A control unit for evaluating the plausibility of data using a generative intelligence model and a method thereof
Amit Arvind Kale; Sonam Singh; Koustav Mullick; Nityanand Mathur
Abstract
A control unit for evaluating the plausibility of data using a generative intelligence model and a method thereof. The control unit 10 comprises an image processing unit 14, a vision transformer 16, a feature generation unit 18. The control unit 10 curates a dataset 11a from a large set of images for training the intelligence model 12. The control unit 10 comprehends the curated dataset 11a by focusing on each of the datapoint physical structure. The control unit 10 extracts feature maps related to each of a datapoints of the curated dataset 11a and generates a first mutli-variate gaussian distribution based on an atleast one parameter. The control unit 10 feeds a synthetic dataset 11b to the intelligence model 12 and extracts the feature maps related to each of the datapoint of the synthetic dataset 11b for generating a second mutli-variate gaussian distribution. The control unit 10 evaluates the plausibility of data of the synthetic dataset 11b from the first and second generated mutli- variate gaussian distribution. (Figures1 & 2)
Official documents
The official publication notice names Nityanand Mathur among the inventors.
Applicants: Bosch Global Software Technologies Private Limited; Robert Bosch GmbH.
Application number: 202541029889. The Patent Office Journal No. 40/2026, page 126814. This record describes a published application; it does not assert that a patent has been granted.
Google Scholar entry · IP India patent search · Official patent journals