MUMBAI, India, Jan. 2 -- Intellectual Property India has published a patent application (202541124468 A) filed by Indian Institute Of Technology, Hyderabad, Telangana, on Dec. 10, 2025, for 'a method and a system for structural depth consolidation of a multi- layer convolutional neural networks.'
Inventor(s) include Pratibha Verma; and Amit Acharyya.
The application for the patent was published on Jan. 2, under issue no. 01/2026.
According to the abstract released by the Intellectual Property India: "The present disclosure relates to relates to a method (300) and a system (100) for for structural depth consolidation of a multi-layer CNN. The method (300) includes receiving (302) a pre-trained multi-layer CNN. The method (300) further includes checking (304) presence of a Batch Normalization (BN) layer between any two consecutive layers of the CNN, and consolidating the BN layer into the immediately preceding layer of the CNN, if the BN layer is detected. Further, the method (300) includes removing (306) an activation function between the two consecutive layers, and merging the consecutive layers and evaluating (308) the accuracy of the CNN after merging of the two consecutive layers. Furthermore, the method (300) includes performing (310) fine-tuning of the CNN if the accuracy of the CNN falls below a target threshold value and processing (312) the next pair of consecutive layers in the CNN if the accuracy of the CNN meets the target threshold."
Disclaimer: Curated by HT Syndication.
