DETAILED ACTION
Claims 1-18 are presented for examination
This office action is in response to submission of application on 6-DECEMBER-2022.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Response to Amendment
The amendment filed on 30-APRIL-2026 in response to the non-final office action mailed 03-MARCH-2026 has been entered. Claims 1-18 remain pending in the application.
With regards to the 103 rejections, the applicant’s amendments to the claims have not overcome the rejections to claims 1-18 as newly found prior art BORIAH sufficiently teaches the newly added limitations of the amended claims.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1, 4, 7, 10, 13, and 16 are rejected under 35 U.S.C. 103 as being unpatentable over TOHIDI (U.S. Pub. No. US 20220016455 A1) in view of BORIAH (U.S. Pub. No. US 20150278603 A1) in view of DIRAC (U.S. Pub. No. US 10635973 B1)
Regarding claim 1, TOHIDI substantially teaches the claim including:
A computer-implemented method, comprising: obtaining a first plurality of data elements, each data element representing a fire-related metric of a geographic region; ([0041] In one example embodiment, a method is provided for receiving, via a computer network and by a fire forecasting system, fire-related inputs including vegetation data, topography data, weather data, and fire-monitoring information including data identifying a physical shape of a fire burning in a region. Further, the method generates, by the fire forecasting system, fire forecast data for the region based on the fire-related inputs. (due to the topography data, i.e. region data, we know that the metrics for the fire take place in a specific geographic region/topological region)) determining, using at least a subset of the first plurality of data elements, one or more derived values representing one or more derived fire-related metrics that characterizes behaviors of a fire; ([0038] In one embodiment, a method includes an operation for accessing a database to obtain values for a plurality of features associated with a fire in a geographical region. The plurality of features include one or more satellite images at a first resolution, vegetation information for the geographical region, and weather data for the geographical region. [0041] In one example embodiment, a method is provided for receiving, via a computer network and by a fire forecasting system, fire-related inputs including vegetation data, topography data, weather data, and fire-monitoring information including data identifying a physical shape of a fire burning in a region. Further, the method generates, by the fire forecasting system, fire forecast data for the region based on the fire-related inputs. The fire forecast data describes a state of the fire in the region at multiple times in the future. The state of the fire comprises a fire perimeter, a fire line intensity, and a flame height. Additionally, the method includes receiving, via the computer network, updated fire-monitoring information regarding a current state of the fire in the region. The fire forecasting system modifies the fire forecast data based on the updated fire-monitoring information. Further, the method causes presentation, in a user interface, of a fire forecast based on the fire forecast data.
(The values are obtained and determined to represent each of the metrics. Further, the features are a subset of the fire-related input of [0041] as seen by how they encompass some but not all data the fire-related inputs do.))
While TOHIDI does teach obtaining fire-related data, it does not explicitly teach:
associating the one or more derived values with the first plurality of data elements as training labels, a null value being associated with a first sub-set of data elements representing non-burn and the one or more derived values being associated with a second sub-set of data elements representing a burn;
However, in analogous art that similarly employs data mapping, BORIAH teaches:
associating the one or more derived values with the first plurality of data elements as training labels, a null value being associated with a first sub-set of data elements representing non-burn and the one or more derived values being associated with a second sub-set of data elements representing a burn; ([0045] In step 300, pixels of the satellite sensor data are clustered based on one of the features 220 by a clustering unit 222. These clusters are an incorrect and inexhaustive set of candidate burned areas. In some embodiments, AF presence is the feature used to cluster the pixels at step 300. The candidate burned areas (pixels) are grouped based on spatial neighborhood (graph G.sub.s) to form candidate burned events 214. FIG. 5(a) shows a true event with areas 500 and 502 being burn events with boundaries 504 and 506 between the burn events and an area 508 outside of the burn events. FIG. 5(b) shows the input candidate burn events 224 created using AF with areas 510 and 512 having AF values of True and area 514 having AF values of false. FIG. 5(c) shows candidate burn events 224 with spurious noise added to test the performance of the various embodiments as discussed further below.
[0046] In step 302, an initial fire probability is assigned to each of the input candidate burned events 224 by a training burn event selection unit 226. In accordance with one embodiment, selection unit 226 determines the initial fire probability based on the AF and temporal texture of its member pixels. At step 304, selection unit 226 selects candidate burn events with sufficiently high initial probabilities to be used as training burn events 228 by classifier trainers 230. In particular, classifier trainers 230 use training burn events 228 to train classifiers 232 at step 306. In accordance with one embodiment, classifiers 232 use V2D and LID features to classify pixels as either being part of a burn event or not being part of a burn event. In accordance with one embodiment, classifiers are trained for each land cover class by using the pixels of training burn events 228 assigned to each land cover class to train the respective classifier for that land cover class.
