Hafiz Syed Muhammad HurFA22-RMT-037Dr. Hafiz Muhammad Afzal SiddiquiLHR TP 93812026-03-172024-03-17https://repository.cuilahore.edu.pk/handle/123456789/2908In this work, we introduce the identification of the MNIST database, which will be in handwritten digits that the machine can identify. The human handwriting form may be de tected and converted into computer language. We use several machine learning algorithms, including Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), and Bidirectional Long Short-Term Memory (BiLSTM). The MNIST database was developed using binary images of handwritten numbers (09) from NIST’s Special Database.Second, introduce the identification of TB photos. Tuber culosis is the biggest cause of mortality worldwide, according to the World Health Organi zation. Inadequate treatment and delayed or incorrect diagnosis have led to several cases of the illness. Accurate and timely diagnosis is critical for successfully managing and preventing tuberculosis. Despite significant progress in deep learning for medical image processing. There are two types of distributions: training and application data.Our findings show that transfer learning from a pre-trained vision transformer outperforms a pre-trained CNN in medical imagingenDepartment of MathematicsMathematicsFA22Artificial IntelligenceVisual PredictionTransfer LearningThesis