The project titled "Manifold Energy Projection Twin Support Vector Machine for Semi-Supervised Classification" focuses on improving the classification performance of traditional Projection Twin Support Vector Machines (PTSVM) by addressing the challenges posed by the reliance on fully labeled datasets in real-world applications. In this research, we propose a novel semi-supervised learning model, the Manifold Energy Projection Twin Support Vector Machine (MEPTSVM), which effectively integrates both labeled and unlabeled data. This method aims to generate more robust and generalizable decision boundaries, leveraging the inherent structure of unlabeled data for enhanced classification. By incorporating graph-regularized techniques and manifold learning, MEPTSVM extends the capabilities of PTSVM, making it a powerful tool for real-world classification problems where labeled data is scarce. The proposed model is evaluated on several semi-supervised classification tasks using the ORL face dataset, demonstrating significant improvements in classification accuracy and generalization performance over existing methods. The project also compares MEPTSVM with other semi-supervised learning techniques such as Graph-Regularized Semi-Supervised SVM (Graph-SSVM), Autoencoder-Based Semi-Supervised Classifier (AE-SSC), and Label Propagation with SVM (LP-SVM), showing its superior performance in terms of both efficiency and accuracy.