[0091] The embodiments above improve the operation of computing devices used to label satellite sensor data by removing the need to perform an exhaustive search for candidate burn areas. Such exhaustive searches require huge amount of processing time. By removing the need for such searches, embodiments described above improve the operation of the computing device when attempting to apply labels to satellite sensor data. In addition, embodiments above overcome a lack of sufficient training data for training classifiers that can classify pixels as being part of a burn event. By identifying an initial cluster of pixels as a burn event and using that initial cluster as training data for training the classifiers, the embodiments make a sensor data labelling computing device possible where it would not have been possible in the past. Lastly, the embodiments improve the operation of the computing devices used to label satellite sensor data by allowing those computing device to more accurately apply such labels to the satellite sensor data.)
It would have been obvious to a person skilled in the art before the effective filing date of the invention to have combined with BORIAH‘s data association and, with TOHIDI‘s fire-related data, with a reasonable expectation of success, a method for associating data to subsets using zero and non-zero values, as in BORIAH, where the subsets comprise of fire related metrics, as found in TOHIDI. A person of ordinary skill would have been motivated to improve reliability. (BORIAH [0003]).
TOHIDI further teaches:
obtaining a second plurality of data elements, each data element representing a fire-related feature of the geographic region, wherein the fire-related features comprise one of terrain-related features and weather-related features; ([0041] In one example embodiment, a method is provided for receiving, via a computer network and by a fire forecasting system, fire-related inputs including vegetation data, topography data, weather data, and fire-monitoring information including data identifying a physical shape of a fire burning in a region. Further, the method generates, by the fire forecasting system, fire forecast data for the region based on the fire-related inputs. (the method of obtaining data of paragraph [0041] is not restricted by data type and thus can be also used to get the second plurality of data elements since the second plurality is not said to be obtained a different way.))
While TOHIDI does teach obtaining fire-related data, it does not explicitly teach:
and training a machine learning (ML) model using (i) at least a subset of the first plurality of data elements, (ii) at least a subset of the second plurality of data elements as input, and the one or more derived values associated with the subset of the first plurality of data elements as training labels to provide a trained ML model.
However, in analogous art that similarly uses data subsets, DIRAC teaches:
and training a machine learning (ML) model using at least a subset of the first plurality of data elements, at least a subset of the second plurality of data elements, and the values associated with the subset of the first plurality of data elements to provide a trained ML model. ((DIRAC claim 6) train a prediction model, wherein training the prediction model includes at least: (1) involves determining a cost function based at least in part on a time-decay function applied to the portion of the second subset; (2) utilizing the first subset corresponding to the input values to predict the second subset (DIRAC uses two subsets to train a model where the first subset is associated with values.))
It would have been obvious to a person skilled in the art before the effective filing date of the invention to have combined with DIRAC‘s training method and, with TOHIDI‘s, as modified by BORIAH, fire-related data, with a reasonable expectation of success, a method for training a model using subsets, as in DIRAC, where the subsets comprise of fire related metrics, as found in TOHIDI, as modified by BORIAH. A person of ordinary skill would have been motivated to improve predictions (DIRAC, 27-34).
Regarding claims 7 and 13, they comprise of limitations similar to those of claim 1 and are therefore rejected for similar rationale.
Claims 2, 8, and 14 are rejected under 35 U.S.C. 103 as being unpatentable over TOHIDI (U.S. Pub. No. US 20220016455 A1), BORIAH (U.S. Pub. No. US 20150278603 A1), and DIRAC (U.S. Pub. No. US 10635973 B1) in further view of Madaio (“Firebird: Predicting Fire Risk and Prioritizing Fire Inspections in Atlanta”)
Regarding claim 2, while TOHIDI does teach claim 1, which claim 2 is dependent upon, it does not explicitly teach:
The computer-implemented method of claim 1, further comprising: generating an input from at least the subset of the first plurality data elements, at least the subset of the second plurality of data elements and the one or more values; and processing the input using the trained ML model that is configured to generate a fire risk prediction output that characterizes predicted future behaviors of a fire.
