Pub Date : 2021-12-29DOI: 10.1109/ICSPIS54653.2021.9729345
Maryam Alizadeh, S. Tabibian
Autism spectrum disorder is one kind of brain developmental disorders. The easiest way to diagnose persons with autism is done through speech processing techniques. However, limited researches have been done in this field. The reason may be due to the lack of valid and suitable datasets in this field. Therefore, in this paper, while analyzing the existing datasets in this field, the process of designing, collecting and evaluating a Persian speaker-independent dataset to diagnose children with autism (PersionSIChASD dataset) using speech processing methods has been discussed. Data collection has been done under the supervision of an autism specialist. The dataset includes those phonetic units that children with autism have difficulty in saying them, correctly. The results of evaluating the proposed dataset have shown speech recognition accuracies equal to 76% and 12% for phonetic units articulated by typical and autism infected children, respectively. The significant difference between the mentioned recognition rates (about 64%) could be exploited to diagnose autism infected children.
{"title":"A Persian speaker-independent dataset to diagnose autism infected children based on speech processing techniques","authors":"Maryam Alizadeh, S. Tabibian","doi":"10.1109/ICSPIS54653.2021.9729345","DOIUrl":"https://doi.org/10.1109/ICSPIS54653.2021.9729345","url":null,"abstract":"Autism spectrum disorder is one kind of brain developmental disorders. The easiest way to diagnose persons with autism is done through speech processing techniques. However, limited researches have been done in this field. The reason may be due to the lack of valid and suitable datasets in this field. Therefore, in this paper, while analyzing the existing datasets in this field, the process of designing, collecting and evaluating a Persian speaker-independent dataset to diagnose children with autism (PersionSIChASD dataset) using speech processing methods has been discussed. Data collection has been done under the supervision of an autism specialist. The dataset includes those phonetic units that children with autism have difficulty in saying them, correctly. The results of evaluating the proposed dataset have shown speech recognition accuracies equal to 76% and 12% for phonetic units articulated by typical and autism infected children, respectively. The significant difference between the mentioned recognition rates (about 64%) could be exploited to diagnose autism infected children.","PeriodicalId":286966,"journal":{"name":"2021 7th International Conference on Signal Processing and Intelligent Systems (ICSPIS)","volume":"509 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121803657","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-29DOI: 10.1109/ICSPIS54653.2021.9729333
Hanieh Malekijou, Vesal Hakami
We consider an energy-harvesting IoT device transmitting delay- and jitter-sensitive data over a wireless fading channel. Given the limited harvested energy, our goal is to compute optimal transmission control policies that decide on how many packets of data should be transmitted from the buffer's head-of-line at each discrete timeslot such that a long-run criterion involving the average delay/jitter is either minimized or never exceeds a pre-specified threshold. We utilize a suite of Q-learning-based techniques (from the reinforcement learning theory) to optimize the transmission policy in a model-free fashion. Compared to prior work, our novelty lies in proposing a model-free learning algorithm that enables jitter-aware transmissions by penalizing control decisions with the variance of the delay cost function. Extensive numerical results are presented for performance evaluation.
{"title":"Model-Free Learning Algorithms for Dynamic Transmission Control in IoT Equipment","authors":"Hanieh Malekijou, Vesal Hakami","doi":"10.1109/ICSPIS54653.2021.9729333","DOIUrl":"https://doi.org/10.1109/ICSPIS54653.2021.9729333","url":null,"abstract":"We consider an energy-harvesting IoT device transmitting delay- and jitter-sensitive data over a wireless fading channel. Given the limited harvested energy, our goal is to compute optimal transmission control policies that decide on how many packets of data should be transmitted from the buffer's head-of-line at each discrete timeslot such that a long-run criterion involving the average delay/jitter is either minimized or never exceeds a pre-specified threshold. We utilize a suite of Q-learning-based techniques (from the reinforcement learning theory) to optimize the transmission policy in a model-free fashion. Compared to prior work, our novelty lies in proposing a model-free learning algorithm that enables jitter-aware transmissions by penalizing control decisions with the variance of the delay cost function. Extensive numerical results are presented for performance evaluation.","PeriodicalId":286966,"journal":{"name":"2021 7th International Conference on Signal Processing and Intelligent Systems (ICSPIS)","volume":"181 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116971931","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-29DOI: 10.1109/ICSPIS54653.2021.9729365
Mousa Alizadeh, Sadegh E Mousavi, M. Beheshti, A. Ostadi
In terms of network topology, one of the extensively utilized technologies is the intrusion detection system (IDS). Despite applying numerous machine learning approaches (supervised and unsupervised) to enhance efficacy, reaching high-grade performance is still a challenging problem for existing intrusion detection algorithms. This study presents a new technique for IDS that focuses on various deep neural networks (DNNs) and their combination for data classification. The proposed model consists of three parts: (1) the feature selection is composed of an intersection of mutual information based on the transductive model (MIT-MIT), Anova F-value, and Genetic Algorithm (GA) methods, (2) the second section is a classifier network using a hybrid CNN-LSTM algorithm, and (3) the hyperparameter optimization module that puts to use Firefly, BAT, and Gray Wolf algorithms. In order to validate and verify the suggested model via accuracy, F1 score, recall, and precision criteria, a benchmark dataset, namely, NSL-KDD, is employed, which compares the proposed method with the highly developed classifiers. The comparison outcomes confirmed the surpassing of the presented strategy over contrast algorithms.
