Megha Nawhal

Title: Innovations by Megha Nawhal

Introduction

Megha Nawhal is an accomplished inventor based in Vancouver, Canada. She has made significant contributions to the field of machine learning and data generation, holding 2 patents that showcase her innovative approach to complex data challenges.

Latest Patents

One of her latest patents is titled "System and method for machine learning architecture for partially-observed multimodal data." This invention addresses the limitations of conventional Variational Autoencoders (VAEs) that operate with fully-observed data during training. Megha's modified VAE framework allows for learning from partially-observed data, conditioned on a fully-observed mask. This model is capable of learning a proper proposal distribution based on the missing data and has been evaluated for both high-dimensional multimodal data and low-dimensional tabular data.

Another notable patent is "System and method for generation of unseen composite data objects." This computer-implemented system generates data structures representing unseen compositions based on two categories observed individually in a training dataset. During the training of a generator, her proposed framework utilizes various discriminators, including three pixel-centric discriminators and one object-centric relational discriminator. These discriminators ensure spatial and temporal consistency across frames while leveraging spatio-temporal scene graphs to reason over object layouts in videos.

Career Highlights

Megha Nawhal currently works at the Royal Bank of Canada, where she applies her expertise in machine learning and data science. Her work focuses on developing innovative solutions that enhance data processing and analysis.

Collaborations

Throughout her career, Megha has collaborated with notable colleagues, including Gregory Mori and Yu Gong. These collaborations have further enriched her research and development efforts in the field of machine learning.

Conclusion

Megha Nawhal's contributions to the field of machine learning and data generation are noteworthy. Her innovative patents reflect her commitment to advancing technology and solving complex data challenges.