Pattern Analysis and Intelligent Systems
shahrzad Oveisi
Articles in Press, Accepted Manuscript, Available Online from 02 June 2022
Abstract
Nowadays, we are witnessing the growth of financial theft cases and thereby a high degree of financial losses concomitant with the growth of IoT technology, as well as development and utilization of this technology in banking and financial fields along with the increase in the volume of transactions. ...
Read More
Nowadays, we are witnessing the growth of financial theft cases and thereby a high degree of financial losses concomitant with the growth of IoT technology, as well as development and utilization of this technology in banking and financial fields along with the increase in the volume of transactions. Financial fraud detection represents the challenge of finding anomalies in networks of financial transactions. In general, anomaly detection is the problem of distinguishing between normal data samples and well-defined patterns or signatures with those that do not conform to the expected profiles. Various methods have been introduced to identify, detect and prevent such thefts. This paper presents a LSTM based approach for detecting fraudulent transactions. I express the fraud detection problem as a sequence classification task and employ LSTM based method to incorporate transaction sequences. Then, the results are assessed using two classifier methods, namely random forest and decision tree and Mean Square Error. In the second case, because the data are imbalanced, undersampling and oversampling techniques have been used, after which I employ LSTM neural network and results assessed with MSE. Finally, the results of the two groups were compared with confusion matrix. According to the evaluations, my proposed method with Random forest classifier gives the best results.
Pattern Analysis and Intelligent Systems
narges jafari; Farhad Soleimanian Gharehchopogh
Volume 6, Issue 3 , August 2020, , Pages 119-132
Abstract
Metaheuristic algorithms are used to solve NP-hard optimization problems. These algorithms have two main components, i.e. exploration and exploitation, and try to strike a balance between exploration and exploitation to achieve the best possible near-optimal solution. The bat algorithm is one of the ...
Read More
Metaheuristic algorithms are used to solve NP-hard optimization problems. These algorithms have two main components, i.e. exploration and exploitation, and try to strike a balance between exploration and exploitation to achieve the best possible near-optimal solution. The bat algorithm is one of the metaheuristic algorithms with poor exploration and exploitation. In this paper, exploration and exploitation processes of Gray Wolf Optimizer (GWO) algorithm are applied to some of the solutions produced by the bat algorithm. Therefore, part of the population of the bat algorithm is changed by two processes (i.e. exploration and exploitation) of GWO; the new population enters the bat algorithm population when its result is better than that of the exploitation and exploration operators of the bat algorithm. Thereby, better new solutions are introduced into the bat algorithm at each step. In this paper, 20 mathematic benchmark functions are used to evaluate and compare the proposed method. The simulation results show that the proposed method outperforms the bat algorithm and other metaheuristic algorithms in most implementations and has a high performance.
Pattern Analysis and Intelligent Systems
Reham Mohammed
Volume 5, Issue 1 , February 2019, , Pages 1-10
Abstract
Quadrotor control has been noted for its trouble as the consequence of the high maneuverability, system nonlinearity and strongly coupled multivariable. This paper deals with the simulation depend on proposed controller of a quadrotor that can overcome this trouble. The mathematical model of quadrotor ...
Read More
Quadrotor control has been noted for its trouble as the consequence of the high maneuverability, system nonlinearity and strongly coupled multivariable. This paper deals with the simulation depend on proposed controller of a quadrotor that can overcome this trouble. The mathematical model of quadrotor is determined using a Newton-Euler formulation. Fractional Order Proportional Integral Derivative (FOPID) controller tuned by genetic algorithm (GA) is investigated to control and stabilization the position and attitude of quadrotor using feedback linearization. This controller is used as a reference to compare its results with Proportional Integral Derivative (PID) controller tuned by GA. The control structure performance is evaluated through the response and minimizing the error of the position and attitude. Simulation results, demonstrates that position and attitude control using FOPID has fast response, better steady state error and RMS error than PID. By simulation the two controllers are tested under the same conditions using SIMULINK under MATLAB2015a.
Pattern Analysis and Intelligent Systems
Sahar Rahmatian; Reza Safabakhsh
Volume 1, Issue 2 , May 2015, , Pages 15-22
Abstract
Multiple people detection and tracking is a challenging task in real-world crowded scenes. In this paper, we have presented an online multiple people tracking-by-detection approach with a single camera. We have detected objects with deformable part models and a visual background extractor. In the tracking ...
Read More
Multiple people detection and tracking is a challenging task in real-world crowded scenes. In this paper, we have presented an online multiple people tracking-by-detection approach with a single camera. We have detected objects with deformable part models and a visual background extractor. In the tracking phase we have used a combination of support vector machine (SVM) person-specific classifiers, similarity scores, the Hungarian algorithm and inter-object occlusion handling. Detections have been used for training person-specific classifiers and to help guide the trackers by computing a similarity score based on them and spatial information and assigning them to the trackers with the Hungarian algorithm. To handle inter-object occlusion we have used explicit occlusion reasoning. The proposed method does not require prior training and does not impose any constraints on environmental conditions. Our evaluation showed that the proposed method outperformed the state of the art approaches by 10% and 15% or achieved comparable performance.