Szerző szerinti böngészés "Kumar, Dheeraj"

Megjelenítve 1 - 2 (Összesen 2)
  • Nincs kép
    TételKorlátozottan hozzáférhető
    Developing AI Agents for Hazard‐Aware Game Environments Using Reinforcement Learning
    Kumar, Dheeraj; Harangi, Balázs; DE--Informatikai Kar
    This project is Unity-based multi-agent using reinforcement learning environment featuring a 2 players. The primary agent and a spider-agent, is trained using ML-Agents to chase prey agent within a 3D arena filled with static obstacles, collectible coins, and hazards like poison food. The environment is designed to challenge agents with navigation complexity, partial observations, and decision-making under pressure. The project highlights the potential of reinforcement learning in game AI and showcases Unity as a powerful platform for simulating complex multi-agent scenarios.
  • Bélyegkép
    TételKorlátozottan hozzáférhető
    Identifying cancer using MRI and CT scans
    Kumar, Dheeraj; Tomán, Henrietta; DE--Informatikai Kar
    This thesis presented a thorough investigation into developing a machine learning model for the classification of Renal Carcinoma Cancer in medical images. The primary aim was to create an accurate and efficient classifier capable of differentiating between malignant and non-cancerous images, thus contributing to earlier cancer detection and improved patient outcomes. The research commenced with an extensive literature review and selection of five renal cancer-related datasets. The chosen dataset was preprocessed using techniques such as outlier removal, denoising, edge detection, histogram equalization, and data augmentation. Several machine learning models, including ANN, SVM, CNN's VGG16, and Transformer's Swin Transformer, were assessed. The SVM model achieved the highest level of accuracy and precision, while the Transformer model exhibited the best recall. The user-friendly website provided instant feedback on whether the uploaded image was malignant or noncancerous.