Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
DETAILED ACTION
This Office Action is sent in response to Application’s Communication received on 04/07/2024 for application number 18/699297. The Office hereby acknowledges receipt of the following and placed of record in file: Specification, Drawing, Abstract, Oath/Declaration, and Claims.
Claims (1-6), (7-11) and (12-16) are presented for examination.
Information Disclosure Statement
The information disclosure statements (IDS) submitted on 04/07/2024 were filed prior to current Office Action. 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 Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102 of this title, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-3, 5-9, 11-14 and 16 rejected under AIA 35 U.S.C. 103(a) as being unpatentable over Fei Toney Liu et al. Isolation Forest 2008 (hereinafter Liu) and further in view of Meghana Madhyastha et al. Pacset (packed serialized trees): Reducing Inference Latency for tree ensemble deployment. 2020 (hereinafter Madhyastha) and further in view of Oberg et al. US Patent Application Publication US 20130179377 A1 (hereinafter Oberg).
Regarding claim 1, Liu teaches A method for generating a prediction…using a decision tree based machine learning model, the method comprising (§2, p. 3, §4, p. 6 wherein Liu describes generating an anomaly score for test instance using an ensemble of isolation trees, wherein the score provides an assessment of whether the instance is anomalous or normal) receiving an input for a prediction request (§4.2, p. 6 wherein Liu explains that each instance is passed through the isolation trees to generate its anomaly score, wherein Algorithm 3 describes input declaration that identifies “x” as an instance and as input to the path length. Therefore, Liu describes accepting a test instance as an input and request to the path length function and processing that instance through the isolation trees to generate an anomaly score) selecting a first tree from the machine learning model (§4.2, p. 6 wherein Liu evaluates a test instance through the individual isolation trees of an isolation forest, wherein the Algorithm 3 receives an insolation tree as an input for the path length evaluation and identifies the particular tree selected from the forest traversal) accumulating a result from the first node (§4.2, p. 6 wherein Liu evaluates the current internal node and increments the current path when proceeding to the selected child node as describes in Algorithm 3, p. 6, lines 4-8. The incremented path length constitutes the accumulated result associated with traversal from the first node) selecting and loading a second node from the first tree into working memory (§5.1, p. 6 wherein Liu teaches selecting a second node from the first tree by choosing the current node’s left or right child based on the input dependent split comparison as described in Algorithm 3, p. 6, lines 5-8).
Liu does not teach prediction in a low resource device using a decision tree; selecting and loading a first node from the first tree into working memory.
However in analogous art of isolation forest, Madhyastha teaches a prediction in a low resource device using a decision tree (§1, p. 2-3, §3, p. 2-3, §5.3, p. 6, §6, p. 6 wherein Madhyastha connects prediction performance with inference throughput in tree ensembles and wherein Madhyastha teaches low resource device by disclosing execution of machine learning inference on embedded edge devices having limited computing and memory resources, with the model stored externally and accessed one block at a time) selecting and loading a first node from the first tree into working memory (§3, p. 3, §5.1, p. 6 wherein Madhyastha teaches selectively accessing a tree node needed for inference and loading the block containing that node from external storage into main memory, wherein the loading its containing block into RAM provides the claimed loading into working memory)
It would have been obvious to a person in the ordinary skill in the art before the effective filing date of the claimed invention to combine Liu with Madhyastha by incorporating the method of prediction in a low resource device using a decision tree; selecting and loading a first node from the first tree into working memory of Madhyastha into the method of generating a prediction…using a decision tree based machine learning model of Liu for the purpose implementing Liu’s isolation forest traversal using Madhyastha’ s external storage and selective loading technique. Thereby enabling local anomaly assessment with a reduced RAM requirement. A reasonable expectation of success follows from the compatible traversal operations: both access a path through each tree. The modification would change the storage and retrieval of Liu’s node records while retaining their tests, child links and path-length calculation (Madhyastha: §1, p. 2, §5.3, p. 6).
Liu does not teach releasing the first node from working memory.
However in analogous art of isolation forest, Oberg teaches releasing the first node from working memory ([0050-0051] wherein Oberg reusing memory occupied by previously traversed tree potion by overwriting that portion with a subsequently needed portion, thereby reducing memory requirement).
It would have been obvious to a person in the ordinary skill in the art before the effective filing date of the claimed invention to combine Liu with Oberg by incorporating the method of releasing the first node from working memory of Oberg into the method of generating a prediction…using a decision tree based machine learning model of Liu for the purpose of applying the memory reuse technique to the combined Liu inference implementation by retaining the accumulated path length and next node identifier, relinquishing the previously processed node data and reusing its working memory storage and reduce RAM requirement (Oberg: ¶¶ [0050-0051]).
