PROPOSED PROJECTS
This project investigates the effects of image permutations on neural network classification performance. By applying various tile-wise permutations to input images, we explore how spatial rearrangements impact classification accuracy. The study focuses on understanding the role of tile resolution and determining whether neural networks adapt to these permutations or rely on inherent image features.
This project explores an innovative approach to vector database searching, particularly for retrieval-augmented text generation applications. By employing a novel thresholding mechanism based on higher criticism, we aim to improve the process of finding relevant matches between a vector query and database elements. The project seeks to address challenges in information retrieval from large-scale datasets, potentially enhancing the accuracy and efficiency of retrieval-augmented generation techniques in natural language processing.
Curabitur varius ex eu est scelerisque, a suscipit diam imperdiet. Proin rutrum dapibus lorem ac aliquet. Morbi faucibus velit sed mi viverra condimentum. Phasellus ornare mauris sollicitudin elementum iaculis. Morbi vel sagittis lacus, vitae semper augue. Integer a posuere ipsum. Sed aliquet lobortis lacus, rutrum ultricies quam porta vitae. Etiam sit amet nisl ac tortor bibendum maximus.
Proin rutrum dapibus lorem ac aliquet. Morbi faucibus velit sed mi viverra condimentum. Phasellus ornare mauris sollicitudin elementum iaculis. Morbi vel sagittis lacus, vitae semper augue. Integer a posuere ipsum. Sed aliquet lobortis lacus, rutrum ultricies quam porta vitae. Etiam sit amet nisl ac tortor bibendum maximus.
Curabitur varius ex eu est scelerisque, a suscipit diam imperdiet. Proin rutrum dapibus lorem ac aliquet. Morbi faucibus velit sed mi viverra condimentum. Phasellus ornare mauris sollicitudin elementum iaculis. Morbi vel sagittis lacus, vitae semper augue. Integer a posuere ipsum. Sed aliquet lobortis lacus, rutrum ultricies quam porta vitae. Etiam sit amet nisl ac tortor bibendum maximus.
Curabitur varius ex eu est scelerisque, a suscipit diam imperdiet. Proin rutrum dapibus lorem ac aliquet. Morbi faucibus velit sed mi viverra condimentum. Phasellus ornare mauris sollicitudin elementum iaculis. Morbi vel sagittis lacus, vitae semper augue. Integer a posuere ipsum. Sed aliquet lobortis lacus, rutrum ultricies quam porta vitae. Etiam sit amet nisl ac tortor bibendum maximus.