However, in analogous art that similarly predicts fire risk, MADAIO teaches:
The computer-implemented method of claim 1, further comprising: generating an input from at least the subset of the first plurality data elements, at least the subset of the second plurality of data elements and the one or more values; ((section 5.2, first sentence) After merging datasets, we had a total of 252 variables for each property. … (section 5.3, first paragraph – second paragraph)A fire risk model would ideally be tested in practice by predicting which properties would have a fire incident in the following year, and then waiting a year to verify which properties actually did catch fire. Because we wanted to effectively evaluate the accuracy of our model without waiting a year to collect data on new fires, we simulated this approach by using data from fire incidents in July 2011 to March 2014 as training data to predict fires in the last year of our data, April 2014 to March 2015. We used grid search with 10-fold cross validation on the training dataset to select the best models and parameters. The models we tried included Logistic Regression [19], Gradient Boosting [11], Support Vector Machine (SVM) [7], and Random Forest [3]. SVM and Random Forest performed the best, with comparable performances (see Table 3). For SVM, the best configuration is using RBF kernel with C = 0.5 and γ = 10 #features . For Random Forest, restricting the maximum depth of each tree to be 10 gave the best performance. Increasing the number of trees in general improves the performance, but we only used 200 trees since adding more trees only obtained insignificant improvement. (section 5.3, third paragraph, first sentence) We then trained SVM and Random Forest on the whole training set using the best parameters and generated predictions on the testing set (they used multiple datasets by combining them to make a training dataset and then used the parameters/values alongside the combined datasets/first and second subsets to generate input for further training.)) and processing the input using the trained ML model that is configured to generate a fire risk prediction output that characterizes predicted future behaviors of a fire. ((section 5.5, first paragraph) After we built the predictive model, we then applied the fire risk scores of each property to the list of current and potential inspectable properties, so that AFRD could focus on inspecting the properties most at risk of fire. To do this, we first computed the raw output of our predictive model for the list of properties we used to train and test the model. This generated a score between 0 and 1, which we then mapped to the discrete range of 1 to 10 that is easier for our AFRD colleagues to work with. (they then used the initial training set to generate further output from the model. In other words, the processed the input and then reinput it through the now trained model to produce a risk prediction output.))
It would have been obvious to a person skilled in the art before the effective filing date of the invention to have combined with MADAIO‘s risk prediction model and, with TOHIDI‘s, as modified by BORIAH, and DARIC, fire-related data, with a reasonable expectation of success, a method predicting fire risk, as in MADAIO, using fire related metric subsets, as found in TOHIDI, as modified by BORIAH, and DARIC. A person of ordinary skill would have been motivated to improve fire predictions (MADAIO, Abstract).
Regarding claims 8 and 14, they comprise of limitations similar to those of claim 2 and are therefore rejected for similar rationale.
Claims 3, 9, and 15 are rejected under 35 U.S.C. 103 as being unpatentable over TOHIDI (U.S. Pub. No. US 20220016455 A1), BORIAH (U.S. Pub. No. US 20150278603 A1), and DIRAC (U.S. Pub. No. US 10635973 B1) in further view of BERRY (U.S. Pub. No. US 20180276977 A1)
While TOHIDI, as modified by BORIAH and DIRAC, does teach claim 1, which claim 3 is dependent upon, it does not explicitly teach:
The computer-implemented method of claim 1, wherein the geographic region is contiguous.
However, in analogous art that similarly handles fire related data, BERRY teaches:
The computer-implemented method of claim 1, wherein the geographic region is contiguous. ([0004] A system as described herein can include a processor and a computer-readable storage medium having instructions stored which, when executed by the processor, cause the processor to perform operations including: receiving an alert notification to be communicated to a plurality of individuals associated with a geographic area, the geographic area comprising a contiguous land area;)
It would have been obvious to a person skilled in the art before the effective filing date of the invention to have combined with BERRY‘s contiguous region data and, with TOHIDI‘s, as modified by BORIAH and DARIC, fire-related data, with a reasonable expectation of success, geographic data that is contiguous, as in BERRY, which is included in fire-related metrics, as found in TOHIDI, as modified by BORIAH, and DARIC. A person of ordinary skill would have been motivated to make this combination to better target relevant areas (BERRY, [0011] ).
Regarding claims 9 and 15, they comprise of limitations similar to those of claim 3 and are therefore rejected for similar rationale.
