MUMBAI, India, July 30 -- Intellectual Property India has published a patent application (202621070079 A) filed by Indian Institute Of Technology Gandhinagar on June 04, 2026, for System And Method For Adaptive Feature Suppression Under Limited-Data Conditions.
Inventors include Maheshwari, Om; and Mohapatra, Nihar Ranjan.
The application for the patent was published on July 24, 2026, under issue no. 30/2026.
Abstract: ABSTRACT SYSTEM AND METHOD FOR ADAPTIVE FEATURE SUPPRESSION UNDER LIMITED-DATA CONDITIONS The present disclosure relates to a method 100 for adaptive feature suppression in predictive modeling under limited-data conditions. A dataset 103 comprising input features 104 and output variables 105 is partitioned by a split module 106 into a training dataset 107 and a testing dataset 108. A trial execution module 109 performs multiple randomized trials by further dividing the training dataset 107 into a sub-training dataset 110 and a validation dataset 111. Candidate feature subsets 114 are generated using correlation thresholds 113 computed by a correlation module 112 and evaluated using an intermediate regression model 116 to obtain a performance score 117. Trials not meeting a performance acceptance threshold 118 are discarded. Removal counters 120 are aggregated and used to compute removal frequencies 122, based on which a feature retention module 124 generates a reduced feature set 126. A final model training module 127 trains a predictive model 128 to generate prediction 129.
Disclaimer: Curated by HT Syndication.