{"title":"Combination of Feature Selection and Hybrid Classifier as to Network Intrusion Detection System Adopting FA, GWO, and BAT Optimizers","authors":"Mousa Alizadeh, Sadegh E Mousavi, M. Beheshti, A. Ostadi","doi":"10.1109/ICSPIS54653.2021.9729365","DOIUrl":"https://doi.org/10.1109/ICSPIS54653.2021.9729365","url":null,"abstract":"In terms of network topology, one of the extensively utilized technologies is the intrusion detection system (IDS). Despite applying numerous machine learning approaches (supervised and unsupervised) to enhance efficacy, reaching high-grade performance is still a challenging problem for existing intrusion detection algorithms. This study presents a new technique for IDS that focuses on various deep neural networks (DNNs) and their combination for data classification. The proposed model consists of three parts: (1) the feature selection is composed of an intersection of mutual information based on the transductive model (MIT-MIT), Anova F-value, and Genetic Algorithm (GA) methods, (2) the second section is a classifier network using a hybrid CNN-LSTM algorithm, and (3) the hyperparameter optimization module that puts to use Firefly, BAT, and Gray Wolf algorithms. In order to validate and verify the suggested model via accuracy, F1 score, recall, and precision criteria, a benchmark dataset, namely, NSL-KDD, is employed, which compares the proposed method with the highly developed classifiers. The comparison outcomes confirmed the surpassing of the presented strategy over contrast algorithms.","PeriodicalId":286966,"journal":{"name":"2021 7th International Conference on Signal Processing and Intelligent Systems (ICSPIS)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117112556","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-29DOI: 10.1109/ICSPIS54653.2021.9729362
S. S. Fard, A. Amirkhani, M. Mosavi
In recent years, object detection with two-stage methods is one of the highest accuracies, like faster R-CNN. One-stage methods which use a typical dense sampling of likely item situations may be speedier and more straightforward. However, it has not exceeded the two-stage detectors' accuracy. This study utilizes a Retina network with a backbone ResNet50 block with multi-head self-attention (MHSA) to enhance one-stage method issues, especially small objects. RetinaNet is an efficient and accurate network and uses a new loss function. We swapped c5 in the ResNet50 block with MHSA, while we also used the features of the Retina network. Furthermore, compared to the ResNet50 block, it contains fewer parameters. The results of our study on the Pascal VOC 2007 dataset revealed that the number 81.86 % mAP was obtained, indicating that our technique may achieve promising performance compared to several current two-stage approaches.
{"title":"RetinaMHSA: Improving in single-stage detector with self-attention","authors":"S. S. Fard, A. Amirkhani, M. Mosavi","doi":"10.1109/ICSPIS54653.2021.9729362","DOIUrl":"https://doi.org/10.1109/ICSPIS54653.2021.9729362","url":null,"abstract":"In recent years, object detection with two-stage methods is one of the highest accuracies, like faster R-CNN. One-stage methods which use a typical dense sampling of likely item situations may be speedier and more straightforward. However, it has not exceeded the two-stage detectors' accuracy. This study utilizes a Retina network with a backbone ResNet50 block with multi-head self-attention (MHSA) to enhance one-stage method issues, especially small objects. RetinaNet is an efficient and accurate network and uses a new loss function. We swapped c5 in the ResNet50 block with MHSA, while we also used the features of the Retina network. Furthermore, compared to the ResNet50 block, it contains fewer parameters. The results of our study on the Pascal VOC 2007 dataset revealed that the number 81.86 % mAP was obtained, indicating that our technique may achieve promising performance compared to several current two-stage approaches.","PeriodicalId":286966,"journal":{"name":"2021 7th International Conference on Signal Processing and Intelligent Systems (ICSPIS)","volume":"54 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128654654","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Most of the learning based single image super-resolution networks employ intensity loss which measures pixel-wise difference between the estimated high resolution image and the ground truth. Since image components are different with respect to their saliency for HVS, it is desired to weight their impact on the loss functions accordingly. In this paper, a simple perceptual loss function is introduced based on the JPEG compression algorithm. In fact, the two compared images are transformed into DCT domain and then divided by the weighted quantization matrix. The difference between the resultant DCT coefficients shows the most effective components for HVS and can be considered as a perceptual loss function. The experimental results illustrate that employing the proposed loss promotes the convergence speed, and also, provides better outputs in terms of qualitative and quantitative measures.
