Gorkem Kar
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Gorkem Kar has a total of 46 co-authors. Here are the top co-authors by number of shared publications:
Marco Gruteser
Rutgers University
5 shared publicationsFan Bai
General Motors Research
3 shared publicationsRamesh Govindan
University of Southern California
3 shared publications- Cagatay Catal
Department of Computer Science and Engineering, Qatar University, Doha, Qatar
2 shared publications - S. Sridevi
Centre for Neuroinformatics, School of Computer Science and Engineering, Vellore Institute of Technology 2 , Chennai
2 shared publications Shubham Jain
Rutgers University
2 shared publications
Latest Publications
- 1 March 2026
- Published by American Vacuum Society in AVS Quantum Science
- Vol. 8 (1)
A classical quantum neural network with dual-channel feature extraction on IonQ QPU for classifying weathered microplastics from Raman spectra
Microplastics are a serious threat to human health and the environment, mainly because of their persistence and entry into the food chain through marine life. The weathering of these particles, caused by prolonged exposure to UV radiation and heat, accelerates chemical changes that can harm ecosystems and human health and make them hard to identify. Therefore, it is important to accurately identify and classify microplastics to assess their impact and creating efficient mitigation measures. This study includes quantum machine learning together with traditional deep learning methods, such as autoencoders, and statistical approaches like principal component analysis and linear discriminant analysis, to classify Raman spectra of different microplastic samples. The model proposed classifies microplastics into weathered and standard samples with an accuracy of 94%, distinguishing between distinct types of plastics, such as polyvinyl chloride, polyethylene terephthalate, polypropylene, polyethylene, and polyamide at 98.5% accuracy. In order to further improve this research forward, the model was run on live quantum computing hardware using Amazon Braket services. The end-to-end pipeline was run on both Braket's SV1 state vector simulator and the IonQ Aria quantum hardware, utilizing the compute resources of AWS's quantum environment. The model recorded a performance level of 97%–98% in the focused tasks, establishing the applicability of embedding quantum computing into practical applications. By improving the precision of microplastic identification, this study adds to the better understanding of their environmental impact and aids wider efforts toward reducing related ecological risks.
- 1 March 2026
- Published by American Vacuum Society in AVS Quantum Science
- Vol. 8 (1)
A classical quantum neural network with dual-channel feature extraction on IonQ QPU for classifying weathered microplastics from Raman spectra
Microplastics are a serious threat to human health and the environment, mainly because of their persistence and entry into the food chain through marine life. The weathering of these particles, caused by prolonged exposure to UV radiation and heat, accelerates chemical changes that can harm ecosystems and human health and make them hard to identify. Therefore, it is important to accurately identify and classify microplastics to assess their impact and creating efficient mitigation measures. This study includes quantum machine learning together with traditional deep learning methods, such as autoencoders, and statistical approaches like principal component analysis and linear discriminant analysis, to classify Raman spectra of different microplastic samples. The model proposed classifies microplastics into weathered and standard samples with an accuracy of 94%, distinguishing between distinct types of plastics, such as polyvinyl chloride, polyethylene terephthalate, polypropylene, polyethylene, and polyamide at 98.5% accuracy. In order to further improve this research forward, the model was run on live quantum computing hardware using Amazon Braket services. The end-to-end pipeline was run on both Braket's SV1 state vector simulator and the IonQ Aria quantum hardware, utilizing the compute resources of AWS's quantum environment. The model recorded a performance level of 97%–98% in the focused tasks, establishing the applicability of embedding quantum computing into practical applications. By improving the precision of microplastic identification, this study adds to the better understanding of their environmental impact and aids wider efforts toward reducing related ecological risks.
