http://repository.iiitd.edu.in/xmlui/handle/123456789/618 Fri, 04 Sep 2026 11:43:22 GMT 2026-09-04T11:43:22Z http://repository.iiitd.edu.in/xmlui/handle/123456789/2093 DNA-Aligner Lakshya; Sengupta, Debarka (Advisor) The DNA-ALIGNER is a novel model designed to align mRNA and gene sequences. In cellular biology, DNA is transcribed into mRNA, which serves as the template for protein synthesis. DNA-ALIGNER addresses the challenge of aligning gene and mRNA sequences by leveraging pre-trained embeddings from the Nucleotide Transformer. The model employs a multi-tower architecture, including mRNATower, GeneTower, and middleTower, to process both gene and mRNA sequences independently and jointly. By performing sequence alignment, DNA-ALIGNER learns meaningful numerical representations of gene data, which can be utilized for downstream tasks in various biological applications. The model uses masked language modeling (MLM) and sequence reconstruction techniques to ensure accurate alignment. Loss functions, including cross-entropy for MLM and DICE loss for sequence reconstruction and Alignment, optimize the alignment and learning process. The DNA-ALIGNER provides a powerful tool for bioinformatics research, enabling a better under- standing of gene sequences, mutation analysis, and application in task-specific networks. Wed, 27 Nov 2024 00:00:00 GMT http://repository.iiitd.edu.in/xmlui/handle/123456789/2093 2024-11-27T00:00:00Z http://repository.iiitd.edu.in/xmlui/handle/123456789/2092 Navigating fashion trends in the digital age Raj, Rohit; Arya, Tarun Kumar; Satyam; Haripriya, Soibam (Advisor) Fashion is an integral part of human life. It extends beyond clothing and accessories, serves as a medium for self-expression, and reflects an individual's personality and cultural identity. In this study, we examine the complex and dynamic relationship between fashion influencers and individuals on social media, by examining the role of social media in constantly changing and reshaping the creation, dissemination and consumption of fashion. Social media has transformed fashion into a more interactive and inclusive space, by allowing individuals to share their views and express their opinion on style and fashion which was previously limited to fashion magazines and Television. One category of such individuals is known as social media based Fashion Influencers. This type of influencers mainly use platforms such as Instagram and YouTube to share their content in which they provide fashion-related tips and suggestions, review products and collaborate with different brands, through which they cater to a wider range of demographics. Fashion influencers have become pivotal in shaping contemporary fashion trends and consumer behavior. Our study mainly focuses on analyzing the content and engagement strategies of top 6 social media based fashion influencers. We have achieved this by examining the content of their posts, captions and the comments they receive from their audience, to identify and capture the underlying themes that define and characterize the interaction between the influencer and their followers and how they communicate and convey fashion to their audience. Apart from that, our study also delves deeper into the broader implications of social media on the field of fashion which includes topics such as democratization of fashion, fast fashion cycles and the potential for promoting sustainability through digital platforms. Wed, 27 Nov 2024 00:00:00 GMT http://repository.iiitd.edu.in/xmlui/handle/123456789/2092 2024-11-27T00:00:00Z http://repository.iiitd.edu.in/xmlui/handle/123456789/2085 Finding optimized path for harvesting and other applicable scenarios Singh, Digvijay; Shankhwar, Kalpana (Advisor) In modern agriculture, the imperative of efficiency and productivity underscores the need for optimized navigation and monitoring practices to meet the demands of a burgeoning global population. Effective traversal and data collection in agricultural fields necessitates meticulous planning to ensure judicious allocation of resources and maximized yield. However, traditional approaches to path optimization for drones have been fraught with challenges, often leading to inefficiencies and resource wastage due to manual assessment and suboptimal route planning. The challenge lies in devising a solution that can automate the pathfinding process, optimize traversal routes, and minimize resource consumption while maximizing yield. To address this challenge, our research aims to develop an innovative approach that integrates advanced algorithms and simulation techniques to streamline the path optimization process for UAVs in agricultural fields. The primary goal is to leverage the capabilities of the Rapidly-exploring Random Tree (RRT) algorithm, renowned for its efficiency in exploring large search spaces and finding feasible paths in complex environments. By combining the RRT algorithm with the immersive capabilities of a 3D simulation environment, we aim to create a powerful tool that can revolutionize pathfinding and navigation for UAVs in agricultural settings. The key innovation of our approach lies in its ability to dynamically adjust UAV traversal routes based on real-time field conditions and obstacles. By modeling the agricultural field with crops, trees, and various obstacles within the 3D simulation environment, researchers can interact with and analyze the pathfinding process, allowing for iterative adjustments and fine-tuning to optimize UAV routes. The RRT algorithm intelligently navigates through the field, taking into account both the feasibility of paths and the need to avoid obstacles. This adaptive approach ensures that the algorithm identifies the most efficient route while dynamically adapting to field conditions and obstacles. Moreover, our approach offers scalability and adaptability to different agricultural environments and field conditions. Whether navigating through large-scale commercial farms or small-scale family-owned operations, the algorithm can be customized to suit specific requirements and optimize UAV routes accordingly. This versatility makes our approach well-suited for a wide range of agricultural applications and challenges. In addition to optimizing UAV routes, our research also aims to enhance monitoring and data collection efficiency. By automating the pathfinding process and strategically navigating through the field, drones can efficiently target areas in trees with less vegetation, areas in need of pesticides, and other critical zones. This capability allows for precise monitoring, data collection, and resource application, leading to higher levels of efficiency and productivity while minimizing resource wastage. The optimized routes generated by our approach result in reduced fuel consumption, machinery wear and tear, and overall operational costs. This transformative impact on UAV efficiency underscores the potential of our approach to drive sustainable productivity in the agricultural sector. Wed, 27 Nov 2024 00:00:00 GMT http://repository.iiitd.edu.in/xmlui/handle/123456789/2085 2024-11-27T00:00:00Z http://repository.iiitd.edu.in/xmlui/handle/123456789/2083 RecipeGPT implementation Gupta, Anmol; Gupta, Bhavya; Bagler, Bagler, Ganesh (Advisor) The culinary domain offers significant opportunities for leveraging AI to simplify and enhance recipe-related interactions. This project focuses on developing an intelligent recipe chatbot that performs two primary tasks: (1) generating cooking instructions from a list of ingredients, and (2) predicting ingredients based on given instructions. Using the ”51k Recipes” dataset, which contains detailed recipe information, we explored and fine-tuned various large language models (LLMs), including GPT-2, T5 (Text-To-Text Transfer Transformer), Gemini, etc, to identify the most effective approach for these tasks. Key challenges addressed include handling large datasets, ensuring the relevance and accuracy of outputs, and optimizing models for bidirectional task processing. Through comprehensive preprocessing and fine-tuning using PyTorch, the chatbot was trained to deliver coherent and contextually appropriate responses. The models were evaluated using metrics such as BLEU scores and human feedback, with notable improvements observed in instruction fluency and ingredient prediction accuracy. This project demonstrates the potential of LLMs in culinary applications and sets the ground- work for creating versatile, user-friendly tools for home cooks, professional chefs, and culinary learners. Future work involves refining dataset quality, exploring larger instruction-tuned models, and integrating the chatbot into practical applications such as voice assistants and cooking platforms. Sun, 08 Dec 2024 00:00:00 GMT http://repository.iiitd.edu.in/xmlui/handle/123456789/2083 2024-12-08T00:00:00Z