CTNF 18/453,651 CTNF 102081 Notice of Pre-AIA or AIA Status 07-03-aia AIA 15-10-aia The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA. Priority 02-27 AIA Acknowledgment is made of applicant’s claim for foreign priority under 35 U.S.C. 119 (a)-(d). The certified copy has been filed in parent Application No. KR-10-2022-0186011 , filed on 27 Dec 2022 . Information Disclosure Statement 06-52 The information disclosure statement (IDS) submitted on August 22, 2023. The submission is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner. Claim Objections Claim 9 and 10 are objected to because of the following informalities: Claim 9 uses "wherein" twice consecutively. Claim 10 uses “off of distribution” instead of “out of distribution”. Appropriate corrections are required. Claim Rejections - 35 USC § 101 07-04-01 AIA 07-04 35 U.S.C. 101 reads as follows: Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title. Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. Regarding Claim 1 Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03 Claim 1 is a method claim thus it falls into one of the four categories of statutory subject matter. Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II. Regarding independent claim 1, following limitations recite a judicial exception: “determining an out of distribution (OOD) index with respect to the spectrum of the semiconductor for each of the at least one semiconductor based on the first data set” [Mathematical Calculation] – comparing two points/vectors from the data set to evaluate indexes for each semiconductor comprises pure mathematical computations that recites to an abstract idea. “ performing a data split on the first data set and the second data set by cluster sampling the first data set and the second data set into at least one learning data set with respect to the OOD index according to the at least one semiconductor” [Mental process] - splitting datasets into multiple groups based on the index, which is just an act of splitting items that is capable of being performed in the human mind with the assistance of paper and pen. “providing an optimal AI algorithm among a plurality of AI algorithms that have been trained on the at least one learning data set” [Mental process] - providing/selecting an optimal AI algorithm is purely comparing the performance of algorithms which is capable of being performed with the assistance of pen and paper. [Mathematical calculation] – training the data set of an optimal algorithm involves mathematical calculation that recites to an abstract idea. Step 2A Prong 2 Regarding Claim 1, the claim recites additional elements of “loading a first data set and a second data set, the first data set representing a spectrum of at least one semiconductor, and the second data set representing a structure of the at least one semiconductor”, Loading data to a computer component is merely data gathering recited at a high level of generality, thus is insignificant extra-solution activity (See MPEP 2106.05(g)). [Even when viewed in combination, the additional elements do not more than automate the mental processes that a person could perform, using computer components as a tool, thus the claim as a whole does not integrate into a practical application.] Step 2B The claim does not provide an inventive concept (significantly more than the abstract idea). The claim is ineligible. As explained above, the additional element [1] is considered an insignificant extra solution activity and at best the equivalent of a mere data gathering recited at a high level of generality and amount to receiving or transmitting data over a network, which is well-understood, routine, conventional activity (See MPEP 2106.05(d), subsection II). This limitation remains insignificant extra-solution activity even upon reconsideration. Even when considered in combination, the additional element represents insignificant extra-solution activity, which cannot provide an inventive concept. Regarding Claim 2 Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03 Claim 2 is a dependent claim of 1, thus it falls within the same category of statutory subject matter. Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II. Regarding dependent claim 2, following limitations recite a judicial exception: extracting a first principal component and a second principal component, with respect to the first data set, by performing principal component analysis (PCA) to reduce a dimension of the first data set [ Mathematical calculation] - performing PCA which is basically a linear algebra technique that finds the directions of greatest variance in high-dimensional data and projects the data in a lower dimension recites to an abstract idea. determining, for each of the at least one semiconductor, a value based on a Euclidean distance between vectors and a cosine distance between the vectors and an origin the vectors including the first principal component and the second principal component [Mathematical calculation] - finding distances between vectors using Euclidean and cosine formulas recites to an abstract idea. extracting a normalized value as the OOD index by normalizing an average value of a product of the Euclidean distance and the cosine distance for each of the at least one semiconductor [ Mathematical calculation] - adjusting scales of the distances found to measure out indexes is mathematical/statistical procedure that recites to an abstract idea. Step 2A Prong 2 The claim 2 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea. Regarding Claim 3 Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03 Claim 3 is a dependent claim of 1, thus it falls within the same category of statutory subject matter. Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II. Regarding dependent claim 3, following limitations recite a judicial exception: “ the performing of the data split includes sampling the first data set and the second data set at a ratio into a training group, a valid group, and a test group of each allocated learning data set according to the OOD indexes of the plurality of semiconductors ” [Mental process] – splitting datasets into multiple groups based on the index, which is just an act of splitting items that is capable of being performed in the human mind with the assistance of paper and pen. Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception? Regarding Claim X, the claim recites additional elements of “ the at least one semiconductor includes a plurality of semiconductors ” Semiconductors are recited at a high level of generality and is merely adding words “apply it” to the judicial exception. (See MPEP 2106.05f(f)) [Even when viewed in combination, the additional elements do not more than creating multiple datasets by sampling them based on the OOD index, thus the claim as a whole does not integrate into a practical application.] Step 2B – whether the claim as a whole amount to significantly more than the judicial exception? I.e. Are there any additional elements (features/limitations/step) recited in the claim beyond the abstract idea? The claim does not provide an inventive concept (significantly more than the abstract idea). The claim is ineligible. As explained above, the additional element [1] is considered a mere instruction to apply an exception to the data of semiconductors. This limitation remains a mere instruction to apply an exception even upon reconsideration. Even when considered in combination, the additional element represents a mere instruction to apply an exception, which cannot provide an inventive concept. Regarding Claim 4 Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03 Claim 4 is a dependent claim of 3, thus it falls within the same category of statutory subject matter. Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II. Regarding dependent claim 4, following limitations recite a judicial exception: “ the performing of the data split includes sampling the first data set and the second data set into a training group, a valid group, and a test group of learning data set, sequentially according to the OOD indexes of the plurality of semiconductors ” [Mental process] – splitting datasets into multiple groups based on the index in a specific order, which is just an act of splitting items by comparing/sorting them using their indexes which is capable of being performed in the human mind with the assistance of paper and pen. Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception? The claim 4 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea. Regarding Claim 5 Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03 Claim 5 is a dependent claim of 3, thus it falls within the same category of statutory subject matter. Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II. Regarding dependent claim 5, following limitations recite a judicial exception: “ the performing of the data split includes sampling the first data set and the second data set into a training group, a valid group, and a test group in ascending order of increasing OOD indexes of the plurality of semiconductors ” [Mental process] – splitting datasets into multiple groups based on the index in a specific order, which is just an act of splitting items by comparing/sorting them using their indexes which is capable of being performed in the human mind with the assistance of paper and pen. Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception? The claim 5 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea. Regarding Claim 6 Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03 Claim 6 is a dependent claim of 3, thus it falls within the same category of statutory subject matter. Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II. Regarding dependent claim 6, following limitations recite a judicial exception: “ the performing of the data split includes sampling the first data set and the second data set into a training group, a valid group, and a test group of the first learning data set in descending order of decreasing OOD indexes of the plurality of semiconductors ” [Mental process] – splitting datasets into multiple groups based on the index in a specific order, which is just an act of splitting items by comparing/sorting them using their indexes which is capable of being performed in the human mind with the assistance of paper and pen. “ the performing of the data split includes sampling the first data set and the second data set into a training group, a valid group, and a test group of the second learning data set in ascending order of increasing OOD indexes of the plurality of semiconductors ” [Mental process] – splitting datasets into multiple groups based on the index in a specific order, which is just an act of splitting items by comparing/sorting them using their indexes which is capable of being performed in the human mind with the assistance of paper and pen. Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception? The claim 6 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea. Regarding Claim 7 – integration into a practical application (sampling [Wingdings font/0xE0] extract indicators from to select best algorithm) Regarding Claim 8 Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03 Claim 8 is a dependent claim of 1, thus it falls within the same category of statutory subject matter. Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II. Regarding dependent claim 8, following limitations recite a judicial exception: extracting a first evaluation indicator, from each of the plurality of AI algorithms, by applying test data classified as a test group, of the at least one learning data set in the first data set and the second data set, to each of the plurality of AI algorithms [ Mental process] – applying datasets to the algorithms and extracting/observing results or performances is simply an action of observing/analyzing which capable of being performed with the assistance of pen and paper. extracting a second evaluation indicator from each of the plurality of AI algorithms, by applying test data classified as a test group, of the at least one learning data set in the first data set and the second data set, to each of the plurality of AI algorithms [ Mental process] – applying datasets to the algorithms and extracting/observing results or performances is simply an action of observing/analyzing which capable of being performed with the assistance of pen and paper. determining a ratio of the second evaluation indicator to the first evaluation indicator for each of the plurality of AI algorithms [ Mathematical calculation] – simply dividing or finding ratio of two numbers/indicators recites to an abstract idea. selecting an AI algorithm having a smallest product of the first evaluation indicator and the ratio from among the plurality of AI algorithms as the optimal AI algorithm [ Mental process, Mathematical calculation] - calculating a product of two numbers is merely a multiplication of numbers and selecting the smallest number from a group of numbers is a mere comparison between numbers which is practically capable of being performed with the assistance of pen and paper. Step 2A Prong 2 The claim 8 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea. Regarding Claim 9 Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03 Claim 9 is a dependent claim of 8, thus it falls within the same category of statutory subject matter. Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II. Regarding in/dependent claim 9, following limitations recite a judicial exception: extracting the first evaluation indicator within the learning area determined by an OOD index of a trained semiconductor and a first evaluation indicator outside the learning area determined by the OOD index of the trained semiconductor [ Mental process] - simply taking a record of the indicator or a number by comparing the indexes of the trained semiconductor or their data in/outside the learning area which is an action of observing/analyzing the data according to the OOD index. selecting a first AI algorithm having a smallest product of the first evaluation indicator within the learning area and the ratio, and a second AI algorithm having a smallest product of the first evaluation indicator outside the learning area and the ratio as the optimal AI algorithms according to an OOD index of a semiconductor to be predicted [ Mental process and mathematical calculation] - calculating a product of two numbers is merely a multiplication of numbers and selecting the smallest number from a group of numbers is a mere comparison between numbers which is practically capable of being performed with the assistance of pen and paper. Step 2A Prong 2 The claim 9 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea. Regarding Claim 10 Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03 Claim 10 is a method claim thus it falls into one of the four categories of statutory subject matter. Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II. Regarding independent claim 10, following limitations recite a judicial exception: receiving spectrum data indicating information of an actually measured spectrum of each of a plurality of semiconductors [ Mental process] - receiving data which can be considered as an action of looking/listening/recording which is capable of being performed with the assistance of pen and paper. generating a plurality of off of distribution (OOD) indexes by determining an OOD index for the spectrum of each of the plurality of semiconductors [ Mathematical Calculation] - comparing two points/vectors from the data set to evaluate indexes for each semiconductor comprises pure mathematical computations that recites to an abstract idea. predicting, from the spectrum data and using the AI algorithm, a structure of a semiconductor, of the plurality of semiconductors, when an OOD index of the semiconductor is smaller than a reference value for the AI algorithm [ Mental process, Mathematical relationship] – going through a sequence of steps as the algorithm, comparing the two numbers, the index and the reference value, and predicting/mapping the spectrum data in to the structure which is capable of being performed with the assistance of pen and paper. providing an optimal AI algorithm, among a plurality of AI algorithms that have been trained, as the AI algorithm predicting the structure of the semiconductor when the OOD index of the semiconductor is greater than or equal to the reference value [ Mental process, Mathematical calculation] – simply comparing the two numbers, the index and the reference value of the semiconductors, and selecting the one that is below the threshold which is an action of observing/analyzing the results which is capable of being performed with the assistance of pen and paper. [Mathematical calculation] – training the data set of an optimal algorithm involves mathematical calculation that recites to an abstract idea. Step 2A Prong 2 The claim 10 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea. Regarding Claim 11 Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03 Claim 11 is a dependent claim of 10, thus it falls within the same category of statutory subject matter. Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II. Regarding dependent claim 11, following limitations recite a judicial exception: calculating the plurality of OOD indexes based on the first data set [Mathematical calculation] - comparing two points/vectors from the data set to evaluate indexes for each semiconductor comprises pure mathematical computations that recite to an abstract idea. performing a data split on the first data set and the second data set, by cluster sampling the first data set and the second data set into at least one learning data set with respect to the OOD indexes according to the plurality of semiconductors [Mental process] - splitting datasets into multiple groups based on the index, which is just an act of splitting items that is capable of being performed in the human mind with the assistance of paper and pen. Step 2A Prong 2 Regarding Claim 11, it recites an additional element of “loading a first data set including the spectrum data and a second data set including structure measurement data representing information of actually measured structures of the plurality of semiconductors respectively having the plurality of OOD indexes” Loading data to a computer component is merely data gathering recited at a high level of generality, thus is insignificant extra-solution activity (See MPEP 2106.05(g)). [Even when viewed in combination, the additional elements do not more than automate the mental processes that a person could perform, using computer components as a tool, thus the claim as a whole does not integrate into a practical application.] Step 2B The claim does not provide an inventive concept (significantly more than the abstract idea). The claim is ineligible. As explained above, the additional element [1] is considered an insignificant extra solution activity and at best the equivalent of a mere data gathering recited at a high level of generality and amount to receiving or transmitting data over a network, which is well-understood, routine, conventional activity (See MPEP 2106.05(d), subsection II). This limitation remains insignificant extra-solution activity even upon reconsideration. Even when considered in combination, the additional element represents insignificant extra-solution activity, which cannot provide an inventive concept. Regarding Claim 12-15 Claims 12,13 and15 have similar limitations of Claims 2, 6, 8 respectively. For the reasons described above with respect to Claim 2, 6, 8, these judicial exceptions are not meaningfully integrated into a practical application, or significantly more than the abstract ideas. The claims do not provide anything more than the abstract ideas of mental processes and mathematical calculations that are practically capable of being performed with the assistance of pen and paper. Therefore, claims 12, 13 and 15 also recite abstract ideas that do not integrate into a practical application or amount to significantly more than judicial exception, and thus are rejected under U.S.C. 101. Regarding Claim 16 Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03 Claim 16 is a method claim thus it falls into one of the four categories of statutory subject matter. Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II. Regarding independent claim 16, following limitations recite a judicial exception: “determining an out of distribution (OOD) index with respect to the spectrum of the semiconductor for each of the at least one semiconductor based on the first data set” [Mathematical Calculation] – comparing two points/vectors from the data set to evaluate indexes for each semiconductor comprises mathematical computations that recite to an abstract idea. “ performing a data split on the first data set and the second data set by cluster sampling the first data set and the second data set into at least one learning data set with respect to the OOD index according to the at least one semiconductor” [Mental process] - splitting datasets into multiple groups based on the index, which is just an act of splitting items that is capable of being performed in the human mind with the assistance of paper and pen. “providing an optimal AI algorithm among a plurality of AI algorithms that have been trained on the at least one learning data set” [Mental process] - providing/selecting an optimal AI algorithm is purely comparing the performance of algorithms which is capable of being performed with the assistance of pen and paper. [Mathematical calculation] – training the data set of an optimal algorithm involves mathematical calculation that recites to an abstract idea. Step 2A Prong 2 Regarding Claim 1, the claim recites additional elements of “a memory storing instructions for executing a method of providing an artificial intelligence (AI) algorithm A memory to store instructions or code-based information is at best the equivalent of merely adding the words apply it to the judicial exception (See MPEP 2106.05(f)). A processor configured to execute the instructions, wherein the processor is configured to, by executing the instructions A processor to execute the instructions or code-based information is at best the equivalent of merely adding the words apply it to the judicial exception (See MPEP 2106.05(f)). “loading a first data set and a second data set, the first data set representing a spectrum of at least one semiconductor, and the second data set representing a structure of the at least one semiconductor”, Loading data to a computer component is merely data gathering recited at a high level of generality, thus is insignificant extra-solution activity (See MPEP 2106.05(g)). [Even when viewed in combination, the additional elements do not more than automate the mental processes that a person could perform, using computer components as a tool, thus the claim as a whole does not integrate into a practical application.] Step 2B The claim does not provide an inventive concept (significantly more than the abstract idea). The claim is ineligible. As explained above, the additional elements [1, 2] are merely computer components that are just to store and execute code-based instructions which are considered a mere instruction to apply an exception and amount to storing and receiving information in memory, which is well-understood, routine, conventional activity (See MPEP 2106.05(d), subsection II). This limitation remains a mere instruction to apply an exception. The additional element [3] is an insignificant extra-solution activity and at best the equivalent of a mere data gathering recited at a high level of generality and amount to receiving or transmitting data over a network, which is well-understood, routine, conventional activity (See MPEP 2106.05(d), subsection II). This limitation remains insignificant extra-solution activity even upon reconsideration. Even when considered in combination, the additional elements represent mere instruction to apply to an exception and an insignificant extra-solution activity, which cannot provide an inventive concept. Regarding Claim 17-20 Claims 17-20 have similar limitations of Claims 2, 3, 8 and 9 respectively. For the reasons described above with respect to Claim 2, 3, 8 and 9 these judicial exceptions are not meaningfully integrated into a practical application, or significantly more than the abstract ideas. The claims do not provide anything more than the abstract ideas of mental processes and mathematical calculations that are practically capable of being performed with the assistance of pen and paper or using computer components as a tool. Therefore, claims 17-20 also recite abstract ideas that do not integrate into a practical application or amount to significantly more than judicial exception, and thus are rejected under U.S.C. 101. Claim Rejections - 35 USC § 103 07-06 AIA 15-10-15 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. 07-20-aia AIA 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. 07-23-aia AIA 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. 