- Open Access
- 16 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 104382-104405
Novel Quanvolutional Neural Network-Based COVID-19 Diagnosis From Raman Spectroscopic Signals of Human Serum
COVID-19 has proven to be one of the most devastating pandemics of the 21st century, causing widespread mortality and overwhelming healthcare systems globally. Early and accurate diagnosis is crucial to managing and controlling the spread of this virus. Although the Polymerase Chain Reaction (PCR) technique is effective for identifying the genetic material of the pathogens causing SARS-CoV-2 COVID-19, its longer processing time has led to the emergence of of Raman spectroscopy as a promising alternative, a non-destructive method that provides detailed molecular insights for detecting COVID-19 through human serum samples. The primary goal of this research is to accurately diagnose COVID-19 by analyzing Raman spectroscopic signals of serum samples with a novel quantum circuit-based Quanvolutional Neural Network (QNN). In this research, although the signals are initially in one dimension (1D), we applied wavelet transforms to convert Raman spectroscopic signals into 2D scaleogram images to capture intricate spectral patterns for efficient classification. We compared the performance of our proposed QNN with existing QNN variant and classical convolutional neural network (CNN) models such as Vanilla CNN, MobileNet CNN, Xception CNN, and VGG-16 CNN. We were surprised to achieve outstanding performance across all seven evaluation metrics namely Jaccard index(JI) of 97.73%, accuracy of 98.92%, F1-score of 98.85%, recall of 97.73%, Matthews correlation coefficient (MCC) of 97.86%, precision of 100% and Area Under the Curve (AUC) score of 98.86%. The results demonstrate that proposed novel QNN outperforms classical models without over-fitting. Our findings highlight the potential of QNN in revolutionizing diagnostic tools for critical diseases like COVID-19.
- 16 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 104382-104405
Open AccessNovel Quanvolutional Neural Network-Based COVID-19 Diagnosis From Raman Spectroscopic Signals of Human Serum
COVID-19 has proven to be one of the most devastating pandemics of the 21st century, causing widespread mortality and overwhelming healthcare systems globally. Early and accurate diagnosis is crucial to managing and controlling the spread of this virus. Although the Polymerase Chain Reaction (PCR) technique is effective for identifying the genetic material of the pathogens causing SARS-CoV-2 COVID-19, its longer processing time has led to the emergence of of Raman spectroscopy as a promising alternative, a non-destructive method that provides detailed molecular insights for detecting COVID-19 through human serum samples. The primary goal of this research is to accurately diagnose COVID-19 by analyzing Raman spectroscopic signals of serum samples with a novel quantum circuit-based Quanvolutional Neural Network (QNN). In this research, although the signals are initially in one dimension (1D), we applied wavelet transforms to convert Raman spectroscopic signals into 2D scaleogram images to capture intricate spectral patterns for efficient classification. We compared the performance of our proposed QNN with existing QNN variant and classical convolutional neural network (CNN) models such as Vanilla CNN, MobileNet CNN, Xception CNN, and VGG-16 CNN. We were surprised to achieve outstanding performance across all seven evaluation metrics namely Jaccard index(JI) of 97.73%, accuracy of 98.92%, F1-score of 98.85%, recall of 97.73%, Matthews correlation coefficient (MCC) of 97.86%, precision of 100% and Area Under the Curve (AUC) score of 98.86%. The results demonstrate that proposed novel QNN outperforms classical models without over-fitting. Our findings highlight the potential of QNN in revolutionizing diagnostic tools for critical diseases like COVID-19.
- Open Access
- 9 January 2024
- Published by MDPI AG in Electronics
- Vol. 13 (2) , 293
Automated Network Incident Identification through Genetic Algorithm-Driven Feature Selection
The cybersecurity landscape presents daunting challenges, particularly in the face of Denial of Service (DoS) attacks such as DoS Http Unbearable Load King (HULK) attacks and DoS GoldenEye attacks. These malicious tactics are designed to disrupt critical services by overwhelming web servers with malicious requests. In contrast to DoS attacks, there exists nefarious Operating System (OS) scanning, which exploits vulnerabilities in target systems. To provide further context, it is essential to clarify that NMAP, a widely utilized tool for identifying host OSes and vulnerabilities, is not inherently malicious but a dual-use tool with legitimate applications, such as asset inventory services in company networks. Additionally, Domain Name System (DNS) botnets can be incredibly damaging as they harness numerous compromised devices to inundate a target with malicious DNS traffic. This can disrupt online services, leading to downtime, financial losses, and reputational damage. Furthermore, DNS botnets can be used for other malicious activities like data exfiltration, spreading malware, or launching other cyberattacks, making them a versatile tool for cybercriminals. As attackers continually adapt and modify specific attributes to evade detection, our paper introduces an automated detection method that requires no expert input. This innovative approach identifies the distinct characteristics of DNS botnet attacks, DoS HULK attacks, DoS GoldenEye attacks, and OS-Scanning, explicitly using the NMAP tool, even when attackers alter their tactics. By harnessing a representative dataset, our proposed method ensures robust detection of such attacks against varying attack parameters or behavioral shifts. This heightened resilience significantly raises the bar for attackers attempting to conceal their malicious activities. Significantly, our approach delivered outstanding outcomes, with a mid 95% accuracy in categorizing NMAP OS scanning and DNS botnet attacks, and 100% for DoS HULK attacks and DoS GoldenEye attacks, proficiently discerning between malevolent and harmless network packets. Our code and the dataset are made publicly available.