{"title":"Perceptually Optimized Loss Function for Image Super-Resolution","authors":"Amirhossein Arezoomand, Pooryaa Cheraaqee, Azadeh Mansouri","doi":"10.1109/ICSPIS54653.2021.9729334","DOIUrl":"https://doi.org/10.1109/ICSPIS54653.2021.9729334","url":null,"abstract":"Most of the learning based single image super-resolution networks employ intensity loss which measures pixel-wise difference between the estimated high resolution image and the ground truth. Since image components are different with respect to their saliency for HVS, it is desired to weight their impact on the loss functions accordingly. In this paper, a simple perceptual loss function is introduced based on the JPEG compression algorithm. In fact, the two compared images are transformed into DCT domain and then divided by the weighted quantization matrix. The difference between the resultant DCT coefficients shows the most effective components for HVS and can be considered as a perceptual loss function. The experimental results illustrate that employing the proposed loss promotes the convergence speed, and also, provides better outputs in terms of qualitative and quantitative measures.","PeriodicalId":286966,"journal":{"name":"2021 7th International Conference on Signal Processing and Intelligent Systems (ICSPIS)","volume":"327 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124612972","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-29DOI: 10.1109/ICSPIS54653.2021.9729389
Mohammed Wadi, Wisam Elmasry, A. Shobole, Mehmet Rida Tur, R. Bayindir, Hossein Shahinzadeh
This paper presents a comprehensive empirical study of five different distribution functions to analysis the wind energy potential, namely, Rayleigh, Gamma, Extreme Value, Logistic, and T Location-Scale. In addition, three metaheuristics optimization methods, Grey Wolf Optimization, Marine Predators Algorithm, and Multi-Verse Optimizer are utilized to determine the optimal parameter values of each distribution. To test the accuracy of the introduced distributions and optimization methods, five error measures are investigated and compared such as mean absolute error, root mean square error, regression coefficient, correlation coefficient, and net fitness. To conduct this analysis, the Catalca site in the Marmara region in Istanbul, Republic of Turkey is selected to be the case study. The experimental results confirm that all introduced distributions based on optimization methods are efficient to model wind speed distribution in the selected site. Rayleigh distribution achieved the best matching while Extreme Value distribution provided the worst matching. Finally, many valuable observations drawn from this study are also discussed. MATLAB 2020b and Excel 365 were used to perform this study.
{"title":"Wind Energy Potential Approximation with Various Metaheuristic Optimization Techniques Deployment","authors":"Mohammed Wadi, Wisam Elmasry, A. Shobole, Mehmet Rida Tur, R. Bayindir, Hossein Shahinzadeh","doi":"10.1109/ICSPIS54653.2021.9729389","DOIUrl":"https://doi.org/10.1109/ICSPIS54653.2021.9729389","url":null,"abstract":"This paper presents a comprehensive empirical study of five different distribution functions to analysis the wind energy potential, namely, Rayleigh, Gamma, Extreme Value, Logistic, and T Location-Scale. In addition, three metaheuristics optimization methods, Grey Wolf Optimization, Marine Predators Algorithm, and Multi-Verse Optimizer are utilized to determine the optimal parameter values of each distribution. To test the accuracy of the introduced distributions and optimization methods, five error measures are investigated and compared such as mean absolute error, root mean square error, regression coefficient, correlation coefficient, and net fitness. To conduct this analysis, the Catalca site in the Marmara region in Istanbul, Republic of Turkey is selected to be the case study. The experimental results confirm that all introduced distributions based on optimization methods are efficient to model wind speed distribution in the selected site. Rayleigh distribution achieved the best matching while Extreme Value distribution provided the worst matching. Finally, many valuable observations drawn from this study are also discussed. MATLAB 2020b and Excel 365 were used to perform this study.","PeriodicalId":286966,"journal":{"name":"2021 7th International Conference on Signal Processing and Intelligent Systems (ICSPIS)","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114810915","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-29DOI: 10.1109/ICSPIS54653.2021.9729393
Shaghayegh Shahiri Tabarestani, A. Aghagolzadeh, M. Ezoji
Using computer-aided diagnosis systems for helping radiologists and reducing the time of diagnosis is vital. In this paper, Faster-RCNN with three different backbone structures for feature extraction is applied for fracture zone prediction on bone X-rays of the MURA database. We used just three subsets of all seven subsets of the database. These subsets contain X-rays from the humerus, elbow, and forearm. The results of the experiments show that Faster-RCNN with Inception-ResNet-Version-2 as the feature extractor has the best performance. AP of this model on test samples in the best condition of parameters setting reaches 66.82 % for IOU=50%.
