# US 2022/0044145 A1 · Machine learning accelerator with decision tree interconnects > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2022/0044145 A1 - Document ID: US20220044145A1 - Status: Published application - Kind: A1 - Application number: 16/986,506 - Filed: 2020-08-06 - Published: 2022-02-10 - Earliest claimed: 2020-08-06 - Assignees: International Business Machines Corporation - Applicants: International Business Machines Corporation - Inventors: Mingu Kang; Seonghoon Woo; Eun Kyung Lee; Sukjay Chey - Web page: https://patentgrep.com/patent/US20220044145A1 ## CPC - G06N 20/00: Machine learning - G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS ## Abstract Techniques for performing improved machine learning using decision trees are disclosed. In one example, a system includes a plurality of decision tree structures, and configuration logic operatively coupled to the plurality of decision tree structures. The configuration logic selectively configures the plurality of decision tree structures to form at least one of: one or more combined decision tree structures, wherein a combined decision tree structure comprises multiple interconnected ones of the plurality of decision tree structures; and one or more individual decision tree structures, wherein an individual decision tree structure comprises a single one of the plurality of decision tree structures. ## Claims ### Claim 1 A system comprising: a plurality of decision tree structures; and configuration logic operatively coupled to the plurality of decision tree structures, wherein the configuration logic selectively configures the plurality of decision tree structures to form at least one of: one or more combined decision tree structures, wherein a combined decision tree structure comprises multiple interconnected ones of the plurality of decision tree structures; and one or more individual decision tree structures, wherein an individual decision tree structure comprises a single one of the plurality of decision tree structures. ### Claim 2 (depends on 1) The system of claim 1, wherein each one of the plurality of decision tree structures is balanced in shape. ### Claim 3 (depends on 2) The system of claim 2, wherein each one of the plurality of decision tree structures is uniform in shape with respect to each other ones of the plurality of decision tree structures. ### Claim 4 (depends on 1) The system of claim 1, wherein the configuration logic further comprises a plurality of multiplexers operatively coupled to at least a subset of the plurality of decision tree structures, wherein each one of the plurality of multiplexers is configured to selectively connect: an input of one of the plurality of decision tree structures to an output of another one of the plurality of decision tree structures; or an input of one of the plurality of decision tree structures to an input independent of other ones of the plurality of decision tree structures. ### Claim 5 (depends on 4) The system of claim 4, wherein the configuration logic further comprises a register operatively coupled to the plurality of multiplexers, wherein the register provides data for controlling the selective connection of ones of the plurality of multiplexers. ### Claim 6 (depends on 1) The system of claim 1, further comprising output generation logic operatively coupled to the plurality of decision tree structures, wherein the output generation logic determines at least one output decision for at least one input based on one or more results from one or more of the plurality of decision tree structures. ### Claim 7 (depends on 1) The system of claim 1, wherein each one of the plurality of decision tree structures comprises a plurality of nodes, wherein at least one node compares an input value to a threshold value and generates a result based on the comparison. ### Claim 8 (depends on 7) The system of claim 7, wherein at least one node outputs a classification label. ### Claim 9 (depends on 1) The system of claim 1, wherein the plurality of decision tree structures is part of a random forest algorithm implementation. ### Claim 10 (depends on 1) The system of claim 1, wherein the system is implemented as part of one or more integrated circuits. ### Claim 11 An apparatus comprising: at least one processor; at least one memory including instruction code; the at least one memory and the instruction code being configured to, with the at least one processor, cause the apparatus at least to: selectively configure a plurality of decision tree structures to form at least one of: one or more combined decision tree structures, wherein a combined decision tree structure comprises multiple interconnected ones of the plurality of decision tree structures; and one or more individual decision tree structures, wherein an individual decision tree structure comprises a single one of the plurality of decision tree structures. ### Claim 12 (depends on 11) The apparatus of claim 11, wherein each one of the plurality of decision tree structures is balanced in shape and uniform in shape with respect to each other ones of the plurality of decision tree structures. ### Claim 13 (depends on 11) The apparatus of claim 11, wherein the selective configuring further comprises selectively connecting: an input of one of the plurality of decision tree structures to an output of another one of the plurality of decision tree structures; or an input of one of the plurality of decision tree structures to an input independent of other ones of the plurality of decision tree structures. ### Claim 14 (depends on 11) The apparatus of claim 11, wherein the at least one memory and the instruction code being further configured to, with the at