Decision List Optimization via Rule Ranking for Prediction Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing decision list techniques face challenges in maintaining prediction performance while preventing increases in processing time and memory usage, especially when determining an optimal decision list with a large value of k.
Innovation Solution
An information processing apparatus that includes a prediction means for calculating prediction results based on predicted values of k top-ranked decision rules and a list determining means for updating variables indicative of the decision list until a predetermined condition is satisfied, thereby determining the decision list to be output.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If k is set to a greater value to improve prediction performance, then prediction accuracy is improved, but processing time and memory usage increase
Solution Approach 1:
The patent segments the decision rule selection process by introducing a ranking mechanism. Instead of considering all possible combinations of k decision rules, the system ranks decision rules based on their satisfaction degrees and selectively applies only the top-ranked rules. This segmentation reduces the search space from exponential to polynomial complexity, maintaining prediction accuracy while significantly reducing processing time.
Solution Approach 2:
The patent changes the parameter representation from binary selection (select/deselect) to continuous ranking (satisfaction degree). By introducing satisfaction degrees as a new parameter and using it to rank decision rules, the system transforms the combinatorial optimization problem into a more efficient ranking-based selection process, reducing computational complexity while preserving prediction performance.
2Measurement precision
If k is set to a greater value to improve prediction performance, then prediction accuracy is improved, but memory usage increases
Solution Approach 1:
The patent segments the set of all decision rules into ranked subsets based on satisfaction degrees. Instead of storing and processing information for all possible combinations of k rules, the system only needs to store the ranking information for individual rules. This segmentation reduces memory requirements from exponential to linear with respect to the number of decision rules.
3Measurement precision
If an optimal decision list is determined by preparing and solving an optimization problem with k decision rules, then prediction performance is improved, but the number of variables increases
Solution Approach 1:
The patent fundamentally changes the parameter space by replacing binary selection variables with continuous satisfaction degree parameters. Instead of having 2^k possible variable combinations, the system uses satisfaction degrees to directly rank and select the top k rules. This parameter transformation reduces the optimization problem from exponential variable space to polynomial variable space, making it computationally tractable while maintaining prediction performance.
Data Source
AI summary
An information processing apparatus 1 includes: a prediction section 11 that for a training example included in a training example set, calculates a prediction result on the basis of predicted values of k top-ranked decision rules which are among decision rules included in a decision list and whose conditions are satisfied by the training example; and a list determining section 12 that by repeatedly carrying out, until a predetermined condition is satisfied by an objective function including an error term of the prediction result, a process for updating a variable indicative of the decision list, determines the decision list, the variable including a variable indicative of a decision rule which is among the decision rules whose conditions are satisfied and which is given kth priority to be used for prediction. This enables promotion of better decision making by a user on the basis of a higher priority decision rule.