Koustava Goswami

Title: Koustava Goswami: Innovator in Text-to-Image Synthesis and Document Question Answering

Introduction

Koustava Goswami is a prominent inventor based in Bengaluru, India. He has made significant contributions to the fields of artificial intelligence and machine learning, particularly in text-to-image synthesis and document question answering. With a total of 4 patents, his work showcases innovative approaches to complex problems in these domains.

Latest Patents

Goswami's latest patents include groundbreaking technologies that enhance the capabilities of neural networks. One of his notable inventions is titled "Text-to-image synthesis utilizing diffusion models with test-time attention segregation and retention optimization." This patent focuses on systems and methods that utilize attention segregation loss and attention retention loss to generate text-conditioned images. By comparing attention maps for multiple concepts, the system effectively reduces overlap and improves information retention across denoising steps.

Another significant patent is "Evidence retrieval for long document question answering using large language models." This invention outlines a method for answering questions related to long documents by leveraging large language models. The process involves obtaining a representation of the document and identifying relevant sections to determine accurate answers to user queries.

Career Highlights

Koustava Goswami is currently employed at Adobe Inc., where he continues to push the boundaries of innovation in technology. His work at Adobe has allowed him to collaborate with other talented professionals in the field, contributing to the development of advanced AI solutions.

Collaborations

Some of his notable coworkers include Inderjeet Nair and Shwetha S. Together, they work on various projects that aim to enhance the functionality and efficiency of AI systems.

Conclusion

Koustava Goswami is a remarkable inventor whose contributions to text-to-image synthesis and document question answering are paving the way for future advancements in artificial intelligence. His innovative patents reflect a deep understanding of complex systems and a commitment to improving technology.