07-21-aia AIA Claim s 1 and 16 are rejected under 35 U.S.C. 103 as being unpatentable over Ivanov et al. ( Ivanov ), WO2022/231837 in view of David et al. ( David), US Patent, US 10,430,719 B2 . As to independent Claim 1, Ivanov teaches a method of providing an artificial intelligence (AI) algorithms, the method comprising: loading a first data set and a second data set, the first data set representing a spectrum of at least one semiconductor, and the second data set representing a structure of the at least one semiconductor ( Ivanov, Pg3[0011] “A machine learning based measurement model must be trained to generate useful estimates of parameters of interest for a particular measurement application. Generally, model training is based on raw measurement signals collected from a specimen having known values of the parameters of interest(i.e., Design of Experiments (DOE) data)”, Abstract, Lines1- 2: “Methods and systems for measurements of semiconductor structures based on a trained parameter…” Pg4, [0016], Lines1-2, “A parameter conditioned measurement model is trained to predict the shape parameterization of a measured structure from DOE measurement data and a set of known DOE values”, wherein the raw measurement signals of DOE data, which the data includes both known spectrum data, which corresponds to the first data set, and known structure data, which corresponds to the second data set, of semiconductors, is used for training, thus it is equivalent to collecting two data sets for the training); determining an out of distribution (OOD) index with respect to the spectrum of the semiconductor for each of the at least one semiconductor based on the first data set( Ivanov, Abstract, “A trained parameter conditioned measurement model predicts a set of values of each non-conditioning parameter based on measurement data and a corresponding set of predetermined values for each conditioning parameter”, Pg4[0015] “The shape of a measured structure is characterized by a geometric model parameterized by one or more conditioning parameters and one or more non-conditioning parameters” , wherein conditioning parameters (the spectrum data) that tag measurement inputs and condition model inference. Functionally, these conditioning parameters serve as an index indicating whether an input lies within the DOE-trained domain, and therefore correspond to the OOD index); performing a data split on the first data set and the second data set by cluster sampling the first data set and the second data set into at least one learning data set with respect to the OOD index according to the at least one semiconductor( Ivanov, Pg8[0049] “In another aspect, a DOE training dataset of measurement data, SjD0E, is generated based on a DOE set of geometric model parameter values including both conditioning and non-conditioning parameters. In this example, the range of index, i, is 1:N, where N is any non-negative integer value. N is the number of discrete points where the shape of the structure is specified in the DOE dataset by the known values of both the conditioning and non-conditioning parameters”, wherein discrete points are inherently meaning partitioning of the data set based on the conditioning and non-conditioning parameters where the discrete or partitioned data sets can be one learning data set, which is inherently equivalent to data split using the OOD index); Ivanov, however, does not teach providing an optimal AI algorithm among a plurality of AI algorithms that have been trained on the at least one learning data set. In the same field of endeavor, David teaches this limitation ( David, Pg25, Column 14, Lines51-54, “In some embodiments, a set of algorithms can be trained simultaneously with the same input and target dataset. The algorithm that gives the best output can be selected for deployment”, wherein it discloses multiple algorithms can be trained and the best or the optimal algorithm can be deployed or provided which is inherently equivalent to the claimed invention). Ivanov and David are analogous to the claimed invention as both are from the same field of endeavor of semiconductor metrology. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to modify the provision of the optimal algorithm from the algorithm library of Ivanov to deploying the optimal algorithm from the trained algorithms based on the input data of David. The motivation is as recited by David (David, Pg20, Overview, Column3, Lines13-21, “machine learning algorithms can be used to create new approaches to data analysis by incorporating new types of input data, and the data can be more effectively correlated, organized and pre-processed, then used to make process adjustments. Data from prior production runs can be used to create a model for a target parameter, and data from a current production run can be input to the model to generate a prediction for the target parameter, and to correlate the prediction with the actual data”) such that training multiple algorithms simultaneously can efficiently reduce the time needed to adjust one single algorithm to adjust to the environment where conditions or parameters rapidly vary or change. As to independent Claim 16, it is a device claim of Claim 1 which contains similar limitations of Claim 1 and thus rejected under the same rationale . 07-21-aia AIA Claim s 2 and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Ivanov and David, as discussed above in Claim 1, in view of Guang-qi et al. ( Guang ), Chinese Patent CN-102039277A, English Translated Version, and further in view of Zhang et al ( Zhang), NPL, “Evaluation of Similarity measure methods for hyperspectral remote sensing data”, published on 2012, Pages: 4 . As to dependent Claim 2, The combination of Ivanov and David, as discussed above, teaches all the limitations of Claim 1. However, the combination does not teach the following limitations but Guang teaches the method of claim 1, wherein the determining OOD index includes 1) extracting a first principal component and a second principal component, with respect to the first data set, by performing principal component analysis(PCA) to reduce a dimension of the first data set ( Guang, Abstract, Lines3-5 “WAT parameters are screened from the WAT data of the known wafers; the screened WAT parameters are subject to orthogonal analysis to obtain principal components which are selected”, pg8 Paragrah3 Line8 “Using this dimensionality reduction idea, a principal component analysis was generated, that is, through dimensionality reduction, several synthesis factors (principal components) were found to represent many variables”, wherein the example here uses the WAT information for the PCA procedure, but it can easily be replaced with the spectrum data as Ivanov collects the both spectrum and structure data, which is thus equivalent to the claimed invention. ) ; The combination of Ivanov and David also does not teach 2) determining, for each of the at least one semiconductor, a value based on a Euclidean distance between vectors and a cosine distance between the vectors and a cosine distance between vectors and an origin, the vectors including the first principal component and the second principal component; 3) extracting a normalized value as the OOD index by normalizing an average value of a product of Euclidean distance and the cosine distance for each of the at least one semiconductor. Guang teaches that the screened parameters are subject to orthogonal analysis to extract principal components that the principal component analysis (PCA) is a specific type of the orthogonal analysis. However, Guang does not teach to use these dimension-reduced components or vectors to calculate Euclidean distances and cosine distances. In the same field of endeavor, Zhang teaches 2) determining, for each of the at least one semiconductor, a value based on a Euclidean distance between vectors and a cosine distance between the vectors and a cosine distance between vectors and an origin, the vectors including the first principal component and the second principal component ( Zhang, subsection2.1, “According to the shape and amplitude features of the hyperspectral data, we choose five widely used similarity measure methods for evaluation. These methods can be divided into four types. The first type is sensitive to spectral amplitude, such as the Euclidean distance (ED) [6]. The second type is sensitive to spectral shape, such as the spectral angel cosine (SAC) [7] and the spectral correlation coefficient (SCC)”, wherein computing distances between vectors or components frequently/widely adopt these two distances ) . 3) extracting a normalized value as the OOD index by normalizing an average value of a product of Euclidean distance and the cosine distance for each of the at least one semiconductor ( Zhang, Abstract, Lines6-13 ,“The results indicate that the spectral angle cosine-Euclidean distance method demonstrates the most superior ability to distinguish various land cover types among five methods because it fully utilizes both the spectral amplitude and shape feature in the hyperspectral data. A combination of the spectral amplitude-sensitive method and the shape-sensitive method will effectively improve the identification accuracy of different land cover types”, conclusion Line18-21, “However, this method only considers the product of the spectral angle cosine and the Euclidean distance. In certain case, this integration is more sensitive to the spectral profile shape of the hyperspectral data, wherein unified testing framework can be interpreted as normalized values of the product of the two distances and the product is used to effectively improve the identification accuracy which inherently encompasses that it can be used as indexing purposes). Ivanov, David, Guang and Zhang are analogous to the claimed invention as they are from the same field of endeavor of hyperspectral image analysis. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to combine Ivanov and David’s parameter-conditioned spectral inference and Guang ’s PCA extraction with the hyperspectral analysis using spectral angle cosine-Euclidean distance(SAC-ED) of Zhang . The motivation to combine is as recited by Guang (Pg8, paragraph3, Lines 4-12, “In most practical problems, there is a certain correlation between variables, so that the information reflected by the observed data overlaps to a certain extent. People naturally want to replace the original variables with fewer variables, and these fewer variables reflect the information of the original variables as much as possible. Using this dimensionality reduction idea, a principal component analysis was generated, that is, through dimensionality reduction, several synthesis factors (principal components) were found to represent many variables, so that these synthesis factors could reflect the information of the original variables as much as possible”) and Zhang (Abstract, Line8-13, “method demonstrates the most superior ability to distinguish various land cover types among five methods because it fully utilizes both the spectral amplitude and shape feature in the hyperspectral data. A combination of the spectral amplitude-sensitive method and the shape-sensitive method will effectively improve the identification accuracy of different land cover types”) such that the synergistic combination of Euclidean and cosine distances within a unified framework maximizes discrimination accuracy by simultaneously capturing both the magnitude (amplitude) and the directional (shape) characteristics inherent ins spectral data, a principle that remains fundamentally applicable even within a dimensionally reduced PCA feature space. As to dependent Claim 17, it is a device claim of Claim 2 which contains similar limitations of Claim 2 and thus rejected under the same rationale . 