# US 2018/0025284 A1 · Methods and systems for identifying patterns in data using delimited feature-regions > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2018/0025284 A1 - Document ID: US20180025284A1 - Status: Published application - Kind: A1 - Application number: 15/658,104 - Filed: 2017-07-24 - Published: 2018-01-25 - Earliest claimed: 2016-07-22 - Inventors: Jonathan Michael Fisher - Web page: https://patentgrep.com/patent/US20180025284A1 ## CPC - G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks - G06N 3/10: Interfaces, programming languages or software development kits, e.g. for simulating neural networks - G06N 20/00: Machine learning - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS ## Abstract A method and system is provided for identifying patters in datasets by identifying delimited regions of feature-space in which patterns occur. The delimited regions are then combined into an ensemble able to make predictions based on the identified regions of feature-space. The method may be used for classification, for regression, for auto-encoding, for simulation, and for other applications of pattern detection. ## Claims ### Claim 1 A method for identifying a pattern in a dataset, comprising: identifying a plurality of feature-regions each in a feature-space wherein points from said dataset form an identifiable pattern and constructing, for each such feature-region, a boundary which approximately encloses said feature-region so as to approximately include said points which form said pattern and approximately exclude other points; and combining said feature-regions into an ensemble. ### Claim 2 (depends on 1) The method according to claim 1, wherein one or more of said feature-regions are identified by a process comprising: identifying a feature-region contained within a feature-space; and projecting said feature-space down to a lower-dimensional feature-space; and modifying the boundary of said feature-region as embedded within said lower-dimensional feature-space so as to approximately include points which form a pattern and approximately exclude other points to form a new identified feature-region to be included in said ensemble. ### Claim 3 (depends on 1) The method according to claim 1, wherein one or more of said feature-regions are identified by a process comprising: identifying a feature-region contained within a feature-space; and embedding said feature-space within a higher-dimensional feature-space; and modifying the boundary of said feature-region as embedded within said higher-dimensional feature-space so as to approximately include points which form a pattern and approximately exclude other points to form a new identified feature-region to be included in said ensemble. ### Claim 4 (depends on 1) The method according to claim 1, wherein one or more of said feature-regions are identified by a process comprising: identifying a plurality of feature-regions each of which is contained within a feature-space; and constructing the Cartesian product of said feature-spaces; and constructing the Cartesian product of said feature-regions as embedded within the Cartesian product of said feature-spaces; and modifying the boundary of said Cartesian product of said feature-regions as embedded within said Cartesian product of said feature-spaces so as to approximately include points which form a pattern and approximately exclude other points to form a new identified feature-region to be included in said ensemble. ### Claim 5 (depends on 1) The method according to claim 1, wherein one or more of said identifiable patterns consists of a feature-region which has a density of points from said dataset differing from the typical or average density of points from said dataset over said feature-space as a whole. ### Claim 6 (depends on 2) The method according to claim 2, wherein one or more of said identifiable patterns consists of a feature-region which has a density of points from said dataset differing from the typical or average density of points from said dataset over said projected lower-dimensional feature-space as a whole. ### Claim 7 (depends on 3) The method according to claim 3, wherein one or more of said identifiable patterns consists of a feature-region which has a density of points from said dataset differing from the typical or average density of points from said dataset over said higher-dimensional feature-space as a whole. ### Claim 8 (depends on 4) The method according to claim 4, wherein one or more of said identifiable patterns consists of a feature-region which has a density of points from said dataset differing from the typical or average density of points from said dataset over said Cartesian product of feature-spaces as a whole. ### Claim 9 (depends on 1) The method according to claim 1, wherein one or more of said identifiable patterns consists of a feature-region which has a skewness of points from said dataset according to one or more classifications or attributes of said points differing from the typical or average skewness of points from said dataset over said feature-space as a whole. ### Claim 10 (depends on 2) The method according to claim 2, wherein one or more of said identifiable patterns consists of a feature-region which has a skewness of points from said dataset according to one or more classifications or attributes of said points differing from the typical or average skewness of points from said dataset over said projected lower-dimensional feature-space as a whole. ### Claim 11 (depends on 3) The method according to claim 3, wherein one or more of said identifiable patterns consists of a feature-region which has a skewness of points from said dataset according to one or more classifications or attributes of said points differing from the typical or average skewness of points from said dataset over said higher-dimensional feature-space as a whole. ### Claim 12 (depends on 4) The method according to claim 4, wherein one or more of said identifiable patterns consists of a feature-region which has a skewness of points from said dataset according to one or more classifications or attributes of said points differing from the typical or average skewness of points from said dataset over said Cartesian product of feature-spaces as a whole. ### Claim 13 (depends on 1) The method according to claim 1, wherein one or more of said identifiable patterns consists of a feature-region which has a trend of points from said dataset according to one or more classifications, statuses, or attributes of said points. ### Claim 14 (depends on 2) The method according to claim 2, wherein one or more of said identifiable patterns consists of a feature-region which has a trend of points from said dataset according to one or more classifications statuses, or attributes of said points. ### Claim 15 (depends on 3) The method according to claim 3, wherein one or more of said identifiable patterns consists of a feature-region which has a trend of points from said dataset according to one or more classifications statuses, or attributes of said points. ### Claim 16 (depends on 4) The method according to claim 4, wherein one or more of said identifiable patterns consists of a feature-region which has a trend of points from said dataset according to one or more classifications statuses, or attributes of said points.