Financial fraud is a growing problem that poses a significant threat to the banking industry, the government sector, and the public. In response, financial institutions must continuously improve their fraud detection systems. Although preventative and security precautions are implemented to reduce financial fraud, criminals are constantly adapting and devising new ways to evade fraud prevention systems. The classification of transactions as legitimate or fraudulent poses a significant challenge for existing classification models due to highly imbalanced datasets. This research aims to develop rules to detect fraud transactions that do not involve any resampling technique. The effectiveness of the rule-based model (RBM) is assessed using a variety of metrics such as accuracy, specificity, precision, recall, confusion matrix, Matthew’s correlation coefficient (MCC), and receiver operating characteristic (ROC) values. The proposed rule-based model is compared to several existing machine learning models such as random forest (RF), decision tree (DT), multi-layer perceptron (MLP), k-nearest neighbor (KNN), naive Bayes (NB), and logistic regression (LR) using two benchmark datasets. The results of the experiment show that the proposed rule-based model beat the other methods, reaching accuracy and precision of 0.99 and 0.99, respectively.

Quantum Cryptography emerged from the limitations of classical cryptography. It will play a vital role in information security after the availability of expected powerful quantum computers. Still many quantum primitives like quantum money, blind quantum computation, quantum copy protection, etc. are theoretical as they require a completely functional quantum computer for their implementation. But one prominent quantum cryptographic primitive, the Quantum Key Distribution (QKD) is possible with current technology. The QKD is a key establishment system having several stages namely raw key generation, key sifting, key reconciliation, and privacy amplification. In this paper, an efficient key sifting scheme has been developed. Successful simulation has shown that the proposed modified key sifting scheme requires less time to build the sifted key compared to the sifted key in conventional BB84 protocol in most cases. This paper also represents Tree Parity Machine (TPM) based key reconciliation analysis using different learning algorithms such as Hebbian, Anti-Hebbian, and Random-Walk. This reconciliation analysis helps to choose the optimum learning algorithm for Artificial Neural Network (ANN) based key reconciliation in future Quantum Key Distribution systems.

With the increasing number of vehicles, the usage of technology has also been increased in the transportation system. Although automobile companies are using advanced technologies to develop high performing transports, traffic safety still remains to be a concerning issue. Drivers' driving behavior is considered as one of the key factors of the traffic safety, which could be monitored from their individual driving maneuvers. In this paper, we present a supervised learning model and a semi-supervised transfer learning model for the classification of driving maneuvers from the sensor fusion time series data. The semi-supervised model consists of an unsupervised long-short term memory (LSTM) autoencoder and a supervised LSTM classifier. The supervised model consists of a supervised LSTM model. Because of using LSTM, both of the models can analyze time-series data. In the semi-supervised model, the LSTM encoder learns from unlabeled data as a compressed low dimensional feature vector, which then transfers the learning to the supervised LSTM classifier to classify the driving maneuvers. With the proposed models, we use domain specific knowledge data of the driving environment, such as data changing rules of various driving maneuvers as well as the temporal features over time. We use class functions for seven driving maneuver types and convert those into binary feature vector to use with the LSTM models. We present a comparative analysis of the per class accuracy of the proposed semi-supervised and supervised models with and without using domain-specific knowledge, where the models with the domain specific knowledge outperform. Our proposed semi-supervised and supervised models are compared with the other existing approaches, where our models trained with the domain specific knowledge provided better performance. We also compared the per class accuracy for both the supervised and semi-supervised models, where all the maneuver class accuracy for supervised model was above 98% and semi-supervised model was above 95%. Although the supervised model outperforms the semi-supervised model, the semi-supervised model would be more beneficial in applications where the labeled driving maneuvers data are hard to capture or insufficient.