MUMBAI, India, July 7 -- Intellectual Property India has published a patent application (202621054750 A) filed by Iiti Drishti Cps Foundation; and Indian Institute Of Technology Indore on April 29, 2026, for Method And System For Multivariate Time-Series Analysis.
Inventors include Shevtekar, Sumit S; and Maurya, Chandresh Kumar.
The application for the patent was published on July 03, 2026, under issue no. 27/2026.
Abstract: A method (100) and system (200) for multivariate time-series analysis are disclosed. The method (100) comprises the steps of: receiving (110), multivariate time-series input data (202) comprising a plurality of variables observed over a temporal sequence; extracting (120), local temporal features at multiple scales from the multivariate time-series input data (202), through a local temporal encoding branch (204); processing (130), the multivariate time-series input data (202), through a global dependency encoding branch (206) to extract long-range temporal dependency features; performing (140), early temporal aggregation (208) on output features of the local temporal encoding branch (204) and the global dependency encoding branch (206) to obtain temporally compressed latent representations; fusing (150), the obtained temporally compressed latent representations to generate a unified latent representation; refining (160), the fused unified latent representation; and generating (170), from the refined unified latent representation, a task- specific output corresponding to at least one multivariate time-series analysis task.
Disclaimer: Curated by HT Syndication.