07-21-aia AIA Claim s 3, 4, 5, 6 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Ivanov and David as discussed in Claim 1 above and in view of Zhi-huang et al. ( Zhihuang) , Chinese patent, CN-111178399A, English Translated Version . As to dependent Claim 3, The combination of Ivanov and David teaches all the limitations as described in Claim 1 above including determining the OOD indexes for the spectrum data and further teaches the following limitation of the method of claim 1, wherein the at least one semiconductor includes a plurality of semiconductors ( Ivanov , Technical Field , “ The described embodiments relate to metrology systems and methods, and more particularly to methods and systems for improved measurement of semiconductor structures “, Pg8[0049] “The range of index, j, is 1:M, where M is any non-negative integer value. M is the number of measurements collected from DOE structures characterized by the known values of both the conditioning and non-conditioning parameters at I discrete points ). Ivanov and David, however, do not teach the performing of the data split includes sampling the first data set and the second data set at a preset ratio into a training group, a valid group, and a test group of each allocated learning data set according to the OOD indexes of the plurality of semiconductors. In the same field of endeavor, Zhihuang teaches this limitation ( Zhihuang, Abstract, “obtaining the relevancy of the features according to the feature information of the object; clustering the objects according to the relevancy of the features to obtain a clustering result, wherein the objects with the small proportion in the clustering result serve as minority class samples, the objects with the large proportion in the clustering result serve as majority class samples, the ratio of the number of the minority class samples to the number of the majority class samples is 1: N, and N is the data imbalance multiplying power and is a positive integer larger than 1” and Pg2 Claim1 “The ratio of the number of minority samples to the number of majority samples is 1:N, N is the data imbalance rate and a positive integer greater than 1”, wherein clustering objects by feature relevance and using the resulting clusters and synthesized minority samples to train a model. Functionally, the clustering output can be partitioned into training, validation, and test sets (e.g., by random split of the majority cluster or hierarchical clustering to produce three clusters), there by corresponding to the three-group data split.) Ivanov, David and Zhihuang are analogous to the claimed invention as they are from the same field of endeavor of collecting and partitioning the data into multiple groups according to an indicator for AI model trainings. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to combine the DOE-conditioned measurement models for semiconductor structure prediction of Ivanov with feature-based partitioning and synthetic-minority augmentation for constructing training datasets of Zhihuang. The motivation to combine is to as recited by Zhihuang (Zhihuang, Abstract, “Through the scheme provided by the embodiment of the invention, the sample can be automatically labeled, and a lot of manpower and material resources are saved”, Pg7, paragraph4, Line11-13, “By adding minority samples, the problem of imbalance of data categories can be solved to ensure that the information of minority samples is not covered”) such that feature-relevance clustering and synthetic minority augmentation enable automated partitioning and balanced training set construction, improving representativeness and robustness of learned models while reducing manual labeling effort. As to dependent Claim 4, Zhihuang teaches the method of claim 3, wherein the performing of the data split includes sampling the first data set and the second data set into the training group, the valid group, and the test group of learning data set, sequentially according to the OOD indexes of the plurality of semiconductors ( Zhihuang, Pg3, Claim4, Lines4-11, “Arrange the features in descending order according to the relevance of the features to obtain an ordered feature sequence; Select the first M1 features from the ordered feature sequence, where M1 is a positive integer greater than or equal to 1; Clustering the object based on the first M1 features; If after clustering the objects based on the first M1 features, the ratio of the number of objects with a small proportion to the number of objects with a large proportion is 1:N, then the clustering result of the object is used as the clustering result”, pg22, paragraph5, Lines4, “Each time a new feature in the ordered feature sequence is added sequentially”, wherein ordering features by relevance and performing clustering on the top M1 features, with iterative expansion to M2 features if the resulting cluster proportion do not meet the required 1:N ratio. This stepwise feature-selection and clustering procedure corresponds to the feature-based data partitioning). As to dependent Claim 5, Zhihuang teaches the method of claim 3, wherein the performing of the data split includes sampling first data set and the second data set into the training group, the valid group and the test group in ascending order of increasing OOD indexes according to the plurality of semiconductors ( Zhihuang, Pg3, Claim4, Lines4-11, “Arrange the features in descending order according to the relevance of the features to obtain an ordered feature sequence; Select the first M1 features from the ordered feature sequence, where M1 is a positive integer greater than or equal to 1; Clustering the object based on the first M1 features; If after clustering the objects based on the first M1 features, the ratio of the number of objects with a small proportion to the number of objects with a large proportion is 1:N, then the clustering result of the object is used as the clustering result”, and Pg21 , paragraph1, Lines1-5, “It should be noted that, here is an example of sorting in descending order according to the relevance of the features, but in other embodiments, it is also possible to sort them in ascending order according to the relevance of the features, but the features are selected for clustering in the following steps When, you can select the last M1 features or the last M2 features”, wherein arranging features by relevance and selecting feature subsets for clustering. The specification explicitly states that although an example uses descending order, in other embodiments features may be sorted in ascending order and, in that case, the last M1 or last M2 features are selected, which is inherently equivalent to producing feature subsets from both descending and ascending orderings, corresponding to the two training sets). As to dependent Claim 6, Zhihuang teaches the method of claim 3, wherein the at least one learning data set includes a first learning data set and a second learning data set, and performing of the data split includes sampling the first data set and the second data set into a training group, a valid group, and a test group of the first learning data set in descending order of decreasing OOD indexes according to the plurality of semiconductors; and sampling the first data set and the second data set into a training group, a valid group, and a test group of the second learning data set in ascending order of increasing OOD indexes according to the plurality of semiconductors ( Zhihuang, Pg3, Claim4, Lines4-11, “Arrange the features in descending order according to the relevance of the features to obtain an ordered feature sequence; Select the first M1 features from the ordered feature sequence, where M1 is a positive integer greater than or equal to 1; Clustering the object based on the first M1 features; If after clustering the objects based on the first M1 features, the ratio of the number of objects with a small proportion to the number of objects with a large proportion is 1:N, then the clustering result of the object is used as the clustering result”, and Pg21 , paragraph1, Lines1-5, “It should be noted that, here is an example of sorting in descending order according to the relevance of the features, but in other embodiments, it is also possible to sort them in ascending order according to the relevance of the features, but the features are selected for clustering in the following steps When, you can select the last M1 features or the last M2 features”, wherein arranging features by relevance and selecting feature subsets for clustering. The specification explicitly states that although an example uses descending order, in other embodiments features may be sorted in ascending order and, in that case, the last M1 or last M2 features are selected, which is inherently equivalent to producing feature subsets from both descending and ascending orderings, corresponding to the two learning data sets in both ascending anhd descending orders). As to dependent Claim 18, it is a device claim of Claim 3 which contains similar limitations of Claim 3 and thus rejected under the same rationale . 07-21-aia AIA Claim 7 is rejected under 35 U.S.C. 103 as being unpatentable over Ivanov , David and Zhihuang as discussed in Claim 6 (which depends on Claims 1 and 3) above and in further view of Donghua et al. ( Donghua), WIPO Application, WO-2017/161646-A1, English Translated Version . As to dependent Claim 7, The combination of Ivanov, David and Zhihuang teaches , as discussed above, all the limitations of Claims 1, 3 and 6. The combination teaches about the two training data sets in both ascending and descending orders where each training data set contains 3 clusters as discussed in Claim 6. But the combination fails to teach the following limitations but Donghua teaches the method of claim 6, wherein the providing of the optimal AI algorithm includes extracting a first evaluation indicator for each of first valid data and first test data from each of the plurality of AI algorithms, by applying the first valid data classified into the valid group of the first learning data in the first data set and the second data set and the firs test data classified into the test group of the first learning data set in the first data set and the second data set to each of the plurality of AI algorithms. However, Donghua teaches this limitation ( Donghua, Pg10, algorithm 3, “(RMSE) is calculated, and then a function of inverse trend is designed according to the size of the error root mean square, and the weight is given to each model”, Pg12, paragraph4, “Firstly, the training data is put into the general model to train and forecast the data and error data of each model”, “with the prediction error of the training model, the prediction error of the training set and the prediction error of the prediction set are calculated respectively” , wherein computing RMSE for each candidate model and extracting per-model error data from training/validation runs (here, the example uses the training data but with the combination of Zhihuang mentioned above can interpret the training data as the two groups of the three are being used for computing RMSE for each group respectively); these extracted error metrics are used to design inverse-trend weighting functions and to select the optimal weighting algorithm by comparing predicted sets to a verification set, thereby it is equivalent to the extracting step of the claimed invention); extracting a second evaluation indicator for each of second valid data and second test data from each of the plurality of AI algorithms, by applying the second valid data classified into the valid group of the second learning data set in the first data set and the second data set and the second test data classified into the test group of the second learning data set in the first data set and the second data set to each of the plurality of AI algorithms( Donghua, Pg10, algorithm 3, “(RMSE) is calculated, and then a function of inverse trend is designed according to the size of the error root mean square, and the weight is given to each model”, Pg12, paragraph4, “Firstly, the training data is put into the general model to train and forecast the data and error data of each model”, “with the prediction error of the training model, the prediction error of the training set and the prediction error of the prediction set are calculated respectively” , wherein the same rationale of the extraction of the first evaluation indicator above applies but to the second learning data set, which was mentioned above in Claim 6, that computing RMSE for each candidate model and extracting per-model error data from training/validation runs; these extracted error metrics are used to design inverse-trend weighting functions and to select the optimal weighting algorithm by comparing predicted sets to a verification set, thereby it is equivalent to the extracting step of the claimed invention); and selecting the optimal AI algorithm from among the plurality of AI algorithms, based on a valid root mean square error (RMSE) for the first valid data, a test RMSE for the first test data, a valid RMSE for the second valid data, and a test RMSE for the second test data( Donghua, Pg11, paragraph1, Lines1-2, “the optimal weight algorithm selection algorithm selects the best weight algorithm”, wherein computing per-model error metrics (RMSE) from training/validation runs, transforms those errors into weights via an inverse-trend weighting function, forms ensemble predictions under multiple weighting algorithms and comparing each ensemble’s validation error and selects the weighting algorithm that yields the minimum error is conceptually equivalent to the selection of the optimal algorithm of the claimed invention). The motivation to combine, before the effective filing date of the claimed invention, is to as recited by Donghua (Donghua, Pg10, algorithm 3, “(RMSE) is calculated, and then a function of inverse trend is designed according to the size of the error root mean square, and the weight is given to each model”, Pg3, paragraph4,5, “The data set predicted by the weight library is used to compare with the verification set to get the error. By the minimum error, the optimal weight algorithm is obtained”) such that computing per-model RMSE and maps RMSE to model weights via an inverse-trend function; RMSE is used because it penalizes large errors, provides a comparable scalar performance metric across models, and yields continuous values suitable for weight-function design, the weighted ensembles are then evaluated on a verification set and the weighting algorithm with minimum validation RMSE is selected as optimal, which enhances the accuracy of model-ran results . 