Tom D J Sercu

Title: The Innovations of Tom D J Sercu

Introduction

Tom D J Sercu is a notable inventor based in New York, NY (US). He has made significant contributions to the field of computing and data analysis. With a total of 2 patents, Sercu's work focuses on advanced methodologies for estimating mutual information and generating attribute-based samples.

Latest Patents

Sercu's latest patents include "Mutual information neural estimation with Eta-trick" and "Generating attribute-based samples." The first patent describes a computing device that receives data X and Y, each containing N samples. A function f(x,y) is defined as a trainable neural network based on the data X and Y. The process involves creating a permuted version of the data Y and computing a loss mean based on the neural network and the permuted data. The goal is to minimize the loss concerning a scalar variable and the neural network, ultimately estimating the mutual information between the test data X and Y. If the estimated mutual information exceeds a predetermined threshold, the data is considered dependent; otherwise, it is deemed independent. The second patent outlines a computer-implemented method that includes training a latent variable model using both labeled and unlabeled data, as well as generating new data with a predetermined set of labels.

Career Highlights

Tom D J Sercu is currently employed at International Business Machines Corporation, commonly known as IBM. His work at IBM has allowed him to explore innovative solutions in data processing and machine learning.

Collaborations

Sercu has collaborated with notable colleagues, including Youssef Mroueh and Pierre L Dognin. Their combined expertise contributes to the advancement of technology in their respective fields.

Conclusion

Tom D J Sercu is a prominent inventor whose work in data analysis and computing continues to influence the industry. His patents reflect a commitment to innovation and the development of advanced methodologies.