{"title":"Bone Fracture Detection and Localization on MURA Database Using Faster-RCNN","authors":"Shaghayegh Shahiri Tabarestani, A. Aghagolzadeh, M. Ezoji","doi":"10.1109/ICSPIS54653.2021.9729393","DOIUrl":"https://doi.org/10.1109/ICSPIS54653.2021.9729393","url":null,"abstract":"Using computer-aided diagnosis systems for helping radiologists and reducing the time of diagnosis is vital. In this paper, Faster-RCNN with three different backbone structures for feature extraction is applied for fracture zone prediction on bone X-rays of the MURA database. We used just three subsets of all seven subsets of the database. These subsets contain X-rays from the humerus, elbow, and forearm. The results of the experiments show that Faster-RCNN with Inception-ResNet-Version-2 as the feature extractor has the best performance. AP of this model on test samples in the best condition of parameters setting reaches 66.82 % for IOU=50%.","PeriodicalId":286966,"journal":{"name":"2021 7th International Conference on Signal Processing and Intelligent Systems (ICSPIS)","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132446693","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-29DOI: 10.1109/ICSPIS54653.2021.9729335
Marzieh Sheikhi, Vesal Hakami
In the new generation networks, the freshness of the data plays a prominent role in real-time systems. The novel metric of the age of information (AoI) measures the elapsed time since the generation of the latest received data. This paper considers a real-time scenario where a source node samples and forwards the measurements to a monitoring center over a millimeter-wave (mmWave) channel. The source node is also equipped with a finite rechargeable battery to harvest energy from the environment. We propose a remote monitoring problem that considers the tradeoff between the minimization of long-term average AoI and the energy usage of the source node. We formulate the problem as an MDP model, and as a model-free reinforcement learning approach, we utilize the Q-learning algorithm to obtain the optimal policy that minimizes the long-term average AoI. Our evaluations investigate the convergence property as well as the impact of changing the problem parameters on the average AoI and average energy consumption. Simulation results show that compared to two other baselines (i.e., random and greedy (myopic) policy), the proposed Q-Learning based algorithm is able to keep the data fresh and consumes less energy by considering the possible future system states.
{"title":"AoI-Aware Status Update Control for an Energy Harvesting Source over an Uplink mmWave Channel","authors":"Marzieh Sheikhi, Vesal Hakami","doi":"10.1109/ICSPIS54653.2021.9729335","DOIUrl":"https://doi.org/10.1109/ICSPIS54653.2021.9729335","url":null,"abstract":"In the new generation networks, the freshness of the data plays a prominent role in real-time systems. The novel metric of the age of information (AoI) measures the elapsed time since the generation of the latest received data. This paper considers a real-time scenario where a source node samples and forwards the measurements to a monitoring center over a millimeter-wave (mmWave) channel. The source node is also equipped with a finite rechargeable battery to harvest energy from the environment. We propose a remote monitoring problem that considers the tradeoff between the minimization of long-term average AoI and the energy usage of the source node. We formulate the problem as an MDP model, and as a model-free reinforcement learning approach, we utilize the Q-learning algorithm to obtain the optimal policy that minimizes the long-term average AoI. Our evaluations investigate the convergence property as well as the impact of changing the problem parameters on the average AoI and average energy consumption. Simulation results show that compared to two other baselines (i.e., random and greedy (myopic) policy), the proposed Q-Learning based algorithm is able to keep the data fresh and consumes less energy by considering the possible future system states.","PeriodicalId":286966,"journal":{"name":"2021 7th International Conference on Signal Processing and Intelligent Systems (ICSPIS)","volume":"58 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128876214","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-29DOI: 10.1109/ICSPIS54653.2021.9729386
Hossein Shahinzadeh, Arezou Mahmoudi, Jalal Moradi, H. Nafisi, E. Kabalci, Mohamed Benbouzid
Security in smart grids has been investigated by many scholars so far. Among the existing security issues, False Data Injection (FDI) attacks in energy, computers, and communication domains are still an ongoing challenge. These attacks have the ability to sabotage the grid through causing misfunctioning of measurements devices as well as changing the state estimation appraisal so that these changes, known as false data, cannot be easily recognized and identified using conventional approaches. In this paper, the degree of network resilience against FDI attacks is analyzed by simulating a randomly generated sample FDI attack, in which the false data vector has different intensity and different quantity. A steady-state AC power flow in accordance with the outage model is employed to simulate and predict the power system response after the incidence of an FDI attack, and the ability of this attack for blackout and shutting down the transmission network has been investigated. In the proposed model, the transmission line outage, load shedding, as well as voltage instability metrics are tested and analyzed on the IEEE 300- bus test network. Given that FDI attacks are considered a serious threat to power systems, the preliminary results imply that the targeted electricity grid is resilient against these attacks in terms of the probability of outage and chain blackouts, but the transient voltage stability can be affected.