07-21-aia AIA Claim s 8 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Ivanov and David in view of Donghua et al. ( Donghua), WIPO Application, WO-2017/161646-A1, English Translated Version . As to dependent Claim 8, The combination of Ivanov and David teaches all the limitations of Claim 1 as discussed above including performing data split using discrete points based on the conditioned or non-conditioned parameters but Ivanov fails to teach the following limitations but Donghua teaches the method of claim 1, wherein the providing of the optimal AI algorithm includes extracting a first evaluation indicator, from each of the plurality of AI algorithms, by applying test data classified as a test group, of the at least one learning data set in the first data set and the second data set, to each of the plurality of AI algorithms( Donghua, Pg10, algorithm 3, “(RMSE) is calculated, and then a function of inverse trend is designed according to the size of the error root mean square, and the weight is given to each model”, Pg12, paragraph4, “Firstly, the training data is put into the general model to train and forecast the data and error data of each model”, “with the prediction error of the training model, the prediction error of the training set and the prediction error of the prediction set are calculated respectively” , wherein computing RMSE for each candidate model and extracting per-model error data from training/validation runs (here, the example uses the training data but as mentioned in Claim 1 by Ivanov , the partitioned data sets can include these test/valid groups as the claimed invention groups them according to the OOD indexes or here by the conditioned or non-conditioned parameters); these extracted error metrics are used to design inverse-trend weighting functions and to select the optimal weighting algorithm by comparing predicted sets to a verification set, thereby it is equivalent to the extracting step of the claimed invention); extracting a second evaluation indicator from each of the plurality of AI algorithms, by applying valid data classified into a valid group, of the at least one learning data set in the first data set and the second data set, to each of the plurality of AI algorithms( Donghua, Pg10, algorithm 3, “(RMSE) is calculated, and then a function of inverse trend is designed according to the size of the error root mean square, and the weight is given to each model”, Pg12, paragraph4, “Firstly, the training data is put into the general model to train and forecast the data and error data of each model”, “with the prediction error of the training model, the prediction error of the training set and the prediction error of the prediction set are calculated respectively” , wherein the same rationale above applies such that computing RMSE for each candidate model and extracting per-model error data from training/validation runs (here, the example uses the training data but as mentioned in Claim 1 by Ivanov , the partitioned data sets can include these test/valid groups as they are just grouped according to the OOD indexes or here by the conditioned or non-conditioned parameters); these extracted error metrics are used to design inverse-trend weighting functions and to select the optimal weighting algorithm by comparing predicted sets to a verification set, thereby it is equivalent to the extracting step of the claimed invention); determining a ratio of the second evaluation indicator to the first evaluation indicator for each of the plurality of AI algorithms ( Donghua, Pg6, Algorithm2, “Algorithm 2: Excludes the relatively poor model of 20% predicted results and give the same weight to the rest of the model”, Pg12, paragraph4, Lines6-7, “Respectively, to the three values of a weight, take 0.3,0.3,0.4. And finally get the comprehensive error value”, wherein determining selection and weighting ratios – for example, excluding the worst 20% of the model results (Algorithm2) and assigning ensemble weights such as 0.3, 0.3, 0.4 – then evaluating resulting ensembles on a verification set to select the best configuration, which is inherently equivalent to the determining a ratio of the two indicators of the claimed invention); selecting an AI algorithm having a smallest product of the first evaluation indicator and the ratio from among the plurality of AI algorithms as the optimal AI algorithm ( Donghua, Pg6, Algorithm2, “Algorithm 2: Excludes the relatively poor model of 20% predicted results and give the same weight to the rest of the model”, Pg3, paragraph4,5, “The data set predicted by the weight library is used to compare with the verification set to get the error. By the minimum error, the optimal weight algorithm is obtained”, Pg11, paragraph4, Lines4-7, “the prediction error of the training model, the prediction error of the training set and the prediction error of the prediction set are calculated respectively. Respectively, to the three values of a weight, take 0.3,0.3,0.4. And finally get the comprehensive error value”, wherein computing per-model error metrics (RMSE), applying selection/weighting ratios, forming ensemble predictions under those ratios, and selecting the weighting/selection algorithm that yields the minimum validation error, which effectively selects the configuration that minimizes the validation error resulting from the indicator-to-ratio combination and this is inherently equivalent to selecting the optimal algorithm using the smallest product of the claimed invention). Ivanov, David and Donghua are analogous to the claimed invention as they are from the same field of endeavor of selecting and providing the optimal algorithm based on the weight of the AI algorithms. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to combine the DOE-conditioned measurement models for semiconductor structure prediction of Ivanov with the multi-model weighting/selection framework based on per-model error metrics (RMSE) of Donghua. The motivation to combine is to as recited by Donghua (Donghua, Pg10, algorithm 3, “(RMSE) is calculated, and then a function of inverse trend is designed according to the size of the error root mean square, and the weight is given to each model”, Pg3, paragraph4,5, “The data set predicted by the weight library is used to compare with the verification set to get the error. By the minimum error, the optimal weight algorithm is obtained”) such that computing per-model RMSE and maps RMSE to model weights via an inverse-trend function; RMSE is used because it penalizes large errors, provides a comparable scalar performance metric across models, and yields continuous values suitable for weight-function design, the weighted ensembles are then evaluated on a verification set and the weighting algorithm with minimum validation RMSE is selected as optimal, which enhances the accuracy of model-ran results. As to dependent Claim 19, it is a device claim of Claim 8 which contains similar limitations of Claim 8 and thus rejected under the same rationale . 07-21-aia AIA Claim s 9 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Ivanov, David and Donghua as discussed in Claims 1 and 8 above and further in view of Inkawhich et al. ( Inkawhich), NPL, “Improving Out-of-Distribution Detection by Learning From the Deployment Environment”, published on Jan 27 2022, pages: 17 . As to dependent Claim 9, The combination of Ivanov, David and Donghua teaches the first evaluation indicator such as RMSE from the at least one learning data set described above in Claim 8. However, the combination fails to teach the following limitations but Inkawhich teaches the method of claim 8, wherein the first evaluation indicator includes a first evaluation indicator within a learning area and a first evaluation indicator outside the learning area ( Inkawhich, Abstract, Lines1-5, “Recognition systems in the remote sensing domain often operate in “open-world” environments, where they must be capable of accurately classifying data from the in-distribution categories while simultaneously detecting and rejecting anomalous/out-of-distribution (OOD) inputs”, Pg3, Section III Methodology, subsection A, System overview, Lines11-16, “The OOD detector then observes the signal from the classifier and produces a real-valued ID/OOD score. If the ID/OOD score is greater than some predefined threshold, the input is predicted to be ID and the classifier releases its prediction. Else, the sample is deemed to be OOD and the system abstains from releasing any prediction”, Pg5, subsection C, OOD detection algorithm, paragraph2, Lines7-9, “O ODIN (x) is then compared to a predefined threshold B thresh . If O ODIN (x) > B thresh , x is deemed to be ID. Else, x is regarded as OOD” , wherein computing a real-valued ID/OOD score from the classifier output and applying a predefined threshold B thresh such that O ODIN (x) > B thresh is classified as inside the learning area (ID) and otherwise as outside (OOD), which is inherently equivalent to the evaluation indicator inside/outside of the learning area), extracting the first evaluation indicator within the learning area determined by an OOD index of a trained semiconductor and a first evaluation indicator outside the learning area determined by the OOD index of the trained semiconductor ( Inkawhich, Pg3, Section III Methodology, subsection A, System overview, Lines11-16, “The OOD detector then observes the signal from the classifier and produces a real-valued ID/OOD score. If the ID/OOD score is greater than some predefined threshold, the input is predicted to be ID and the classifier releases its prediction. Else, the sample is deemed to be OOD and the system abstains from releasing any prediction”, Pg5, subsection C, OOD detection algorithm, paragraph2, Lines7-9, “O ODIN (x) is then compared to a predefined threshold B thresh . If O ODIN (x) > B thresh , x is deemed to be ID. Else, x is regarded as OOD”, wherein computing a real-valued ID/OOD score from the classifier output and applying a predefined threshold B thresh such that O ODIN (x) > B thresh is classified as inside the learning area (ID) and otherwise as outside (OOD), which is inherently equivalent to the evaluation indicator inside/outside of the learning area). The combination of Donghua and Inkawhich teaches the providing of the optimal AI algorithm includes selecting a first AI algorithm having a smallest product of the first evaluation indicator within the learning area and the ratio, and a second AI algorithm having a smallest product of the first evaluation indicator outside the learning area and the ratio as the optimal AI algorithms according to an OOD index of a semiconductor to be predicted(as mentioned in Claim 8, Donghua teaches applying candidate ratio/weight configurations and selecting the configuration that minimizes verification error; as mentioned above Inkawhich teaches to compute a real-valued ID/OOD score and thresholding it to classify inside/outside status. Combining these teachings, the system can compute the first evaluation indicator (ID/OOD score), applies each candidate ratio to form a combined metric on a verification set, and selects the AI algorithm whose combined metric is smallest, which is conceptually equivalent to the claimed invention). Ivanov , David, Donghua and Inkawhich are analogous to the claimed invention as they are from the same field of endeavor of collecting data and use some kind of an indicator to cluster them so they can be used to train the AI models of metrology. