Microsoft had 599 patents in artificial intelligence during Q2 2024. Microsoft Corp’s Q2 2024 patents include a neural transcompilation model for identifying and correcting syntax translation defects, a method for detecting potential model inversion attacks in data processing requests, a computing apparatus for generating video transcripts and organizing keyframes based on topics, a technology for generating high-resolution images from low-resolution images using machine learning models, and an automated frame skipping approach for efficiently processing frames in surveillance imagery. GlobalData’s report on Microsoft gives a 360-degree view of the company including its patenting strategy. Buy the report here.
Smarter leaders trust GlobalData
Access deeper industry intelligence
Experience unmatched clarity with a single platform that combines unique data, AI, and human expertise.
Microsoft had no grants in artificial intelligence as a theme in Q2 2024.
Recent Patents
Application: Syntax unit testing and fine-tuning of neural transcompilation models (Patent ID: US20240211224A1)
The patent filed by Microsoft describes a neural transcompilation model that is tested with syntax unit tests to identify syntax elements in a source code program that fail to translate properly into a target programming language. These identified syntax elements are ranked based on translation failure rates, and the model is fine-tuned using training samples of elements with the highest failure rates to teach the model the correct associations between syntax elements causing translation defects and their correct translations. The system and method outlined in the patent involve obtaining syntax unit tests, executing the neural transcompilation model, detecting syntax translation defects, and fine-tuning the model with training data to improve translation accuracy.
The system includes a processor and memory storing a program that executes actions such as deploying the fine-tuned model in an integrated development environment, executing translations with input values to detect defects, transforming defective syntax elements, and generating correct translations. The method involves generating translations, detecting defects, training the model with paired samples, computing failure rates, and ranking defects based on these rates. The patent also covers hardware storage devices with executable instructions for executing the neural transcompilation model, identifying defects, creating training datasets, and training the model to improve translation accuracy. The model can be a neural transformer model with attention or a recurrent neural network, and the training dataset includes programs with the highest failure rates for effective learning.
To know more about GlobalData’s detailed insights on Microsoft, buy the report here.
Data Insights
From
The gold standard of business intelligence.
Blending expert knowledge with cutting-edge technology, GlobalData’s unrivalled proprietary data will enable you to decode what’s happening in your market. You can make better informed decisions and gain a future-proof advantage over your competitors.
GlobalData, the leading provider of industry intelligence, provided the underlying data, research, and analysis used to produce this article.
GlobalData Patent Analytics tracks bibliographic data, legal events data, point in time patent ownerships, and backward and forward citations from global patenting offices. Textual analysis and official patent classifications are used to group patents into key thematic areas and link them to specific companies across the world’s largest industries.