least one processor, cause the apparatus to determine at least one output decision for at least one input based on one or more results from one or more of the plurality of decision tree structures. ### Claim 15 (depends on 11) The apparatus of claim 11, wherein the plurality of decision tree structures is part of a random forest algorithm implementation. ### Claim 16 A method, comprising steps of: training a plurality of decision tree structures, wherein each one of the plurality of decision tree structures is balanced in shape and uniform in shape with respect to each other ones of the plurality of decision tree structures; obtaining a set of control data; and selectively configuring the plurality of decision tree structures based on the obtained control data to form at least one of: one or more combined decision tree structures, wherein a combined decision tree structure comprises multiple interconnected ones of the plurality of decision tree structures; and one or more individual decision tree structures, wherein an individual decision tree structure comprises a single one of the plurality of decision tree structures; wherein one or more of the steps are executed by processing circuitry configured to execute instruction code. ### Claim 17 (depends on 16) The method of claim 16, wherein the selective configuring further comprises selectively connecting: an input of one of the plurality of decision tree structures to an output of another one of the plurality of decision tree structures; or an input of one of the plurality of decision tree structures to an input independent of other ones of the plurality of decision tree structures. ### Claim 18 (depends on 16) The method of claim 16, further comprising determining at least one output decision for at least one input based on one or more results from one or more of the plurality of decision tree structures. ### Claim 19 (depends on 16) The method of claim 16, wherein the plurality of decision tree structures is part of a random forest algorithm implementation. ### Claim 20 An article of manufacture comprising a non-transitory computer-readable storage medium having embodied therein executable instruction code that when executed by a processor causes the processor to perform steps of: training a plurality of decision tree structures, wherein each one of the plurality of decision tree structures is balanced in shape and uniform in shape with respect to each other ones of the plurality of decision tree structures; obtaining a set of control data; and selectively configuring the plurality of decision tree structures based on the obtained control data to form at least one of: one or more combined decision tree structures, wherein a combined decision tree structure comprises multiple interconnected ones of the plurality of decision tree structures; and one or more individual decision tree structures, wherein an individual decision tree structure comprises a single one of the plurality of decision tree structures. ### Claim 21 (depends on 20) The article of claim 20, wherein the selective configuring further comprises selectively connecting: an input of one of the plurality of decision tree structures to an output of another one of the plurality of decision tree structures; or an input of one of the plurality of decision tree structures to an input independent of other ones of the plurality of decision tree structures. ### Claim 22 (depends on 20) The article of claim 20, further comprising determining at least one output decision for at least one input based on one or more results from one or more of the plurality of decision tree structures. ### Claim 23 A system comprising: a plurality of decision tree structures, wherein each one of the plurality of decision tree structures is balanced in shape and uniform in shape with respect to each other ones of the plurality of decision tree structures; configuration logic operatively coupled to the plurality of decision tree structures, wherein the configuration logic selectively configures the plurality of decision tree structures to form at least one of: one or more combined decision tree structures, wherein a combined decision tree structure comprises multiple interconnected ones of the plurality of decision tree structures; and one or more individual decision tree structures, wherein an individual decision tree structure comprises a single one of the plurality of decision tree structures; wherein the configuration logic comprises: a plurality of multiplexers operatively coupled to at least a subset of the plurality of decision tree structures, wherein each one of the plurality of multiplexers is configured to selectively connect: an input of one of the plurality of decision tree structures to an output of another one of the plurality of decision tree structures; or an input of one of the plurality of decision tree structures to an input independent of other ones of the plurality of decision tree structures; and a register operatively coupled to the plurality of multiplexers, wherein the register provides data for controlling the selective connection of ones of the plurality of multiplexers; and output generation logic operatively coupled to the plurality of decision tree structures and register, wherein the output generation logic determines at least one output decision for at least one input based on one or more results from one or more of the plurality of decision tree structures. ### Claim 24 (depends on 23) The system of claim 23, wherein the system is configured to implement a random forest algorithm. ### Claim 25 (depends on 23) The system of claim 23, wherein the system is implemented as part of one or more integrated circuits.