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to combine the metrology prediction model of Ivanov and the multi-model weighting/selection framework based on per-model error metrics (RMSE) of Donghua with the threshold-based classifier using OOD/ID score of Inkawhich. The motivation to combine is to as recited by Inkawhich (Inkawhich, Pg3 Section III (Methodology), Subsection A (System overview), Lines11-16, “The OOD detector then observes the signal from the classifier and produces a real-valued ID/OOD score. If the ID/OOD score is greater than some predefined threshold, the input is predicted to be ID and the classifier releases its prediction”, Abstract, Lines13-16, “We then propose a novel ‘learn-online’ workflow for updating the DNNs during deployment using a small library of carefully collected samples from the operating environment” ) such that combination of a calibrated numeric score, thresholded decision rule, OOD-aware training, and a lightweight online update workflow delivers a practical, auditable, and cost-effective way to reduce erroneous predictions in real-world measurement systems. As to dependent Claim 20, it is a device claim of Claim 9 which contains similar limitations of Claim 9 and thus rejected under the same rationale . 07-21-aia AIA Claim 10 and 11 are rejected under 35 U.S.C. 103 as being unpatentable over Ivanov in view of Inkawhich et al. ( Inkawhich), NPL, “Improving Out-of-Distribution Detection by Learning From the Deployment Environment”, published on Jan 27 2022, pages: 17 . As to independent Claim 10, Ivanov teaches an operation method of a computer configured to operate an artificial intelligence (AI) algorithm, the method comprising: receiving spectrum data indicating information of an actually measured spectrum of each of a plurality of semiconductors ( Ivanov, Abstract, Lines1-2: “Methods and systems for measurements of semiconductor structures based on a trained parameter…”, Pg7, [0039], Lines11-13, “The radiation received by the spectrometer 104 in the collection beam 108 is analyzed with regard to polarization state, allowing for spectral analysis by the spectrometer of radiation passed by the analyzer”, Pg16 [0106], Lines10-12, “In one embodiment, computing system 130 determines values of one or more parameters of interest during process based on measured signals 111 received from a measurement system”, wherein the spectrometer actually collects the spectrum data to analyze); generating a plurality of off of distribution (OOD) indexes by determining an OOD index for the spectrum of each of the plurality of semiconductors( Ivanov, Abstract, “A trained parameter conditioned measurement model predicts a set of values of each non-conditioning parameter based on measurement data and a corresponding set of predetermined values for each conditioning parameter”, Pg4[0015] “The shape of a measured structure is characterized by a geometric model parameterized by one or more conditioning parameters and one or more non-conditioning parameters” , wherein conditioning parameters that tag measurement inputs and condition model inference. Functionally, these conditioning parameters serve as an index indicating whether an input lies within the DOE-trained domain, and therefore correspond to the OOD index); predicting, from the spectrum data and using the AI algorithm, a structure of a semiconductor, of the plurality of semiconductors, when an OOD index of the semiconductor is smaller than a reference value for the AI algorithm ( Ivanov, Abstract, Lines4-10, “A trained parameter conditioned measurement model predicts a set of values of each non-conditioning parameter based on measurement data and a corresponding set of predetermined values for each conditioning parameter. In this manner, the trained parameter conditioned measurement model predicts the shape of a measured structure. Although a parameter conditioned measurement model is trained at discrete geometric points of a structure, the trained model predicts values of non-conditioning parameters for any corresponding conditioning parameter value”, Pg4, [0017], Lines1-4, “Due to the continuous nature of the trained parameter conditioned measurement model, the trained model learns the shape of the measured structure at all values of the conditioning parameters within the range of DOE values of the conditioning parameters employed during training” ); Ivanov teaches predicting, from the spectrum data and using the AI algorithm, a structure of a semiconductor, of the plurality of semiconductors ( Ivanov, Abstract, Lines4-7, “a trained parameter conditioned measurement model predicts a set of values of each non-conditioning parameter based on measurement data and a corresponding set of predetermined values for each conditioning parameter” , wherein a trained parameter-conditioned measurement model that predicts structure parameters, non-conditioning parameters, from measurement/spectrum inputs, conditioning parameters ) . However, Ivanov does not teach of a model that predicts the structure of the semiconductor when an OOD index of the semiconductor is smaller than a reference value for the AI algorithm. In the same field of endeavor, Inkawhich teaches this limitation ( Inkawhich, Pg3 Section III (Methodology), Subsection A (System overview), Lines11-16, “The OOD detector then observes the signal from the classifier and produces a real-valued ID/OOD score. If the ID/OOD score is greater than some predefined threshold, the input is predicted to be ID and the classifier releases its prediction”, wherein computing an ID/OOD score and applying a predefined threshold such that samples are classified as within or outside the learning area ). Ivanov and Inkawhich are analogous to the claimed invention as both are from the same field of endeavor of machine learning for measurement and OOD detection in sensing systems for metrology. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to combine the metrology prediction model of Ivanov with the threshold-based classifier using OOD/ID score of Inkawhich. The motivation to combine is to as recited by Inkawhich (Inkawhich, Pg3 Section III (Methodology), Subsection A (System overview), Lines11-16, “The OOD detector then observes the signal from the classifier and produces a real-valued ID/OOD score. If the ID/OOD score is greater than some predefined threshold, the input is predicted to be ID and the classifier releases its prediction”, Abstract, Lines13-16, “We then propose a novel ‘learn-online’ workflow for updating the DNNs during deployment using a small library of carefully collected samples from the operating environment” ) such that combination of a calibrated numeric score, thresholded decision rule, OOD-aware training, and a lightweight online update workflow delivers a practical, auditable, and cost-effective way to reduce erroneous predictions in real-world measurement systems. As to dependent Claim 11, Ivanov and Inkawhich, as discussed in Claim 10, teach about collecting actually measured spectrum data and generating OOD indexes for the data. Ivanov further teaches the method of claim 10, further comprising: loading a first data set including the spectrum data and a second data set including structure measurement data representing information of actually measured structures of the plurality of semiconductors respectively having the plurality of OOD indexes ( Ivanov, Pg3[0011] “A machine learning based measurement model must be trained to generate useful estimates of parameters of interest for a particular measurement application. Generally, model training is based on raw measurement signals collected from a specimen having known values of the parameters of interest”, Abstract, Lines1-2: “Methods and systems for measurements of semiconductor structures based on a trained parameter…”, Pg7, [0039], Lines11-13, “The radiation received by the spectrometer 104 in the collection beam 108 is analyzed with regard to polarization state, allowing for spectral analysis by the spectrometer of radiation passed by the analyzer” and [0045], Lines2-5, “In one example, scanning electron microcopy (SEM) images are employed to directly measure the shape of a semiconductor structure and the measured shapes are employed to parameterize the geometric model”, wherein the raw measurement signals include both spectrum and structure data of a semiconductor. The spectrometer 104 is specifically used to collect the information of spectrum of a semiconductor. And as mentioned above about collecting actually measured spectrum data with using SEM to collect actually measured structure data is equivalent to loading the two data sets of the claimed invention); calculating the plurality of OOD indexes based on the first data set( Ivanov, Abstract, “A trained parameter conditioned measurement model predicts a set of values of each non-conditioning parameter based on measurement data and a corresponding set of predetermined values for each conditioning parameter”, Pg4[0015] “The shape of a measured structure is characterized by a geometric model parameterized by one or more conditioning parameters and one or more non-conditioning parameters” , wherein conditioning parameters (the spectrum data) that tag measurement inputs and condition model inference. Functionally, these conditioning parameters serve as an index indicating whether an input lies within the DOE-trained domain, and therefore correspond to the OOD index); performing a data split on the first data set and the second data set, by cluster sampling the first data set and the second data set into at least one learning data set with respect to the OOD indexes according to the plurality of semiconductors ( Ivanov, Pg8[0049] “In another aspect, a DOE training dataset of measurement data, SjD0E, is generated based on a DOE set of geometric model parameter values including both conditioning and non-conditioning parameters. In this example, the range of index, i, is 1:N, where N is any non-negative integer value. N is the number of discrete points where the shape of the structure is specified in the DOE dataset by the known values of both the conditioning and non-conditioning parameters”, wherein discrete points are inherently meaning partitioning of the data set based on the conditioning and non-conditioning parameters where the discrete or partitioned data sets can be one learning data set, which is inherently equivalent to data split using the OOD index); wherein the optimal AI algorithm was trained on the at least one learning data set ( Ivanov, Pg4[0016] “A parameter conditioned measurement model is trained to predict the shape parameterization of a measured structure from DOE measurement data and a set of known, DOE values of both the conditioning and the non-conditioning parameters at a number of discrete locations of the structure”, Pg3[0008] “the floating parameters are resolved by a search through a library of pre-computed solutions to find the closest match”, Pg6[0036] “The number of training samples required to achieve a high resolution predictive model is relatively small compared to a measurement model trained without conditioning parameters. Thus, a high resolution shape predictive model is trained from relatively a low resolution set of shape references using model conditioning as described herein”, Pg7[0041] “In a further embodiment, system 100 may include one or more computing systems 130 employed to perform measurements of structures based on measurement models developed in accordance with the methods described herein”, wherein the parameter-conditioned measurement model is trained on DOE measurement data at discrete DOE points or partitioned data where the clusters can be considered as one learning data set, and when a solution cannot be resolved directly the floating parameters may be resolved by searching a library of pre-computed solutions to find the closest match, which is conceptually equivalent to training the optimal algorithm and providing an optimal algorithm if it is not considered as the optimal) . 