Regarding claim 2, Liu as modified by Madhyastha and Oberg teaches determining a number of trees in the machine learning model (p. 5 wherein Liu, as described in Algorithm 1, lines 3-7, provides the ensemble size parameter (t) which specifies the number of isolation trees constructed for the model) and dividing the result by the number of trees (§2, p.3, §4.2, p. 6 wherein Liu calculates the average path length across the isolation trees wherein the arithmetic average is obtained by dividing the accumulated path lengths by the number of trees).
Regarding claim 3, Liu as modified by Madhyastha and Oberg teaches wherein the result is a path length in at least the first tree (§2, p. 3, §4.2, p. 6 wherein Liu calculates a path length for an input instance traversing an individual isolation tree, wherein Algorithm 3 receives the selected isolation tree, maintains the current path length and returns the resulting path length).
Regarding claim 5, Liu as modified by Madhyastha and Oberg teaches accumulating the result from the second node. (§4, p. 6, wherein Liu describes isolation forest and wherein the pathlength procedure processes successive nodes of the selected isolation tree and updates the accumulated path length by passing e +1, as describes in Algorithm 3, lines 4-8, when proceeding from an internal node to its selected child. Accordingly, processing the second internal node further increments the previously accumulated result).
Regarding claim 6, Liu as modified by Madhyastha and Oberg teaches wherein the machine learning model is an isolation forest model (Abstract, §4, wherein Liu discloses an isolation forest model, designated iForest for detecting anomalies).
Regarding claim 7, Liu as modified by Madhyastha and Oberg teaches A non-transitory machine-readable medium comprising computer program code which when executed by a computer carries out a set of operations of a method for generating a prediction in a low resource device using a decision tree based machine learning model, the set of operations comprising (Madhyastha: §§1-2, pp. 1-2). Claim 7 is similar in scope to claim 1 therefore the claims are rejected under similar rationale.
Regarding claim 12, Liu as modified by Madhyastha and Oberg teaches an electronic device comprising (Liu: §1, p. 1, Madhyastha: §§2, pp. 1-2). Claim 12 is similar in scope to claim 1 therefore the claims are rejected under similar rationale.
Claims 8 and 13 are similar in scope to claim 2 therefore the claims are rejected under similar rationale.
Claims 9 and 14 are similar in scope to claim 3 therefore the claims are rejected under similar rationale.
Claims 11 and 16 are similar in scope to claim 5 therefore the claims are rejected under similar rationale.
Claims 4, 10 and 15 are rejected under AIA 35 U.S.C. 103(a) as being unpatentable over Fei Toney Liu et al. Isolation Forest 2008 (hereinafter Liu) and further in view of Meghana Madhyastha et al. Pacset (packed serialized trees): Reducing Inference Latency for tree ensemble deployment. 2020 (hereinafter Madhyastha) and further in view of Oberg et al. US Patent Application Publication US 20130179377 A1 (hereinafter Oberg) and further in view of Jurg Gutknecht et al. Project Oberon, The Design of an Operating System and Compiler, 2005 (hereinafter Gutknecht).
Regarding claim 4, Liu, Madhyastha and Oberg do not teach wherein a single node of the selected tree is allotted working memory at a given time.
However in analogous art of isolation forest, Gutknecht teaches wherein a single node of the selected tree is allotted working memory at a given time (§7.4, p. 174, p. 183, p. 184, p. 190 wherein Gutknecht identifies each B-tree nodes as a page and describes a disk pages that are fetched into DirPage buffer that expressly describes allocating one search buffer and declares one a DirPage, repeatedly reads the selected page into a, determines the descendant address and reuses a on the next iteration).
It would have been obvious to a person in the ordinary skill in the art before the effective filing date of the claimed invention to combine Gutknecht with Liu, Madhyastha and Oberg by incorporating the method of wherein a single node of the selected tree is allotted working memory at a given time of Gutknecht into the method of generating a prediction…using a decision tree based machine learning model of with Liu, Madhyastha and Oberg for the purpose of modifying the tree traversal implementation of Liu, as adapted for memory constrained inference by Madhyastha, to allocate and reuse a single node sized working memory buffer according to Gutknecht, thereby reducing the working memory required to hold tree nodes during traversal. (Gutknecht: $7.4, pp. 183-184).
Claims 10 and 15 are similar in scope to claim 4 therefore the claims are rejected under similar rationale.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to HASSAN MRABI whose telephone number is (571)272-8875. The examiner can normally be reached on Monday-Friday, 7:30am-5pm. Alt, Friday, EST.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Viker Lamardo can be reached on 571-270-5871. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/HASSAN MRABI/Examiner, Art Unit 2144
