Gábor Bartók

Title: Gábor Bartók: Innovator in Multi-Task Learning

Introduction

Gábor Bartók is a notable inventor based in Zurich, Switzerland. He has made significant contributions to the field of machine learning, particularly in the area of multi-task learning. His innovative approach has garnered attention in the tech community.

Latest Patents

Gábor Bartók holds a patent for "Flexible parameter sharing for multi-task learning." This patent outlines systems and methods that enable flexible parameter sharing across multiple tasks. The training method described in the patent includes obtaining a test input, selecting a specific task, and training a multi-task machine-learned model. This is achieved by performing a forward pass using the test input and connection probability matrices to generate a sample distribution of test outputs. The components of the model are trained based on this distribution, followed by a backwards pass to refine the connection probability matrix using a straight-through Gumbel-softmax approximation.

Career Highlights

Gábor Bartók is currently employed at Google Inc., where he continues to push the boundaries of machine learning technologies. His work is instrumental in developing systems that enhance the efficiency and effectiveness of multi-task learning.

Collaborations

Gábor has collaborated with talented individuals such as Effrosyni Kokiopoulou and Krzysztof Stanislaw Maziarz. These collaborations have contributed to the advancement of innovative solutions in the field.

Conclusion

Gábor Bartók is a pioneering inventor whose work in multi-task learning is shaping the future of machine learning technologies. His contributions are vital to the ongoing evolution of this field.