Julio Zamora Esquivel

Dr. Julio Zamora is Principal Engineer and Senior Research Scientist Manager at Intel Labs, Leading globally the Human Robot Collaboration Group as a part of the Intelligent System Research Group. He received a master's degree in Computer Sciences and a PhD in Electric Engineering from CINVESTAV. Dr. Zamora had a post-doctoral position at KAIST, Korea. He was nominated for the W.K. Clifford international prize for his contributions to geometric algebra, introducing Quadric Geometric Algebra, and the formulation of Robot dynamics in terms of octonions. He is a member of the National Research System, the Mexican Association for Computer Vision, Neural Computing and Robotics, and a Senior member of IEEE. He has more than 106 patents in process and more than 30 publications in journals, book chapters, and conference proceedings. His research interests include Artificial Intelligence, Computer Vision, Geometric Algebras, Robotics, and Image Processing.

Contact: gepadiza@yahoo.com

Abstracts

Introduction mathematical framework

To model a robotic arm and compute the differential kinematics of the end effector represented by a circle in a three dimensional space, described as a bi-vector of conformal geometric algebra. Additionally by using a circle to describe the grasping pose on the object. create a differential kinematics based control law to guide the robot arm in order to minimize the error between the griper's circle  and the target circle. Since the circle has 3 degrees of freedom for the center, two degrees for the orientation and one more for radius we can use to describe the End-Effector pose and our control law adjusts the position and the orientations simultaneously

Introduction of Deformable Fractional Filters (DFFs) for Convolutional Neural Networks (CNNs). 

DFFs enhance the efficiency of conventional Deformable Convolutional Filters by introducing a compression mechanism rooted in techniques from fractional calculus. Concretely, our method reduces the parameter overhead requirement of convolutional filters by replacing the kernel with a fractional approximation, which can be trained using only three parameters -- regardless of the kernel size. DFFs present a compelling use case for the compression of networks that require large kernel sizes. To demonstrate the benefits of DFFs, we report experimental results across a diverse set of computer vision problem domains, including classification and semantic segmentation.

Region: Latin America