Advanced Micro Devices had 27 patents in artificial intelligence during Q3 2023. Advanced Micro Devices Inc filed 9 patents in Artificial Intelligence theme during Q3 2023. Advanced Micro Devices Inc filed several patents during Q3 2023. One patent describes an apparatus that modifies the behavior of an artificial agent based on its similarity to human behavior. Another patent discusses an inference server that processes batches of inference requests from client systems. A third patent introduces a rearranger circuit that rearranges data elements of raw images for neural network training or inference. Lastly, a patent presents a safety monitor framework for a safety-critical inference application that detects erroneous processing by comparing outputs to a known good result. GlobalData’s report on Advanced Micro Devices gives a 360-degreee view of the company including its patenting strategy. Buy the report here.

Advanced Micro Devices grant share with artificial intelligence as a theme is 59% in Q3 2023. Grant share is based on the ratio of number of grants to total number of patents.

Recent Patents

Application: Quantifying the human-likeness of artificially intelligent agents using statistical methods and techniques (Patent ID: US20230274168A1)

The patent filed by Advanced Micro Devices Inc. describes an apparatus and method for modifying the behavior of an artificial agent based on its similarity to human behavior. The apparatus includes a processor that determines a first distribution of behavior associated with the artificial agent and a second distribution of behavior associated with a user. The processor then generates a human-likeness similarity measurement by comparing the first distribution to the second distribution. If the similarity measurement fails to meet a similarity threshold, the behavior of the artificial agent is modified.

The method involves determining the first distribution and the second distribution by vectorizing sets of behavior information into a plurality of vectors. These vectors represent different subsequences of behavior. The method also includes transforming the vectors into visual representations. The first distribution consists of the vectors or visual representations from the first set of behavior information, while the second distribution consists of the vectors or visual representations from the second set of behavior information.

To generate the similarity measurement, a two-sample hypothesis test is performed. This involves operating in two modes. In the first mode, the first distribution and the second distribution are evaluated separately to generate a distribution of distances. A test statistic is then generated based on this distribution. In the second mode, the first distribution and the second distribution are combined to create a pooled sample distribution of behaviors. The pooled sample distribution is then evaluated to generate a second distribution of distances. The similarity measurement is generated based on the test statistic and the second distribution of distances.

The method also includes modifying the behavior of the artificial agent based on the similarity measurement. This modification can involve changing one or more behaviors of the artificial agent.

Additionally, the patent describes an apparatus that includes a processor configured to perform the method. The processor determines the first distribution and the second distribution, generates the similarity measurement, and modifies the behavior of the artificial agent based on the similarity measurement.

In another aspect of the invention, the method involves determining the similarity between the behavior of two different artificial agents. The behavior of each artificial agent is represented by an episode, which is a sequence of behaviors from an initial state to a terminal state. The method includes sampling trajectories from each episode, vectorizing the trajectories into vectors, and determining the first distribution and the second distribution based on these vectors. The similarity measurement is then generated and used to modify the behavior of at least one of the artificial agents.

Overall, the patent describes a system and method for assessing the human-likeness of an artificial agent's behavior and modifying it based on this assessment.

To know more about GlobalData’s detailed insights on Advanced Micro Devices, buy the report here.

Premium 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

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.