07-21-aia AIA Claim 12 is rejected under 35 U.S.C. 103 as being unpatentable over Ivanov and Inkawhich discussed above in Claim 11, in view of Guang and further in view Zhang . As to dependent Claim 12, Ivanov and Inkawhich teach all the limitations of Claim 11 as described above but they fail to teach the following limitations but Guang teaches the method of claim 1, wherein the determining OOD index includes 1) extracting a first principal component and a second principal component, with respect to the first data set, by performing principal component analysis(PCA) to reduce a dimension of the first data set ( Guang, Abstract, Lines3-5 “WAT parameters are screened from the WAT data of the known wafers; the screened WAT parameters are subject to orthogonal analysis to obtain principal components which are selected”, pg8 Paragrah3 Line8 “Using this dimensionality reduction idea, a principal component analysis was generated, that is, through dimensionality reduction, several synthesis factors (principal components) were found to represent many variables”, wherein the example here uses the WAT information for the PCA procedure, but it can easily be replaced with the spectrum data as Ivanov collects the both spectrum and structure data, which is thus equivalent to the claimed invention. ) ; Ivanov and Inkawhich also do not teach 2) determining, for each of the at least one semiconductor, a value based on a Euclidean distance between vectors and a cosine distance between the vectors and a cosine distance between vectors and an origin, the vectors including the first principal component and the second principal component; 3) extracting a normalized value as the OOD index by normalizing an average value of a product of Euclidean distance and the cosine distance for each of the at least one semiconductor. Guang teaches that the screened parameters are subject to orthogonal analysis to extract principal components that the principal component analysis (PCA) is a specific type of the orthogonal analysis. However, Ivanov and Inkawhich do not teach to use these dimension-reduced components or vectors to calculate Euclidean distances and cosine distances. In the same field of endeavor, Zhang teaches 2) determining, for each of the at least one semiconductor, a value based on a Euclidean distance between vectors and a cosine distance between the vectors and a cosine distance between vectors and an origin, the vectors including the first principal component and the second principal component ( Zhang, subsection2.1, “According to the shape and amplitude features of the hyperspectral data, we choose five widely used similarity measure methods for evaluation. These methods can be divided into four types. The first type is sensitive to spectral amplitude, such as the Euclidean distance (ED) [6]. The second type is sensitive to spectral shape, such as the spectral angel cosine (SAC) [7] and the spectral correlation coefficient (SCC)”, wherein computing distances between vectors or components frequently/widely adopt these two distances ) . 3) extracting a normalized value as the OOD index by normalizing an average value of a product of Euclidean distance and the cosine distance for each of the at least one semiconductor ( Zhang, Abstract, Lines6-13 ,“The results indicate that the spectral angle cosine-Euclidean distance method demonstrates the most superior ability to distinguish various land cover types among five methods because it fully utilizes both the spectral amplitude and shape feature in the hyperspectral data. A combination of the spectral amplitude-sensitive method and the shape-sensitive method will effectively improve the identification accuracy of different land cover types”, conclusion Line18-21, “However, this method only considers the product of the spectral angle cosine and the Euclidean distance. In certain case, this integration is more sensitive to the spectral profile shape of the hyperspectral data, wherein unified testing framework can be interpreted as normalized values of the product of the two distances and the product is used to effectively improve the identification accuracy which inherently encompasses that it can be used as indexing purposes). The motivation to combine, before the effective filing date of the claimed invention, is as recited by Guang (Pg8, paragraph3, Lines 4-12, “In most practical problems, there is a certain correlation between variables, so that the information reflected by the observed data overlaps to a certain extent. People naturally want to replace the original variables with fewer variables, and these fewer variables reflect the information of the original variables as much as possible. Using this dimensionality reduction idea, a principal component analysis was generated, that is, through dimensionality reduction, several synthesis factors (principal components) were found to represent many variables, so that these synthesis factors could reflect the information of the original variables as much as possible”) and Zhang (Abstract, Line8-13, “method demonstrates the most superior ability to distinguish various land cover types among five methods because it fully utilizes both the spectral amplitude and shape feature in the hyperspectral data. A combination of the spectral amplitude-sensitive method and the shape-sensitive method will effectively improve the identification accuracy of different land cover types”) such that the synergistic combination of Euclidean and cosine distances within a unified framework maximizes discrimination accuracy by simultaneously capturing both the magnitude (amplitude) and the directional (shape) characteristics inherent ins spectral data, a principle that remains fundamentally applicable even within a dimensionally reduced PCA feature space . 07-21-aia AIA Claim 13 is rejected under 35 U.S.C. 103 as being unpatentable over Ivanov and Inkawhich discussed above in Claim 11, in view of Zhihuang . As to dependent Claim 13, The combination of Ivanov and Inkawhich teaches all the limitations described in Claim 11. However, the combination fails to teach the following limitations but Zhihuang teaches the method of claim 11, wherein the at least one learning data set includes a first learning data set and a second learning data set, and performing of the data split includes sampling the first data set and the second data set into a training group, a valid group, and a test group of the first learning data set in descending order of decreasing OOD indexes according to the plurality of semiconductors; and sampling the first data set and the second data set into a training group, a valid group, and a test group of the second learning data set in ascending order of increasing OOD indexes according to the plurality of semiconductors ( Zhihuang, Pg3, Claim4, Lines4-11, “Arrange the features in descending order according to the relevance of the features to obtain an ordered feature sequence; Select the first M1 features from the ordered feature sequence, where M1 is a positive integer greater than or equal to 1; Clustering the object based on the first M1 features; If after clustering the objects based on the first M1 features, the ratio of the number of objects with a small proportion to the number of objects with a large proportion is 1:N, then the clustering result of the object is used as the clustering result”, and Pg21 , paragraph1, Lines1-5, “It should be noted that, here is an example of sorting in descending order according to the relevance of the features, but in other embodiments, it is also possible to sort them in ascending order according to the relevance of the features, but the features are selected for clustering in the following steps When, you can select the last M1 features or the last M2 features”, wherein arranging features by relevance and selecting feature subsets for clustering. The specification explicitly states that although an example uses descending order, in other embodiments features may be sorted in ascending order and, in that case, the last M1 or last M2 features are selected, which is inherently equivalent to producing feature subsets from both descending and ascending orderings, corresponding to the two learning data sets in both ascending and descending orders). Ivanov, Inkawhich and Zhihuang are analogous to the claimed invention as they are from the same field of endeavor of collecting and partitioning the data into multiple groups according to an indicator for AI model trainings. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to combine the DOE-conditioned measurement models for semiconductor structure prediction of Ivanov and the threshold-based classifier using OOD/ID score of Inkawhich with feature-based partitioning and synthetic-minority augmentation for constructing training datasets of Zhihuang. The motivation to combine is to as recited by Zhihuang (Zhihuang, Abstract, “Through the scheme provided by the embodiment of the invention, the sample can be automatically labeled, and a lot of manpower and material resources are saved”, Pg7, paragraph4, Line11-13, “By adding minority samples, the problem of imbalance of data categories can be solved to ensure that the information of minority samples is not covered”) such that feature-relevance clustering and synthetic minority augmentation enable automated partitioning and balanced training set construction, improving representativeness and robustness of learned models while reducing manual labeling effort . 07-21-aia AIA Claim 14 is rejected under 35 U.S.C. 103 as being unpatentable over Ivanov, Inkawhich and Zhihuang as discussed in Claim 13 above and in further view of Donghua . As to dependent Claim 14, The combination of Ivanov, Inkawhich and Zhihuang teaches all the limitations in Claim 13 as described. The combination teaches about the two training data sets in both ascending and descending orders where each training data set contains 3 clusters. But the combination fails to teach the following limitations but Donghua teaches the method of claim 13, wherein the providing of the optimal AI algorithm includes extracting a first evaluation indicator for each of first valid data and first test data from each of the plurality of AI algorithms, by applying the first valid data classified into the valid group of the first learning data in the first data set and the second data set and the firs test data classified into the test group of the first learning data set in the first data set and the second data set to each of the plurality of AI algorithms. However, Donghua teaches this limitation ( Donghua, Pg10, algorithm 3, “(RMSE) is calculated, and then a function of inverse trend is designed according to the size of the error root mean square, and the weight is given to each model”, Pg12, paragraph4, “Firstly, the training data is put into the general model to train and forecast the data and error data of each model”, “with the prediction error of the training model, the prediction error of the training set and the prediction error of the prediction set are calculated respectively” , wherein computing RMSE for each candidate model and extracting per-model error data from training/validation runs (here, the example uses the training data but with the combination of Zhihuang mentioned above can interpret the training data as the two groups of the three are being used for computing RMSE for each group respectively); these extracted error metrics are used to design inverse-trend weighting functions and to select the optimal weighting algorithm by comparing predicted sets to a verification set, thereby it is equivalent to the extracting step of the claimed invention); extracting a second evaluation indicator for each of second valid data and second test data from each of the plurality of AI algorithms, by applying the second valid data classified into the valid group of the second learning data set in the first data set and the second data set and the second test data classified into the test group of the second learning data set in the first data set and the second data set to each of the plurality of AI algorithms( Donghua, Pg10, algorithm 3, “(RMSE) is calculated, and then a function of inverse trend is designed according to the size of the error root mean square, and the weight is given to each model”, Pg12, paragraph4, “Firstly, the training data is put into the general model to train and forecast the data and error data of each model”, “with the prediction error of the training model, the prediction error of the training set and the prediction error of the prediction set are calculated respectively” , wherein the same rationale of the extraction of the first evaluation indicator above applies but to the second learning data set, which was mentioned above in Claim 6, that computing RMSE for each candidate model and extracting per-model error data from training/validation runs; these extracted error metrics are used to design inverse-trend weighting functions and