Mohammad K Ebrahimpour

Title: Innovations of Mohammad K Ebrahimpour

Introduction

Mohammad K Ebrahimpour is a notable inventor based in Fremont, CA, who has made significant contributions to the field of machine learning and image processing. With a total of five patents to his name, he has developed innovative techniques that enhance the capabilities of machine learning models in image captioning and object detection.

Latest Patents

One of his latest patents is titled "Image captioning with weakly-supervised attention penalty." This invention discloses techniques for training a machine-learning model to caption images. The process involves generating a plurality of feature vectors and visual attention maps based on an input image. Each feature vector corresponds to different regions of the image, and caption attention maps are generated based on these feature vectors. An attention penalty is calculated by comparing the caption attention maps with the visual attention maps, leading to the calculation of a loss function that is used to train the visual and attention models.

Another significant patent is "Ventral-dorsal neural networks: object detection via selective attention." This methodology focuses on efficient object classification within a visual medium. It utilizes a first neural network for attention-based object localization, generating a visual mask that is applied to the visual medium. The masked visual medium is then processed by a second neural network to detect and classify objects effectively.

Career Highlights

Mohammad K Ebrahimpour is currently employed at Ancestry.com Operations Inc., where he continues to innovate and contribute to advancements in technology. His work has been instrumental in enhancing the functionality of machine learning applications in various domains.

Collaborations

He has collaborated with talented coworkers, including Yen-Yun Yu and Jiayun Li, to further his research and development efforts in the field.

Conclusion

Mohammad K Ebrahimpour's contributions to machine learning and image processing through his patents demonstrate his expertise and innovative spirit. His work continues to influence advancements in technology and improve the capabilities of machine learning models.