17457684. ACCELERATING DECISION TREE INFERENCES BASED ON TENSOR OPERATIONS simplified abstract (INTERNATIONAL BUSINESS MACHINES CORPORATION)
ACCELERATING DECISION TREE INFERENCES BASED ON TENSOR OPERATIONS
Organization Name
INTERNATIONAL BUSINESS MACHINES CORPORATION
Inventor(s)
Nikolaos Papandreou of Thalwil (CH)
Charalampos Pozidis of Thalwil (CH)
Milos Stanisavljevic of Adliswil (CH)
Jan Van Lunteren of Rueschlikon (CH)
Cedric Lichtenau of Stuttgart (DE)
Andrew M. Sica of Oxford CT (US)
ACCELERATING DECISION TREE INFERENCES BASED ON TENSOR OPERATIONS - A simplified explanation of the abstract
This abstract first appeared for US patent application 17457684 titled 'ACCELERATING DECISION TREE INFERENCES BASED ON TENSOR OPERATIONS
Simplified Explanation
The abstract describes a method for accessing a specific number of top levels in multiple decision trees. The method involves identifying the top nodes of each decision tree and then determining the subtrees that are connected to the remaining nodes of each tree. Each input record is processed through the top nodes of each decision tree to associate it with a specific subtree.
- The method allows for accessing a specific number of top levels in multiple decision trees.
- It involves identifying the top nodes of each decision tree and determining the subtrees connected to the remaining nodes.
- Each input record is associated with a specific subtree by processing it through the top nodes of each decision tree.
- The method can be applied to multiple decision trees and input records, resulting in a total of K × N associations.
Potential Applications
- This method can be used in various fields that involve decision-making processes, such as finance, healthcare, and marketing.
- It can be applied in recommendation systems to associate user preferences with specific subsets of decision trees.
- The method can be utilized in fraud detection systems to identify patterns and associate them with specific decision tree subtrees.
Problems Solved
- The method solves the problem of efficiently accessing and processing specific levels of multiple decision trees.
- It addresses the challenge of associating input records with the appropriate subtrees in a scalable manner.
- The method solves the problem of managing large amounts of data and decision trees by providing a systematic approach to processing and associating records.
Benefits
- The method allows for efficient access and processing of specific levels in multiple decision trees.
- It enables accurate association of input records with the appropriate subtrees, improving decision-making processes.
- The method provides scalability, allowing for the management of large amounts of data and decision trees.
- It can enhance the performance and accuracy of recommendation systems, fraud detection systems, and other applications that involve decision trees.
Original Abstract Submitted
Accessing a value M identifying M top levels of one or more N decision trees, wherein 1 ≤ M < Min(L, ...., L) and wherein a M top levels defines top nodes for each of the N decision trees, and wherein for each decision tree T of the N decision trees. Identifying one or more subtrees subtended by respective subsets of remaining nodes of each decision tree T, a remaining nodes including all of the nodes of said each decision tree T but its top nodes. Processing each of the K input records through a top nodes of said each decision tree T to associate each of the K input records with a single, respective one of the subtrees of each decision tree T, wherein K × N associations are obtained in total for the N decision trees and the K input records.
