Po-Yu Wu is a notable inventor based in Palo Alto, CA (US). He has made significant contributions to the field of machine learning, holding a total of 3 patents. His work focuses on enhancing the efficiency and effectiveness of machine learning processes through innovative methods.

Latest Patents

One of Po-Yu Wu's latest patents is titled "Overlapping Gradient Synchronization in Machine Learning." In this method, an orchestrator of a computing system determines that results of machine learning model computations are available and dispatches a worker to perform model computations that include computing gradients of the results. The orchestrator also determines that a set of gradients of the results is available and dispatches a gradient worker to compute a sum of the gradients. Additionally, the orchestrator determines that a second set of gradients is available and dispatches a second gradient worker to compute a sum of the second set of gradients. The orchestrator then determines that the sums of the first and second gradients are available and dispatches a third gradient worker to compute synchronized gradients. This process allows gradient workers to compute sums and synchronized gradients concurrently with training workers computing additional model computation results and/or gradients. A computer program product can include this method, and a computing system can incorporate the orchestrator.

Another significant patent is "Dependency-Aware Server Processing of Dataflow Applications." This computer-implemented method involves a server processing work requests from a work requester. The work requester can communicate a processing dependency of one work request on a second work request to the server. The server associates the dependency with the work requests and/or a queue of work requests. The dependency includes a condition to be met in association with processing the work requests, which may involve an action for the server to take. A computing system can comprise a work requester, a server, and a set of dependency-aware queues for processing a set of work requests. The server processes work requests enqueued to the queues in an order based on the dependencies.

Career Highlights

Po-Yu Wu is currently employed at Sambanova Systems, Inc., where he continues to innovate in the field of machine learning. His work has been instrumental in advancing the capabilities of data processing and machine learning applications.

Collaborations

Some of his notable coworkers include Subhra Mazumdar and Greg Dykema, who contribute to the

Profile summary based on public USPTO records.

Please report any incorrect information to [email protected]