to select the optimal weighting algorithm by comparing predicted sets to a verification set, thereby it is equivalent to the extracting step of the claimed invention); and selecting the optimal AI algorithm from among the plurality of AI algorithms, based on a valid root mean square error (RMSE) for the first valid data, a test RMSE for the first test data, a valid RMSE for the second valid data, and a test RMSE for the second test data( Donghua, Pg11, paragraph1, Lines1-2, “the optimal weight algorithm selection algorithm selects the best weight algorithm”, wherein computing per-model error metrics (RMSE) from training/validation runs, transforms those errors into weights via an inverse-trend weighting function, forms ensemble predictions under multiple weighting algorithms and comparing each ensemble’s validation error and selects the weighting algorithm that yields the minimum error is conceptually equivalent to the selection of the optimal algorithm of the claimed invention). The motivation to combine, before the effective filing date of the claimed invention, is to as recited by Donghua (Donghua, Pg10, algorithm 3, “(RMSE) is calculated, and then a function of inverse trend is designed according to the size of the error root mean square, and the weight is given to each model”, Pg3, paragraph4,5, “The data set predicted by the weight library is used to compare with the verification set to get the error. By the minimum error, the optimal weight algorithm is obtained”) such that computing per-model RMSE and maps RMSE to model weights via an inverse-trend function; RMSE is used because it penalizes large errors, provides a comparable scalar performance metric across models, and yields continuous values suitable for weight-function design, the weighted ensembles are then evaluated on a verification set and the weighting algorithm with minimum validation RMSE is selected as optimal, which enhances the accuracy of model-ran results . 07-21-aia AIA Claim s 15 is rejected under 35 U.S.C. 103 as being unpatentable over Ivanov and Inkawhich discussed above in Claim 11, in view of Donghua . As to dependent Claim 15, The combination of Ivanov and Inkawhich teaches all the limitations of Claim 11 as discussed above including performing data split using discrete points based on the conditioned or non-conditioned parameters but the combination fails to teach the following limitations but Donghua teaches the method of claim 11, wherein the providing of the optimal AI algorithm includes extracting a first evaluation indicator, from each of the plurality of AI algorithms, by applying test data classified as a test group, of the at least one learning data set in the first data set and the second data set, to each of the plurality of AI algorithms( Donghua, Pg10, algorithm 3, “(RMSE) is calculated, and then a function of inverse trend is designed according to the size of the error root mean square, and the weight is given to each model”, Pg12, paragraph4, “Firstly, the training data is put into the general model to train and forecast the data and error data of each model”, “with the prediction error of the training model, the prediction error of the training set and the prediction error of the prediction set are calculated respectively” , wherein computing RMSE for each candidate model and extracting per-model error data from training/validation runs (here, the example uses the training data but as mentioned in Claim 1 by Ivanov , the partitioned data sets can include these test/valid groups as the claimed invention groups them according to the OOD indexes or here by the conditioned or non-conditioned parameters); these extracted error metrics are used to design inverse-trend weighting functions and to select the optimal weighting algorithm by comparing predicted sets to a verification set, thereby it is equivalent to the extracting step of the claimed invention); extracting a second evaluation indicator from each of the plurality of AI algorithms, by applying valid data classified into a valid group, of the at least one learning data set in the first data set and the second data set, to each of the plurality of AI algorithms( Donghua, Pg10, algorithm 3, “(RMSE) is calculated, and then a function of inverse trend is designed according to the size of the error root mean square, and the weight is given to each model”, Pg12, paragraph4, “Firstly, the training data is put into the general model to train and forecast the data and error data of each model”, “with the prediction error of the training model, the prediction error of the training set and the prediction error of the prediction set are calculated respectively” , wherein the same rationale above applies such that computing RMSE for each candidate model and extracting per-model error data from training/validation runs (here, the example uses the training data but as mentioned in Claim 1 by Ivanov , the partitioned data sets can include these test/valid groups as they are just grouped according to the OOD indexes or here by the conditioned or non-conditioned parameters); these extracted error metrics are used to design inverse-trend weighting functions and to select the optimal weighting algorithm by comparing predicted sets to a verification set, thereby it is equivalent to the extracting step of the claimed invention); determining a ratio of the second evaluation indicator to the first evaluation indicator for each of the plurality of AI algorithms( Donghua, Pg6, Algorithm2, “Algorithm 2: Excludes the relatively poor model of 20% predicted results and give the same weight to the rest of the model”, Pg12, paragraph4, Lines6-7, “Respectively, to the three values of a weight, take 0.3,0.3,0.4. And finally get the comprehensive error value”, wherein determining selection and weighting ratios – for example, excluding the worst 20% of the model results (Algorithm2) and assigning ensemble weights such as 0.3, 0.3, 0.4 – then evaluating resulting ensembles on a verification set to select the best configuration, which is inherently equivalent to the determining a ratio of the two indicators of the claimed invention); selecting an AI algorithm having a smallest product of the first evaluation indicator and the ratio from among the plurality of AI algorithms as the optimal AI algorithm ( Donghua, Pg6, Algorithm2, “Algorithm 2: Excludes the relatively poor model of 20% predicted results and give the same weight to the rest of the model”, Pg3, paragraph4,5, “The data set predicted by the weight library is used to compare with the verification set to get the error. By the minimum error, the optimal weight algorithm is obtained”, Pg11, paragraph4, Lines4-7, “the prediction error of the training model, the prediction error of the training set and the prediction error of the prediction set are calculated respectively. Respectively, to the three values of a weight, take 0.3,0.3,0.4. And finally get the comprehensive error value”, wherein computing per-model error metrics (RMSE), applying selection/weighting ratios, forming ensemble predictions under those ratios, and selecting the weighting/selection algorithm that yields the minimum validation error, which effectively selects the configuration that minimizes the validation error resulting from the indicator-to-ratio combination and this is inherently equivalent to selecting the optimal algorithm using the smallest product of the claimed invention). Ivanov, Inkawhich and Donghua are analogous to the claimed invention as they are from the same field of endeavor of selecting and providing the optimal algorithm based on the weight of the AI algorithms. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to combine the DOE-conditioned measurement models for semiconductor structure prediction of Ivanov and the threshold-based classifier using OOD/ID score of Inkawhich with the multi-model weighting/selection framework based on per-model error metrics (RMSE) of Donghua. The motivation to combine is to as recited by Donghua (Donghua, Pg10, algorithm 3, “(RMSE) is calculated, and then a function of inverse trend is designed according to the size of the error root mean square, and the weight is given to each model”, Pg3, paragraph4,5, “The data set predicted by the weight library is used to compare with the verification set to get the error. By the minimum error, the optimal weight algorithm is obtained”) such that computing per-model RMSE and maps RMSE to model weights via an inverse-trend function; RMSE is used because it penalizes large errors, provides a comparable scalar performance metric across models, and yields continuous values suitable for weight-function design, the weighted ensembles are then evaluated on a verification set and the weighting algorithm with minimum validation RMSE is selected as optimal, which enhances the accuracy of model-ran results. Conclusion Any inquiry concerning this communication or earlier communications from the examiner should be directed to DONG YOON JUNG whose telephone number is (571)270-0198. The examiner can normally be reached 8am-5pm. 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, Cesar Paula can be reached at (571) 272-4128. 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. /DONG YOON JUNG/Examiner, Art Unit 2145 /CESAR B PAULA/Supervisory Patent Examiner, Art Unit 2145 Application/Control Number: 18/453,651 Page 2 Art Unit: 2145 Application/Control Number: 18/453,651 Page 3 Art Unit: 2145 Application/Control Number: 18/453,651 Page 4 Art Unit: 2145 Application/Control Number: 18/453,651 Page 5 Art Unit: 2145 Application/Control Number: 18/453,651 Page 6 Art Unit: 2145 Application/Control Number: 18/453,651 Page 7 Art Unit: 2145 Application/Control Number: 18/453,651 Page 8 Art Unit: 2145 Application/Control Number: 18/453,651 Page 9 Art Unit: 2145 Application/Control Number: 18/453,651 Page 10 Art Unit: 2145 Application/Control Number: 18/453,651 Page 11 Art Unit: 2145 Application/Control Number: 18/453,651 Page 12 Art Unit: 2145 Application/Control Number: 18/453,651 Page 13 Art Unit: 2145 Application/Control Number: 18/453,651 Page 14 Art Unit: 2145 Application/Control Number: 18/453,651 Page 15 Art Unit: 2145 Application/Control Number: 18/453,651 Page 16 Art Unit: 2145 Application/Control Number: 18/453,651 Page 17 Art Unit: 2145 Application/Control Number: 18/453,651 Page 18 Art Unit: 2145 Application/Control Number: 18/453,651 Page 19 Art Unit: 2145 Application/Control Number: 18/453,651 Page 20 Art Unit: 2145 Application/Control Number: 18/453,651 Page 21 Art Unit: 2145 Application/Control Number: 18/453,651 Page 22 Art Unit: 2145 Application/Control Number: 18/453,651 Page 23 Art Unit: 2145 Application/Control Number: 18/453,651 Page 24 Art Unit: 2145 Application/Control Number: 18/453,651 Page 25 Art Unit: 2145 Application/Control Number: 18/453,651 Page 26 Art Unit: 2145 Application/Control Number: 18/453,651 Page 27 Art Unit: 2145 Application/Control Number: 18/453,651 Page 28 Art Unit: 2145 Application/Control Number: 18/453,651 Page 29 Art Unit: 2145 Application/Control Number: 18/453,651 Page 30 Art Unit: 2145 Application/Control Number: 18/453,651 Page 31 Art Unit: 2145 Application/Control Number: 18/453,651 Page 32 Art Unit: 2145 Application/Control Number: 18/453,651 Page 33 Art Unit: 2145 Application/Control Number: 18/453,651 Page 34 Art Unit: 2145 Application/Control Number: 18/453,651 Page 35 Art Unit: 2145 Application/Control Number: 18/453,651 Page 36 Art Unit: 2145 Application/Control Number: 18/453,651 Page 37 Art Unit: 2145 Application/Control Number: 18/453,651 Page 38 Art Unit: 2145 Application/Control Number: 18/453,651 Page 39 Art Unit: 2145 Application/Control Number: 18/453,651 Page 40 Art Unit: 2145 Application/Control Number: 18/453,651 Page 41 Art Unit: 2145 Application/Control Number: 18/453,651 Page 42 Art Unit: 2145 Application/Control Number: 18/453,651 Page 43 Art Unit: 2145 Application/Control Number: 18/453,651 Page 44 Art Unit: 2145 Application/Control Number: 18/453,651 Page 45 Art Unit: 2145 Application/Control Number: 18/453,651 Page 46 Art Unit: 2145 Application/Control Number: 18/453,651 Page 47 Art Unit: 2145 Application/Control Number: 18/453,651 Page 48 Art Unit: 2145 Application/Control Number: 18/453,651 Page 49 Art Unit: 2145 Application/Control Number: 18/453,651 Page 50 Art Unit: 2145 Application/Control Number: 18/453,651 Page 51 Art Unit: 2145 Application/Control Number: 18/453,651 Page 52 Art Unit: 2145 Application/Control Number: 18/453,651 Page 53 Art Unit: 2145 Application/Control Number: 18/453,651 Page 54 Art Unit: 2145
Prosecution Insights