Abhay Nath

Project ID: PRJ2025DIT013
Guide: Ms. Chintal Raval

This thesis utilizes Deep Learning and Generative AI to address oncology issues with diagnostics and therapeutics improvement. It presents a hybrid model for protein-ligand binding prediction (molecular docking), novel algorithms for anti-cancer drugs discovery, and superior CNNs for analysis of histology images. Also, it uses Generative Adversarial Networks for drug design, and generation of radiological (MRI & CT) images. Making… Show more

Project ID: PRJ2025DIT013
Guide: Ms. Chintal Raval

This thesis utilizes Deep Learning and Generative AI to address oncology issues with diagnostics and therapeutics improvement. It presents a hybrid model for protein-ligand binding prediction (molecular docking), novel algorithms for anti-cancer drugs discovery, and superior CNNs for analysis of histology images. Also, it uses Generative Adversarial Networks for drug design, and generation of radiological (MRI & CT) images. Making major advancements towards accuracy, efficiency, and scalability, this work opens doors for precision medicine with reducing the subjectivity of diagnosis and simplifying the drug development process. All the results are validated through in silico, in vitro and in vivo biological studies. Show less

A robust, Fusion-based Deep Learning Model to Enhance Calcium Detection Accuracy in Echocardiography Images. By integrating information from multiple imaging modalities, the study sought to improve diagnostic precision and support early intervention.

Pioneering the future of cancer research, my AI model seamlessly integrates advanced algorithms and interactive visualizations to accelerate the discovery of anti-cancer drugs, offering a transformative approach to precision medicine.

Developed an enhanced YOLOv8s-based model for real-time road object detection, achieving 97.9% accuracy using advanced optimization techniques like He initialization, LARS, SWA, and AGC. Innovations such as SE blocks and FPN improve tiny object recognition, making it ideal for autonomous driving and smart traffic management with high precision and efficiency.

We implemented a dynamic association rule mining algorithm to uncover valuable item relationships and frequent itemsets. We enhanced pattern interpretation using a custom synergy score combined with Zhang’s score, yielding a more sophisticated understanding of associations than standard metrics. Our approach achieved a highest synergy score of 1.097 and an average above 1, indicating more accurate, relevant rules. It outperformed comparative deep learning models and offers retailers actionable… Show more

We implemented a dynamic association rule mining algorithm to uncover valuable item relationships and frequent itemsets. We enhanced pattern interpretation using a custom synergy score combined with Zhang’s score, yielding a more sophisticated understanding of associations than standard metrics. Our approach achieved a highest synergy score of 1.097 and an average above 1, indicating more accurate, relevant rules. It outperformed comparative deep learning models and offers retailers actionable insights for personalized marketing, inventory control, and data-driven strategy development to better serve customers. Show less

Developed a machine learning-based system for lime disease classification using spectrometry data, achieving up to 99.98% precision with models like LSTM, RNN, and BiLSTM combined with classifiers such as SVM and XGBoost. This approach significantly improves disease detection accuracy, aiding in better crop protection and agricultural efficiency.

Identification of Different Medicinal Plants/Raw materials through Using Machine Learning Algorithms.

Roles-

🔹Abhay Nath: AI/ML
🔹Mann Mendapara: AI/ML
🔹Nirav Thummar: Android
🔹Rishi Vachhani: Android
🔹Goldi Ladani: Dataset
🔹Urvesh Bhesdania: Dataset

• RUDRA - India's Own Artificial Intelligence: a project unveiling India's AI-powered virtual voice assistant. RUDRA, short for Responsive, Unsupervised, Data-driven, Researchable, and Assistant, has gained widespread acclaim, becoming viral on platforms like LinkedIn and attracting attention from international startups in the USA, Canada, Bangladesh and Saudi Arabia. It has also garnered interest from colleges across India, spanning Punjab, Kerala, Hyderabad, and Assam. RUDRA's success has not… Show more

• RUDRA - India's Own Artificial Intelligence: a project unveiling India's AI-powered virtual voice assistant. RUDRA, short for Responsive, Unsupervised, Data-driven, Researchable, and Assistant, has gained widespread acclaim, becoming viral on platforms like LinkedIn and attracting attention from international startups in the USA, Canada, Bangladesh and Saudi Arabia. It has also garnered interest from colleges across India, spanning Punjab, Kerala, Hyderabad, and Assam. RUDRA's success has not only brought recognition but also opened doors for my research opportunities in AI. Achieving 1st positions in various competitions further solidifies its leadership in the AI domain. Celebal Technologies from Jaipur expressed interest in this project, reaching out to our university's Training and Placement Cell with an offer for placement while I was in my early third year of B.Tech, highlighting its significant impact and potential in the AI landscape. Show less

Advancing Cancer Histology Analysis through AI-driven Machine Learning and Deep Learning Integration. This project focuses on leveraging Artificial Intelligence (AI) alongside Machine Learning (ML) and Deep Learning (DL) to revolutionize cancer histology analysis. By integrating advanced algorithms and neural networks, our system can accurately interpret intricate patterns within histological images. Through comprehensive training on diverse datasets, our AI model adeptly identifies subtle… Show more

Advancing Cancer Histology Analysis through AI-driven Machine Learning and Deep Learning Integration. This project focuses on leveraging Artificial Intelligence (AI) alongside Machine Learning (ML) and Deep Learning (DL) to revolutionize cancer histology analysis. By integrating advanced algorithms and neural networks, our system can accurately interpret intricate patterns within histological images. Through comprehensive training on diverse datasets, our AI model adeptly identifies subtle biomarkers indicative of various cancer types and stages, providing clinicians with precise diagnostic insights. Continuously evolving, our adaptable system bridges the gap between traditional histopathology and AI technologies, facilitating earlier detection, improved patient outcomes, and advancements in personalized medicine. Show less

AI Powered Platformer (Super Mario) Game with Dynamic Level Generation and AI-Powered Adaptive Difficulty. Crafted a unique version of the classic Super Mario game integrating dynamic level generation algorithms driven by artificial intelligence. Utilizing AI-powered adaptive difficulty, the game adjusts the level's complexity in real-time based on the player's skill level and performance. Each playthrough offers dynamically generated levels tailored to the player's abilities, ensuring a… Show more

AI Powered Platformer (Super Mario) Game with Dynamic Level Generation and AI-Powered Adaptive Difficulty. Crafted a unique version of the classic Super Mario game integrating dynamic level generation algorithms driven by artificial intelligence. Utilizing AI-powered adaptive difficulty, the game adjusts the level's complexity in real-time based on the player's skill level and performance. Each playthrough offers dynamically generated levels tailored to the player's abilities, ensuring a challenging yet enjoyable gaming experience. This innovative fusion of AI and dynamic level generation redefines traditional gameplay, providing players with personalized and engaging adventures.

Partner: Sneha Dass has significantly simplified the code complexity, making it easier to maintain and eliminating the unnecessary load on the main file. Her contributions have significantly enhanced the scalability of this project. Show less