http://dspace.dtu.ac.in:8080/jspui/handle/123456789/26 2026-04-28T04:03:40Z http://dspace.dtu.ac.in:8080/jspui/handle/repository/22695 Title: OPTIMIZING INVESTMENT PORTFOLIOS IN BANKING USING INTEGER PROGRAMMING TECHNIQUES Authors: AGGARWAL, MANAN; ANAND, NAMAN; DAS, L.N. (SUPERVISOR) Abstract: In the modern banking system, optimizing the allocation of capital across various loan aspects is a critical task that directly impacts profitability and risk management. Traditional portfolio optimization methods often rely on linear programming techniques that assume continuous investment decisions. However, real-world banking constraints—such as regulatory limits, discrete investment units, and risk thresholds—demand more realistic and implementable models. This thesis explores the application of Integer Programming (IP) techniques to optimize investment portfolios in the banking domain. The primary objective is to maximize net profit from three major loan categories—Home Loans, Personal Loans, and Business Loans—while adhering to operational constraints such as investment caps, borrower creditworthiness, and diversification rules. The dataset used in this study was manually created to simulate realistic banking scenarios, including data on expected profits, borrower creditworthiness, and administrative costs. A Mixed Integer Linear Programming (MILP) model was formulated to reflect these constraints, with investment decisions modelled in discrete ₹1 lakh units. The results indicated an optimal allocation of ₹50 lakhs each to Home and Personal Loans, yielding a maximum net profit of ₹8.50 lakhs. Business loans, though offering a competitive return, were excluded from the final allocation due to relatively lower risk-adjusted performance and constraint tightness. Graphical visualizations were used to interpret allocation patterns and profit contributions, while sensitivity analysis highlighted the binding nature of budget and diversification constraints. The study demonstrates that Integer Programming not only improves the practical feasibility of financial decisions but also allows banks to manage risk while achieving profitability. The model’s structure provides a robust foundation for extending into multi-period investment strategies, stochastic interest rate environments, or incorporating credit scoring models in future work. 2025-06-01T00:00:00Z http://dspace.dtu.ac.in:8080/jspui/handle/repository/22646 Title: RECENT APPLICATIONS OF PCA AND SVD Authors: AHUJA, ADITI Abstract: In this thesis, Principal Component Analysis is a powerful dimensionality reduction technique that transforms high-dimensional data into a lower-dimensional space while preserving variance. By comput- ing the covariance matrix and its eigenvectors, PCA finds principal components that best represent the data. It is used on large scale in image compression, face recognition, and feature extraction, simplifying complex datasets without losing critical information. SVD is a matrix factorization method that decom- poses any matrix into three distinct matrices: A = U ΣV T . This decomposition reveals hidden patterns in data and has applications in data compression, noise reduction, and recommendation systems. Unlike PCA, which relies on eigenvectors of the covariance matrix, SVD works directly on the data matrix, making it more versatile. PCA and SVD are two fundamental techniques in linear algebra that have revolutionized data science, machine learning, and image processing. This presentation explores their mathematical foundations, geometric interpretations, and real-world applications. 2025-12-01T00:00:00Z http://dspace.dtu.ac.in:8080/jspui/handle/repository/22640 Title: IMAGE SEGMENTATION Authors: GUPTA, DIKSHA; DHINGRA, SUSHANT Abstract: Fuzzy C-Means (FCM) is one of the popular techniques used for segmenting sci- entific images. It is suggested in the literature to use intuitionistic fuzzy c-means (IFCM), which is based on the notion of intuitionistic fuzzy sets (IFSs), to handle the ambiguity and uncertainty related to real data.The hesitation and member- ship degrees are used to determine the objective function.However, FCM is used to achieve the approximate answer rather than analytically computing the objective function. Even though there are numerous variations of intuitionistic fuzzy set the- ory, all of them struggle with the issue of noise in images during the segmentation process.In order to address this issue, we have proposed using a picture fuzzy set theoretic approach, which improves the data’s ability to be represented and aids in handling the noise structures present in the image. In our proposed work, the picture fuzzy Euclidean distance is swapped out for the Manhattan distance (City Block Distance), as Manhattan distance produces significantly better noise suppres- sion.The method was applied to a fake image that had been ”Gaussian” and ”salt and pepper” distorted.Partition efficiency, average segmentation accuracy (ASA), and dice score (DS) were the performance metrics used. We can utilize the distance measure and dissimilarity between fuzzy sets to calculate the difference between two fuzzy sets or intuitionistic sets as it can be used for pattern recognition, and image segmentation.Results show that the proposed method gives the better result. 2023-05-01T00:00:00Z http://dspace.dtu.ac.in:8080/jspui/handle/repository/22281 Title: KWG-RFE: A TRI-STAGE HYBRID FEATURE REDUCTION FRAMEWORK FOR ANDROID MALWARE DETECTION Authors: DWIVEDI, KUNIKAA Abstract: Mobile computing has been transformed by the quick development and broad use of An- droid smartphones, but this has also given cybercriminals a larger attack surface. Among these, mobile malware poses the greatest threat to data, privacy, and system integrity for both individuals and companies. As a result, reliable, effective, and precise malware detection techniques that are exclusive to the Android ecosystem are desperately needed. This thesis introduces KWG-RFE, a hybrid feature selection strategy that offers a fresh approach to Android malware detection. This strategy combines three complementing techniques: Recursive Feature Elimination (RFE), Graph-based feature analysis, and the Kruskal-Wallis statistical test. These elements function in concert to filter and rank fea- tures according to their significance and effect on classification performance. The proposed method was validated using a large dataset of over 111,000 Android appli- cations, which included both malicious and benign samples. Each program had a num- ber of components removed, including hardware-related parts, intent filters, permissions, and API calls. In order to reduce dimensionality while maintaining crucial information pertinent to malware identification, these features were subsequently put through the KWG-RFE selection procedure. Both the full and reduced feature sets were used to train and assess a number of ma- chine learning classifiers, such as Random Forests, Decision Trees, and Support Vector Machines. With a full feature set of 97.75% and a competitive 94.50% accuracy after ap- plying the KWG-RFE feature reduction, the Random Forest classifier showed the highest detection accuracy among them. The findings show that, without compromising classi- fication performance, the suggested hybrid feature selection approach is quite successful at removing superfluous and unnecessary characteristics. All things considered, this study shows how effective it is to combine algorithmic, struc- tural, and statistical feature selection methods when it comes to Android malware de- tection. The suggested KWG-RFE approach is a useful addition to the field of mobile cybersecurity since it maintains a high level of accuracy while increasing detection effi- ciency by lowering computational overhead. 2025-05-01T00:00:00Z