, important, or well-known, wherein important is defined: of much or great significance or consequence, wherein significance is defined: importance (Dictionary.com)
62 detect: to discover the existence of, wherein discover is defined: to notice or realize, wherein notice is defined: to perceive; become aware of., wherein perceive is defined: to recognize, discern, envision, or understand, wherein recognize is defined: to identify as something or someone previously seen, known, etc. (Dictionary.com)
63 “based” a past-participle contributing to the action of “corresponds to the at least one area”
64 The pronoun of “in which” refers to claim 18’s “at least one area”; claim 18’s “the activation map”; and/or claim 18’s “the at least one first object”.
65 For the pronoun of “in which” see claim 18’s footnote that specifies the pronoun “in which”: The pronoun of “in which” refers to claim 18’s “at least one area”; claim 18’s “the activation map”; and/or claim 18’s “the at least one first object”
66 Claim 5 is not interpreted under the narrow subset of claim scope of the broadest reasonable interpretation as discussed in the rejection of claim 1.
67 any: one, a, an, or some; one or more without specification or identification. (Dictionary.com)
68 {curly brackets} represent claim limitations already taught
69 correspond: to be similar or analogous; be equivalent in function, position, amount, etc. (Dictionary.com)
70 the: (used to mark a noun as being used generically). (Dictionary.com)
71 “the
72 correspond: to be similar or analogous; be equivalent in function, position, amount, etc. (Dictionary.com)
73 template: anything that determines or serves as a pattern; a model, wherein pattern is defined: a distinctive style, model, or form, wherein form is defined: the shape of a thing or person, wherein shape is defined: the quality of a distinct object or body in having an external surface or outline of specific form or figure, wherein outline is defined: the line by which a figure or object is defined or bounded; contour.
74 correspond: to be similar or analogous; be equivalent in function, position, amount, etc. (Dictionary.com)
75 similarity: the state of being similar; likeness; resemblance, wherein likeness id defined: the state or fact of being like, wherein like is defined: corresponding or agreeing in general or in some noticeable respect, wherein noticeable is defined: attracting notice or attention; capable of being noticed, wherein notice is defined: to perceive; become aware of, wherein perceive is defined: to recognize, discern, envision, or understand, wherein recognize is defined: to identify as something or someone previously seen, known, etc. (Dictionary.com)
76 template: anything that determines or serves as a pattern; a model, wherein pattern is defined: a distinctive style, model, or form, wherein form is defined: the shape of a thing or person, wherein shape is defined: the quality of a distinct object or body in having an external surface or outline of specific form or figure, wherein outline is defined: the line by which a figure or object is defined or bounded; contour.
77 detect: to discover the existence of, wherein discover is defined: to notice or realize, wherein notice is defined: to perceive; become aware of., wherein perceive is defined: to recognize, discern, envision, or understand, wherein recognize is defined: to identify as something or someone previously seen, known, etc. (Dictionary.com)
78 (italics) represent claim limitations already taught above
79 Ellipsis (…) represent claim limitations already taught above
80 (italics) represent claim limitations already taught above
81 Ellipsis (…) represent claim limitations already taught above
82 enhance: (tr) to intensify or increase in quality, value, power, etc; improve; augment (Dictionary.com)
83 signal-to-noise ratio: the ratio of one parameter, such as power of a wanted signal to the same parameter of the noise at a specified point in an electronic circuit, etc (Dictionry.com)
84 detection: the act or process of extracting information, esp at audio or video frequencies, from an electromagnetic wave See also demodulation (Dictionary.com)
85 participle
86 “feature importance”: (1) appears redundant or (2) suggests that there are contextually two features or (3) may be a term of art.
87 participle
88 “feature importance”: (1) appears redundant or (2) suggests that there are contextually two features or (3) may be a term of art.
89 past-participle participating with the action of (“identifying”).
90 “feature importance”: (1) appears redundant or (2) suggests that there are contextually two features or (3) may be a term of art.
91 parameter: Computers., a variable that must be given a specific value during the execution of a program or of a procedure within a program.. wherein variable is defined: Mathematics, Computers.
a quantity or function that may assume any given value or set of values, wherein value is defined: Mathematics. magnitude; quantity; number represented by a figure, symbol, or the like, wherein magnitude is defined: size; extent; dimensions. (Dictiionary.com)
92 correspond: to be similar or analogous; be equivalent in function, position, amount, etc. (Dictionary.com)
93 “feature importance”: (1) appears redundant or (2) suggests that there are contextually two features or (3) may be a term of art.
94 whether: (used to introduce a single alternative, the other being implied or understood, or some clause or element not involving alternatives). See whether or not she has come. I doubt whether we can do any better. (Dictionary.com)
95 correspond: to be similar or analogous; be equivalent in function, position, amount, etc. (Dictionary.com)
96 Markush element follows: [A.B and/or C]
97 modify: to change somewhat the form or qualities of; alter partially; amend: Synonyms: reform, shape, adjust, vary, wherein form is defined: the shape of a thing or person. (Dictionary.com)
98 correspond: to be similar or analogous; be equivalent in function, position, amount, etc. (Dictionary.com)
99 “based” is past participle contributing to the action of the claimed “adjusting”
100 whether: (used to introduce a single alternative, the other being implied or understood, or some clause or element not involving alternatives). See whether or not she has come. I doubt whether we can do any better. (Dictionary.com)
101 whether: (used to introduce a single alternative, the other being implied or understood, or some clause or element not involving alternatives). See whether or not she has come. I doubt whether we can do any better. (Dictionary.com)
102 Since Markush alternative (B) is taught the Markush element [A,B and/or C] is taught under the broadest reasonable interpretation of claim 12.