{"title":"Anomaly Detection and Resilience-Oriented Countermeasures against Cyberattacks in Smart Grids","authors":"Hossein Shahinzadeh, Arezou Mahmoudi, Jalal Moradi, H. Nafisi, E. Kabalci, Mohamed Benbouzid","doi":"10.1109/ICSPIS54653.2021.9729386","DOIUrl":"https://doi.org/10.1109/ICSPIS54653.2021.9729386","url":null,"abstract":"Security in smart grids has been investigated by many scholars so far. Among the existing security issues, False Data Injection (FDI) attacks in energy, computers, and communication domains are still an ongoing challenge. These attacks have the ability to sabotage the grid through causing misfunctioning of measurements devices as well as changing the state estimation appraisal so that these changes, known as false data, cannot be easily recognized and identified using conventional approaches. In this paper, the degree of network resilience against FDI attacks is analyzed by simulating a randomly generated sample FDI attack, in which the false data vector has different intensity and different quantity. A steady-state AC power flow in accordance with the outage model is employed to simulate and predict the power system response after the incidence of an FDI attack, and the ability of this attack for blackout and shutting down the transmission network has been investigated. In the proposed model, the transmission line outage, load shedding, as well as voltage instability metrics are tested and analyzed on the IEEE 300- bus test network. Given that FDI attacks are considered a serious threat to power systems, the preliminary results imply that the targeted electricity grid is resilient against these attacks in terms of the probability of outage and chain blackouts, but the transient voltage stability can be affected.","PeriodicalId":286966,"journal":{"name":"2021 7th International Conference on Signal Processing and Intelligent Systems (ICSPIS)","volume":"198 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116371918","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-29DOI: 10.1109/ICSPIS54653.2021.9729379
Kosar Bakhshi, B. Bahrak, H. Mahini
Advances in technology and the human tendency to use virtual services are constantly increasing in all areas of life. Online ride-hailing services are not an exception to this rule. Due to the financial transactions in these systems, the possibility of fraud by profiteers also increases which can affect the revenue of such services significantly. In this paper, we propose a system that can detect fraud in online ride-hailing systems. We address frauds that occur using the ride collusion method or creating a fake ride using GPS spoofing applications. We have used real unlabeled data from one of the largest ride-hailing companies in Iran for this purpose. Our system first identifies the most important features that help us distinguish real rides from fake rides, then it uses unsupervised learning methods to detect ride anomalies. After identifying the anomalies and examining these rides, we label the data, and use supervised learning methods to construct the fraud detection model.
{"title":"Fraud Detection System in Online Ride-Hailing Services","authors":"Kosar Bakhshi, B. Bahrak, H. Mahini","doi":"10.1109/ICSPIS54653.2021.9729379","DOIUrl":"https://doi.org/10.1109/ICSPIS54653.2021.9729379","url":null,"abstract":"Advances in technology and the human tendency to use virtual services are constantly increasing in all areas of life. Online ride-hailing services are not an exception to this rule. Due to the financial transactions in these systems, the possibility of fraud by profiteers also increases which can affect the revenue of such services significantly. In this paper, we propose a system that can detect fraud in online ride-hailing systems. We address frauds that occur using the ride collusion method or creating a fake ride using GPS spoofing applications. We have used real unlabeled data from one of the largest ride-hailing companies in Iran for this purpose. Our system first identifies the most important features that help us distinguish real rides from fake rides, then it uses unsupervised learning methods to detect ride anomalies. After identifying the anomalies and examining these rides, we label the data, and use supervised learning methods to construct the fraud detection model.","PeriodicalId":286966,"journal":{"name":"2021 7th International Conference on Signal Processing and Intelligent Systems (ICSPIS)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127034022","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：