- 9 January 2024
- Published by MDPI AG in Electronics
- Vol. 13 (2) , 293
Open AccessAutomated Network Incident Identification through Genetic Algorithm-Driven Feature Selection
The cybersecurity landscape presents daunting challenges, particularly in the face of Denial of Service (DoS) attacks such as DoS Http Unbearable Load King (HULK) attacks and DoS GoldenEye attacks. These malicious tactics are designed to disrupt critical services by overwhelming web servers with malicious requests. In contrast to DoS attacks, there exists nefarious Operating System (OS) scanning, which exploits vulnerabilities in target systems. To provide further context, it is essential to clarify that NMAP, a widely utilized tool for identifying host OSes and vulnerabilities, is not inherently malicious but a dual-use tool with legitimate applications, such as asset inventory services in company networks. Additionally, Domain Name System (DNS) botnets can be incredibly damaging as they harness numerous compromised devices to inundate a target with malicious DNS traffic. This can disrupt online services, leading to downtime, financial losses, and reputational damage. Furthermore, DNS botnets can be used for other malicious activities like data exfiltration, spreading malware, or launching other cyberattacks, making them a versatile tool for cybercriminals. As attackers continually adapt and modify specific attributes to evade detection, our paper introduces an automated detection method that requires no expert input. This innovative approach identifies the distinct characteristics of DNS botnet attacks, DoS HULK attacks, DoS GoldenEye attacks, and OS-Scanning, explicitly using the NMAP tool, even when attackers alter their tactics. By harnessing a representative dataset, our proposed method ensures robust detection of such attacks against varying attack parameters or behavioral shifts. This heightened resilience significantly raises the bar for attackers attempting to conceal their malicious activities. Significantly, our approach delivered outstanding outcomes, with a mid 95% accuracy in categorizing NMAP OS scanning and DNS botnet attacks, and 100% for DoS HULK attacks and DoS GoldenEye attacks, proficiently discerning between malevolent and harmless network packets. Our code and the dataset are made publicly available.
- Top Cited PapersOpen Access
- 2 February 2023
- Published by Springer Nature in Journal of Ambient Intelligence and Humanized Computing
- Vol. 14 (3) , 2887-2913
Machine learning in crime prediction
Predicting crimes before they occur can save lives and losses of property. With the help of machine learning, many researchers have studied predicting crimes extensively. In this paper, we evaluate state-of-the-art crime prediction techniques that are available in the last decade, discuss possible challenges, and provide a discussion about the future work that could be conducted in the field of crime prediction. Although many works aim to predict crimes, the datasets they used and methods that are applied are numerous. Using a Systematic Literature Review (SLR) methodology, we aim to collect and synthesize the required knowledge regarding machine learning-based crime prediction and help both law enforcement authorities and scientists to mitigate and prevent future crime occurrences. We focus primarily on 68 selected machine learning papers that predict crime. We formulate eight research questions and observe that the majority of the papers used a supervised machine learning approach, assuming that there is prior labeled data, and however in some cases, there is no labeled data in real-world scenarios. We have also discussed the main challenges found while conducting some of the studies by the researchers. We consider that this research paves the way for further research to help governments and countries fight crime and decrease this for better safety and security.
