Lusine Abrahamyan

Title: Lusine Abrahamyan: Distributed Learning Innovation

Introduction

Lusine Abrahamyan is an inventor based in Leuven, Belgium. Her work includes a patented method for training learning models through distributed computing.

Latest Patents

Abrahamyan is associated with 1 patent, titled “Method for a Distributed Learning.” The invention describes a computer-implemented approach in which computing nodes train a learning model using training data and derive gradient information for updating it. The nodes encode that information by exploiting correlations among their respective gradients, exchange the encoded information, determine an aggregate gradient, and use it to update the model. This approach is designed to support distributed model training.

Career Highlights

Abrahamyan has worked with imec and Vrije Universiteit Brussel. Her patent reflects work in distributed learning and machine-learning systems.

Collaborations

Her listed coworker is Nikolaos Deligiannis, with whom she is associated professionally.

Conclusion

Lusine Abrahamyan’s patented method addresses the exchange and aggregation of gradient information across computing nodes, contributing to approaches for distributed learning.