Claims 5, 11, and 17 are rejected under 35 U.S.C. 103 as being unpatentable over TOHIDI (U.S. Pub. No. US 20220016455 A1), BORIAH (U.S. Pub. No. US 20150278603 A1), and DIRAC (U.S. Pub. No. US 10635973 B1) in further view of RYDER (U.S. Pub. No. US 20170169683 A1)
While TOHIDI does teach claim 1, which claim 5 is dependent upon, it does not explicitly teach:
The computer-implemented method of claim 1, wherein the derived fire-related metrics comprise one or more of speed, size, duration, and expansion.
However, in analogous art that similarly deals with fire related data, RYDER teaches:
The computer-implemented method of claim 1, wherein the derived fire-related metrics comprise one or more of speed, size, ([0067] In the implementation described above in which the system 100 continues to monitor a fire after confirmation and transmission of a fire alarm to an emergency responder, the system 100 can also transmit fire classification, size, intensity, temperature, location, and/or other fire-related data to an emergency responder while a fire burns near the system 100. ) duration, and expansion.
It would have been obvious to a person skilled in the art before the effective filing date of the invention to have combined with RYDER‘s fire-related metrics involving size and, with TOHIDI‘s, as modified by BORIAH and DARIC, fire-related data, with a reasonable expectation of success, data relating to the size of the fire, as in RYDER, which is included in fire-related metrics, as found in TOHIDI, as modified by BORIAH and DARIC. A person of ordinary skill would have been motivated to make this combination to better predict fires (RYDER, [0002] ).
Regarding claims 11 and 17, they comprise of limitations similar to those of claim 5 and are therefore rejected for similar rationale.
Claims 6, 12, and 18 are rejected under 35 U.S.C. 103 as being unpatentable over TOHIDI (U.S. Pub. No. US 20220016455 A1), BORIAH (U.S. Pub. No. US 20150278603 A1), and DIRAC (U.S. Pub. No. US 10635973 B1) in further view of ZHOU (U.S. Pub. No. US 20190371147 A1)
While TOHIDI, as modified by BORIAH and DIRAC, does teach claim 1, which claim 6 is dependent upon, it does not explicitly teach:
The computer-implemented method of claim 1, wherein the ML model comprises one of a gradient boosted decision tree, a random forest, and a convolutional neural network.
However, in analogous art that similarly deals with fire related data, ZHOU teaches:
The computer-implemented method of claim 1, wherein the ML model comprises one of a gradient boosted decision tree, a random forest, and a convolutional neural network. ([0014] Optionally, the sensing module includes a visual sensor, and the n types of environmental data comprise an image acquired by the visual sensor at the current moment, wherein determining a fire probability corresponding to each type of environmental data according to the n types of environmental data includes: inputting the image acquired by the visual sensor at the current moment into a fire model to acquire the fire probability corresponding to the image acquired by the visual sensor at the current moment, the fire model being configured to determine a fire occurrence probability according to the image; wherein the fire model is a fire model acquired by training a convolutional neural network with a sample set as training data)
It would have been obvious to a person skilled in the art before the effective filing date of the invention to have combined with ZHOU‘s ML model type and, with TOHIDI‘s, as modified by BORIAH and DARIC, fire-related data and model, with a reasonable expectation of success, a CNN, as in ZHOU, that predicts fire risk, as found in TOHIDI, as modified by BORIAH and DARIC. A person of ordinary skill would have been motivated to make this combination to improve model accuracy (ZHOU, abstract ).
Regarding claims 12 and 18, they comprise of limitations similar to those of claim 6 and are therefore rejected for similar rationale.
Response to Arguments
Applicant’s arguments filed 30-APRIL-2026 have been fully considered, but they are found to be non-persuasive
With regards to the applicant’s remarks regarding the 103 rejection in the non-final action, the applicant argues that the prior art does not teach the newly amended claims 1, 7, and 13. The examiner acknowledges this argument and has introduced newly found prior art BORIAH. BORIAH teaches associating data to sub-data sets based on a zero or non-zero evaluation as training labels. BORIAH is relevant due to its association in the art of mapping data, which is relevant to the applicant’s case. Further, the examiner has adjusted all dependent claims accordingly.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to SKIELER A KOWALIK whose telephone number is (571)272-1850. The examiner can normally be reached 8-5.
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, Mariela D Reyes can be reached at (571)270-1006. 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.
/SKIELER ALEXANDER KOWALIK/Examiner, Art Unit 2142 /Mariela Reyes/Supervisory Patent Examiner, Art Unit 2142
Prosecution Insights