MUMBAI, India, Aug. 17 -- Intellectual Property India has published a patent application (202641095982 A) filed by Samala Bhavana; and Dr. M. Janga Reddy on August 08, 2026, for Predictive Semantic Data Compression System For Large-Scale Machine Learning Datasets With Adaptive Feature Reconstruction.
Inventors include Dr. G. Jagan Naik; Ch. Swapna; K. Venkata Balamurali Krishna; N. Sowmya; J. S. Geetha Priya; and Guduru Mounika.
The application for the patent was published on August 14, 2026, under issue no. 33/2026.
Abstract: The present disclosure relates to a predictive semantic data compression system (100) for large-scale machine learning datasets with adaptive feature reconstruction. The system (100) comprises a memory (102), a processor (104), and a communication module (106) configured to receive and transmit distributed dataset streams. A semantic redundancy analysis module (108) identifies contextual relationships and redundancy patterns within machine learning datasets by generating semantic representations. A predictive compression module (110) forecasts latent feature correlations and anticipated redundancy to enable proactive compression of data. An adaptive feature reconstruction module (112) reconstructs task-relevant features from compressed semantic representations based on downstream machine learning requirements. A compression optimization module (114) iteratively refines compression and reconstruction performance using semantic loss and fidelity metrics. The system (100) enables prediction-driven semantic compression and reconstruction, reducing storage and computational overhead while preserving learning-relevant feature integrity across distributed and heterogeneous machine learning environments.
Disclaimer: Curated by HT Syndication.
