MUMBAI, India, Feb. 13 -- Intellectual Property India has published a patent application (202641008775 A) filed by S. Nagasree; Muntha Raju; G Shirisha; S. Sindhu; S. Tejaswini; and S. Vishnu, Hyderabad, Telangana, on Jan. 28, for 'auto ml: automatic algorithm selection tool.'
Inventor(s) include S. Nagasree; Muntha Raju; G Shirisha; S. Sindhu; S. Tejaswini; and S. Vishnu.
The application for the patent was published on Feb. 13, under issue no. 07/2026.
According to the abstract released by the Intellectual Property India: "The invention relates to an Automatic Algorithm Selection Tool for automated construction, selection, and deployment of machine learning pipelines. The system comprises a data ingestion module, meta feature extractor, meta knowledge base, pipeline generator, constraint aware optimization engine, evaluation module, and deployment interface. For a given dataset and task specification, the meta feature extractor computes dataset descriptors that are supplied to a meta model trained on historical pipeline performance records stored in the meta knowledge base. The meta model outputs candidate pipeline templates, which the pipeline generator instantiates into executable pipelines. The constraint aware optimization engine filters pipelines that violate user specified hard constraints and ranks remaining pipelines using multi objective optimization over predictive performance, resource usage, and other criteria. The evaluation module trains and validates selected pipelines, and the deployment interface packages a recommended pipeline for production use and registers it with external monitoring systems. Telemetry from deployed pipelines is logged back to the meta knowledge base to continuously improve the meta model. The invention thus enables fast, transparent, and extensible AutoML for diverse applications."
Disclaimer: Curated by HT Syndication.