- 2 February 2023
- Published by Springer Nature in Journal of Ambient Intelligence and Humanized Computing
- Vol. 14 (3) , 2887-2913
Top Cited PapersOpen AccessMachine learning in crime prediction
Predicting crimes before they occur can save lives and losses of property. With the help of machine learning, many researchers have studied predicting crimes extensively. In this paper, we evaluate state-of-the-art crime prediction techniques that are available in the last decade, discuss possible challenges, and provide a discussion about the future work that could be conducted in the field of crime prediction. Although many works aim to predict crimes, the datasets they used and methods that are applied are numerous. Using a Systematic Literature Review (SLR) methodology, we aim to collect and synthesize the required knowledge regarding machine learning-based crime prediction and help both law enforcement authorities and scientists to mitigate and prevent future crime occurrences. We focus primarily on 68 selected machine learning papers that predict crime. We formulate eight research questions and observe that the majority of the papers used a supervised machine learning approach, assuming that there is prior labeled data, and however in some cases, there is no labeled data in real-world scenarios. We have also discussed the main challenges found while conducting some of the studies by the researchers. We consider that this research paves the way for further research to help governments and countries fight crime and decrease this for better safety and security.
- 30 June 2022
- Published by Muhendislik Bilimleri ve Tasarim Dergisi in Mühendislik Bilimleri ve Tasarım Dergisi
- Vol. 10 (2) , 504-519
TÜRKÇE DİLİNDE YAZILAN BİLİMSEL METİNLERİN DERİN ÖĞRENME TEKNİĞİ UYGULANARAK ÇOKLU SINIFLANDIRILMASI
Ekim 2018 yılında Google tarafından geliştirilen BERT derin öğrenme tekniği, makine öğrenimi ve doğal dil işleme dünyasında çok popüler oldu. Transformatörlerin Çift Yönlü Kodlayıcı Gösterimleri anlamına gelen BERT, yapay zeka ve makine öğrenimi teknolojilerini bir arada kullanan bir doğal dil işleme tekniği olarak açıklanabilir. Günümüzde, gözetimli öğrenme metodolojisinin bir parçası olan sınıflandırma problemleriyle çokça karşılaşılmaktadır. Sınıflandırmanın temeli eğitilen bir makinenin yeni gelen bir veri hakkında tahminleme yapabilmesine ve sınıflandırabilmesine dayanır. Buradaki amaç bir veri kümesi üzerinde tanımlı olan sınıflar arasında veriyi dağıtabilmektir. Türkçe'nin morfolojisinin zengin ama karmaşık olması, sondan eklemeli bir dil olması ve dil bilgisinden kaynaklanan zorluklar çoklu sınıflandırma problemlerinin çözümünde başlıca sorun teşkil etmekte iken BERT derin öğrenme tekniği ile bu sorun daha kolay çözülebilir hale gelmiştir. Bu çalışmada, son 10 yıl içinde Türkçe dili ile yazılmış akademik araştırma ve bilimsel çalışmalar veri seti olarak kullanıldı. Çoklu sınıflandırma problemlerinde kullanmak üzere, veri setine BERT derin öğrenme tekniği uygulanarak önceden eğitilmiş Türkçe bir BERT modeli üzerinde ince ayar (fine-tuning) yapıldı. Deneylerin sonucunda, eğitilmiş olan sistemin doğruluğu %96 başarım oranına sahip olmuştur.
