MUMBAI, India, Jan. 9 -- Intellectual Property India has published a patent application (202521108352 A) filed by Alpesh Prakash Sonar; and Divyani Alpesh Sonar, Thane, Maharashtra, on Nov. 8, 2025, for 'wizard-based no-code machine learning system and method for automated task detection.'
Inventor(s) include Alpesh Prakash Sonar; and Divyani Alpesh Sonar.
The application for the patent was published on Dec. 12, under issue no. 50/2025.
According to the abstract released by the Intellectual Property India: "The present invention relates to a distributed, wizard-based system (100) for automated machine learning (ML) task detection across multi-modal datasets. A first computing device (102) receives user data in tabular, text, or image formats and employs a wizard interface generator (108) to initiate the workflow. The data is transmitted to a second computing device (104), which performs statistical and structural analysis using parallel hardware pipelines, identifying schema patterns, annotation formats, and modality-specific characteristics. The analyzed data is then processed by a third computing device (106) that combines rule-based logic and machine learning models to classify the dataset into predefined ML tasks such as classification, regression, question answering, token classification, or object detection. A confidence evaluation module computes certainty scores and prompts users for confirmation when thresholds are not met. The wizard interface (108) dynamically adapts user interface steps with configuration guidance, explanations, and model recommendations, thereby simplifying ML workflow creation for non-technical users."
Disclaimer: Curated by HT Syndication.
