MUMBAI, India, July 13 -- Intellectual Property India has published a patent application (202631068741 A) filed by National Institute Of Technology Rourkela Nit Rourkela on June 01, 2026, for Computer-Implemented System And Method For Forecasting Non-Linear Stochastic Time-Series Data.
Inventors include Samota, Mr. Sandeep Kumar; Gupta, Ms. Reema; and Chakraverty, Dr. Snehashish.
The application for the patent was published on July 10, 2026, under issue no. 28/2026.
Abstract: The present invention relates to a computer-implemented system (1100) for forecasting non-linear stochastic time-series data. The non-invasive system (1100) comprises a hardware processor (1102) coupled with a memory storing executable instructions, a data ingestion and normalization module (1104 and 1106) configured to receive multivariate time-series input data associated with at least one stochastic variable and normalize the received data into a standardized representation, a variational encoding module (1108) configured to transform the normalized input data into a probabilistic latent representation, a neural stochastic differential solver (1110) configured to propagate the latent initial state in continuous time according to a stochastic differential formulation, a probabilistic decoding module (1112) configured to reconstruct evolved latent trajectories into forecast output data, and a training engine (1114) configured to train the variational encoding module (1108), the neural stochastic differential solver (1110), and the probabilistic decoding module (1112) in an end-to-end manner by using an ELBO framework.
Disclaimer: Curated by HT Syndication.