- 30 June 2022
- Published by Muhendislik Bilimleri ve Tasarim Dergisi in Mühendislik Bilimleri ve Tasarım Dergisi
- Vol. 10 (2) , 504-519
TÜRKÇE DİLİNDE YAZILAN BİLİMSEL METİNLERİN DERİN ÖĞRENME TEKNİĞİ UYGULANARAK ÇOKLU SINIFLANDIRILMASI
Ekim 2018 yılında Google tarafından geliştirilen BERT derin öğrenme tekniği, makine öğrenimi ve doğal dil işleme dünyasında çok popüler oldu. Transformatörlerin Çift Yönlü Kodlayıcı Gösterimleri anlamına gelen BERT, yapay zeka ve makine öğrenimi teknolojilerini bir arada kullanan bir doğal dil işleme tekniği olarak açıklanabilir. Günümüzde, gözetimli öğrenme metodolojisinin bir parçası olan sınıflandırma problemleriyle çokça karşılaşılmaktadır. Sınıflandırmanın temeli eğitilen bir makinenin yeni gelen bir veri hakkında tahminleme yapabilmesine ve sınıflandırabilmesine dayanır. Buradaki amaç bir veri kümesi üzerinde tanımlı olan sınıflar arasında veriyi dağıtabilmektir. Türkçe'nin morfolojisinin zengin ama karmaşık olması, sondan eklemeli bir dil olması ve dil bilgisinden kaynaklanan zorluklar çoklu sınıflandırma problemlerinin çözümünde başlıca sorun teşkil etmekte iken BERT derin öğrenme tekniği ile bu sorun daha kolay çözülebilir hale gelmiştir. Bu çalışmada, son 10 yıl içinde Türkçe dili ile yazılmış akademik araştırma ve bilimsel çalışmalar veri seti olarak kullanıldı. Çoklu sınıflandırma problemlerinde kullanmak üzere, veri setine BERT derin öğrenme tekniği uygulanarak önceden eğitilmiş Türkçe bir BERT modeli üzerinde ince ayar (fine-tuning) yapıldı. Deneylerin sonucunda, eğitilmiş olan sistemin doğruluğu %96 başarım oranına sahip olmuştur.
- Top Cited PapersOpen Access
Hybrid Blockchain Platforms for the Internet of Things (IoT): A Systematic Literature Review
In recent years, research into blockchain technology and the Internet of Things (IoT) has grown rapidly due to an increase in media coverage. Many different blockchain applications and platforms have been developed for different purposes, such as food safety monitoring, cryptocurrency exchange, and secure medical data sharing. However, blockchain platforms cannot store all the generated data. Therefore, they are supported with data warehouses, which in turn is called a hybrid blockchain platform. While several systems have been developed based on this idea, a current state-of-the-art systematic overview on the use of hybrid blockchain platforms is lacking. Therefore, a systematic literature review (SLR) study has been carried out by us to investigate the motivations for adopting them, the domains at which they were used, the adopted technologies that made this integration effective, and, finally, the challenges and possible solutions. This study shows that security, transparency, and efficiency are the top three motivations for adopting these platforms. The energy, agriculture, health, construction, manufacturing, and supply chain domains are the top domains. The most adopted technologies are cloud computing, fog computing, telecommunications, and edge computing. While there are several benefits of using hybrid blockchains, there are also several challenges reported in this study.
Top Cited PapersOpen AccessHybrid Blockchain Platforms for the Internet of Things (IoT): A Systematic Literature Review
In recent years, research into blockchain technology and the Internet of Things (IoT) has grown rapidly due to an increase in media coverage. Many different blockchain applications and platforms have been developed for different purposes, such as food safety monitoring, cryptocurrency exchange, and secure medical data sharing. However, blockchain platforms cannot store all the generated data. Therefore, they are supported with data warehouses, which in turn is called a hybrid blockchain platform. While several systems have been developed based on this idea, a current state-of-the-art systematic overview on the use of hybrid blockchain platforms is lacking. Therefore, a systematic literature review (SLR) study has been carried out by us to investigate the motivations for adopting them, the domains at which they were used, the adopted technologies that made this integration effective, and, finally, the challenges and possible solutions. This study shows that security, transparency, and efficiency are the top three motivations for adopting these platforms. The energy, agriculture, health, construction, manufacturing, and supply chain domains are the top domains. The most adopted technologies are cloud computing, fog computing, telecommunications, and edge computing. While there are several benefits of using hybrid blockchains, there are also several challenges reported in this study.