103 “using” is participle contributing to “identifying”
104 apply: to bring into action, wherein action is defined: something done or performed; act; deed. (Dictionary.com)
105 to: (used for expressing contact or contiguity) on; against; beside; upon. (Dictionary.com)
106 apply: to bring into action, wherein action is defined: something done or performed; act; deed. (Dictionary.com)
107 Applicant’s disclosure:
[0039] The terms and words used in the following description and claims are not limited to the bibliographical meanings, but, are merely used by the inventor to enable a clear and consistent understanding of the disclosure. Accordingly, it should be apparent to those skilled in the art that the following description of various embodiments of the disclosure is provided for illustration purpose only and not for the purpose of limiting the disclosure as defined by the appended claims and their equivalents.
108 on: in connection, association, or cooperation with; as a part or element of. (Dictionary.com)
109 interface: The layout of an application's graphic or textual controls in conjunction with the way the application responds to user activity, wherein textual is defined: of or relating to a text. (Dictionary.com)
110 past participle
111 Claim scope/Claim range: The participle “included” can modify over the range of the nouns (1) “attributes”: first-image-included- attributes; (2) “text”: first-image- included-text; and (3) “attributes” & “text”: first-image-included-attributes- &-text, wherein scope is defined: Linguistics, Logic., the range of words or elements of an expression over which a modifier (i.e. patent examiner) or operator (i.e., patent examiner) has control.
In “old men and women,” “old” may either take “men and women” or just “men” in its scope. (Dictionary.com)
112 apply: to bring into action, wherein action is defined: something done or performed; act; deed. (Dictionary.com)
113 past participle
114 ellipsis (…) represent claim limitations taught above
115 (italics) represent claim limitations taught above
116 past participle
117 (italics) represent claim limitations taught above
118 “performance” is directly receiving the action of “controls”
119 THE CLAIMED INVENTION AS A WHOLE regarding the claimed “policy”:
The CPU problem is discussed in the rejection of claim 1.
Another problem regarding accuracy is via applicant’s disclosure:
[0003] An electronic device may adjust the performance (which may be, but is not limited to, for example, a central processing unit (CPU) clock within the electronic device) of the electronic device, based on a current state (which may be, but is not limited to, for example, frames per second (FPS) and/or temperature for the output of a display). The electronic device may select a policy for a performance control based on the current state, and may control performance based on the selected policy. For example, in an overheat state, the electronic device may select a policy for reducing a CPU clock, and may control the performance of the electronic device, based on the selected policy. When a specific application (e.g., a game application) is executed, the electronic device may identify whether an index required by the application is satisfied. If the index required by the application is not satisfied, the electronic device may control performance by changing a policy. In order to accurately determine whether the index required for the specific application is satisfied, more accurate monitoring of the current state may be required.
The solution (figs. 8A,8B,9: training) to the accuracy problem is:
[0076] FIGS. 3E and 3F may illustrate an Al accuracy for each of different electronic device types according to various embodiments of the disclosure. For example, the Al model may be trained for an electronic device of a first type. FIG. 3E illustrates a scene prediction accuracy341 and a receiver operating characteristic (ROC) curve342 when the Al model is used in the electronic device of the first type. It may be identified that the scene prediction accuracy341 and the ROC curve342 are at relatively high levels. FIG. 3F illustrates a scene prediction accuracy351 and an ROC curve352 when the Al model is used in an electronic device of another type which is different from the first type. It may be identified that the scene prediction accuracy351 and the ROC curve352 are at relatively low levels. Accordingly, it may be required to precisely train the Al model for each application.
Claim 21 does not claim “precisely train the Al model for each application” (figs. 8A,8B,9: training): an indication of obviousness
120 “performance” is directly receiving the action of “controls”
121 BROAD CLAIM LANGUAGE: policy: a definite course of action adopted for the sake of expediency, facility, etc.. (Dictionary.com)
122 program: a plan of action to accomplish a specified end, wherein plan is defined: a scheme or method of acting, doing, proceeding, making, etc., developed in advance, wherein method is defined: a procedure, technique, or way of doing something, especially in accordance with a definite plan, wherein procedure is defined: a particular course or mode of action. (Dictionary.com) .
123 strategy: a plan, method, or series of maneuvers or stratagems for obtaining a specific goal or result, wherein plan is defined: a scheme or method of acting, doing, proceeding, making, etc., developed in advance, wherein scheme is defined: a plan, program, or policy officially adopted and followed, as by a government or business. (Dictionary.com)
124 adjust: (Dictionary.com)
125 “performance” is directly receiving the action of “controls”
126 -ing (of using): a suffix of nouns formed from verbs (use), expressing the action of the verb (use) or its result (“index determined”), product, material, etc. (the art of building; a new building; cotton wadding ), wherein express is defined: to put (thought) into words; utter or state (Dictionary.com)
127 application: the act of putting to a special use or purpose. (Dictionary.com)
Prosecution Insights