José Lezama Torres De La Llosa

Title: José Lezama Torres De La Llosa: Inventor of Temporal Sequence Generation Technology

Introduction

José Lezama Torres De La Llosa is an inventor based in Atlanta, Georgia, and works at Google Inc. His patent work concerns methods for generating temporal sequences with diffusion transformer neural networks. He is credited with 1 patent.

Latest Patents

Lezama Torres De La Llosa is named on the patent “Generating Temporal Sequences Using Diffusion Transformer Neural Networks.” The invention describes methods, systems, and software for generating an output sequence of data elements conditioned on an input.

The method begins with an input that includes noise representations for multiple segments of a temporal sequence. A latent denoising neural network updates the representation for each segment over multiple iterations. For segments after the first, the network can use a memory vector representing hidden states produced while processing preceding segments. The updated latent representations are then processed to generate the output sequence.

Career Highlights

His listed professional affiliation is Google Inc. The patent reflects work on neural-network techniques for producing temporally structured data, with memory from earlier segments informing the processing of later ones.

Collaborations

The patent’s listed coworkers include Irfan Essa and David A. Ross. Their association highlights a collaborative context for the invention.

Conclusion

José Lezama Torres De La Llosa’s patent focuses on using diffusion transformer neural networks and segment-level memory to generate conditioned temporal sequences.