- 1 November 2020
- Published by Elsevier in Applied Soft Computing
- Vol. 96
Distracted driver detection by combining in-vehicle and image data using deep learning
- 1 November 2020
- Published by Elsevier in Applied Soft Computing
- Vol. 96
Distracted driver detection by combining in-vehicle and image data using deep learning
- 1 April 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Scotto: Real-Time Driver Behavior Scoring Using In-Vehicle Data
This paper explores the minimal dataset to be collected from vehicles, to efficiently score drivers, using common vehicle sensors. This can lead to important results for insurance companies, advertisements and personalization. Existing work relies on several sensor information that are collected over a drive including acceleration/deceleration patterns or average trip duration. Vehicular companies make vehicular sensor information available to many external services. To explore how to score driving behaviors from such a data, we consider a system that interfaces to vehicle bus and executes supervised learning methods on this data. To facilitate this analysis, we collect invehicle data from 20 drivers on a test route and have less than %10 error in our scoring algorithm.
- 1 April 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Scotto: Real-Time Driver Behavior Scoring Using In-Vehicle Data
This paper explores the minimal dataset to be collected from vehicles, to efficiently score drivers, using common vehicle sensors. This can lead to important results for insurance companies, advertisements and personalization. Existing work relies on several sensor information that are collected over a drive including acceleration/deceleration patterns or average trip duration. Vehicular companies make vehicular sensor information available to many external services. To explore how to score driving behaviors from such a data, we consider a system that interfaces to vehicle bus and executes supervised learning methods on this data. To facilitate this analysis, we collect invehicle data from 20 drivers on a test route and have less than %10 error in our scoring algorithm.
- Open Access
- 9 November 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 67 (3) , 1909-1922
Towards Robust Vehicular Context Sensing
In-vehicle context sensing can detect many aspects of driver behavior and the environment, such as drivers changing lanes, stop signs, road grade and potholes, and these features can be used to improve driver safety and comfort, and engine efficiency. In general, detecting these features can use either on-board sensors on the vehicle (car sensors), or sensors built into mobile devices (phone sensors) carried by one or more occupants, or both. Furthermore, traces of sensor read- ings from different cars, when crowd-sourced, can provide increased spatial coverage as well as disambiguation. In this paper, we explore, by designing novel detection algorithms for the three different features discussed above, three related questions: How is the accuracy of detection related to the choice of phone vs. car sensors? To what extent, and in what ways, does crowd-sourcing contribute to detec- tion accuracy? How is accuracy affected by phone position? We have collected hundreds of miles of vehicle traces with annotated groundtruth, and demonstrated through evaluation that our detection algorithms can achieve high accuracy for each task (e.g. > 90% for lane change determinations) and that crowd-sensing plays an indispensable role in improving the detection performance (e.g. improving recall by 35% for lane change determinations on curves). Our results can give car manufacturers insight into how to augment their internal sensing capabilities with phone sensors, or give mobile app developers insight into what car sensors to use in order to complement mobile device sensing capabilities.
- 9 November 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 67 (3) , 1909-1922
Open AccessTowards Robust Vehicular Context Sensing
In-vehicle context sensing can detect many aspects of driver behavior and the environment, such as drivers changing lanes, stop signs, road grade and potholes, and these features can be used to improve driver safety and comfort, and engine efficiency. In general, detecting these features can use either on-board sensors on the vehicle (car sensors), or sensors built into mobile devices (phone sensors) carried by one or more occupants, or both. Furthermore, traces of sensor read- ings from different cars, when crowd-sourced, can provide increased spatial coverage as well as disambiguation. In this paper, we explore, by designing novel detection algorithms for the three different features discussed above, three related questions: How is the accuracy of detection related to the choice of phone vs. car sensors? To what extent, and in what ways, does crowd-sourcing contribute to detec- tion accuracy? How is accuracy affected by phone position? We have collected hundreds of miles of vehicle traces with annotated groundtruth, and demonstrated through evaluation that our detection algorithms can achieve high accuracy for each task (e.g. > 90% for lane change determinations) and that crowd-sensing plays an indispensable role in improving the detection performance (e.g. improving recall by 35% for lane change determinations on curves). Our results can give car manufacturers insight into how to augment their internal sensing capabilities with phone sensors, or give mobile app developers insight into what car sensors to use in order to complement mobile device sensing capabilities.