I am a PhD student at Florida International University. My current research focus is on detecting anomaly at system level by leveraging Artificial Intelligence and deep learning techniques. It is currently funded by Department of Defense. I also work as a Teaching Assistant on machine learning class. Supervisors: Advisor
In the current world, the applications of anomaly detection range from fraud detection to diagnos... more In the current world, the applications of anomaly detection range from fraud detection to diagnosis in the medical area. Most of the current methodologies are applicable only when a particular dataset pertains to certain assumptions and a distinct domain. Such assumptions require prior knowledge of the dataset. The training development cycle time to find the best single model is time-consuming and challenging. Unsupervised anomaly detection methods do not use the target label for training. However, they result in high false positive rates. In this paper, we address the problem of the ensemble anomaly detection approach that generalizes well across multiple domains. We design a multi- level hybrid approach. At the first level, we train several weak classifiers (weak one class classifiers). Next, we utilize deep learning-based AutoEncoder to reduce the dimension of the dataset. These are the two sets of hybrid features. Next, different one-class classifiers have their strength and lim...
Machine Learning for Societal Improvement, Modernization, and Progress
Cybersecurity attacks are rising both in rate and complexity over time. More development and cons... more Cybersecurity attacks are rising both in rate and complexity over time. More development and constant improvement in defensive approaches are needed to secure the operational systems against such attacks. Several malicious attacks pose severe security threats to organizations and users in today's internet age. It is vital to train enhanced malware classification systems to capture the variation in the malware type that belongs to the same family type. In this chapter, the author addresses the malware detection issue using a learning-based approach. First, the author explains various machine learning and deep learning algorithms to solve the problem. Next, the author provides practical implementation by proposing a deep learning-based framework on the open-source benchmark dataset on API calls. The dataset contains API calls during normal and malware-infected processes. The proposed framework trains a hybrid model of convolution neural network followed by long short-term memory t...
Computer Science & Engineering: An International Journal
Information spread through online social media or sites has increased drastically with the swift ... more Information spread through online social media or sites has increased drastically with the swift growth of the Internet. Unverified or fake news reaches numerous users without concern about the trustworthiness of the info. Such fake news is created for political or commercial interests to mislead the users. In current society, the spread of misinformation is a big challenge. Hence, we propose a deep learning-based Long Short Term Memory (LSTM) classifier for fake news classification. Textual content is the primary unit in the fake news scenario. Therefore, natural language processing-based feature extraction is used to generate language-driven features. Experimental results show that NLP-based featured extraction with LSTM model achieves a higher accuracy rate in discernible less time.
International Journal of Computer Science and Information Security (IJSCIS), 2020
The utility companies are heavily investing in the production of renewable energy to meet the cur... more The utility companies are heavily investing in the production of renewable energy to meet the current demand for energy consumption by the commercial industries and the households. Renewable energy is derived from the earth's natural sources. The most common renewable energy resources are Solar, Biogas, Geothermal, Hydroelectric, Wind, Solar Thermal, etc. Conventional energy sources are making a negative impact on the environment and the economy; hence it's a need of an hour to have a blend of sources of energy with a majority percentage of sustainable energy sources. Renewable energy is an alternative to conventional energy, and it tends to be much less harmful to the environment. The prediction of energy generation from sustainable sources is required to gauge the amount of energy produced in the future to meets higher energy consumption needs. In this paper, we provide an approach for predicting the geothermal energy production using a LSTM (Long Short Term-Memory) Sequence to Sequence Encoder-Decoder Neural Network architecture. Keywords— Renewable Energy, Geothermal, Prediction, Deep Learning, Recurrent neural networks, Long Short-Term Memory.
2019 International Conference on Computational Science and Computational Intelligence (CSCI), 2019
With the advent of technology, sophisticated malware presents a significant threat to computer se... more With the advent of technology, sophisticated malware presents a significant threat to computer security. In this work, we propose anomaly detection techniques that learn three different behaviors of windows system-call sequences. We apply Long-Short-Term-Memory (LSTM) for temporal behavior, Cosine Similarity for frequency distribution behavior, and Jaccard Similarity for commonality behavior. The proposed framework monitors the processes in a hypervisor-based environment to detect compromised virtual machines. System call sequences of normal processes and malware-infected processes were extracted with memory forensic techniques. Our proposed anomaly detection techniques were able to learn the above three behavior of the system call sequences with 99% accuracy.
Online social networks (OSN) are one of the favorite places where people share posts like their p... more Online social networks (OSN) are one of the favorite places where people share posts like their photos, videos, and text to gain popularity. On the other hand, the marketing industry tries to gain the popularity of their advertisement using such OSNs. Popularity of a particular post depends on the number of likes received by that post. To increase one’s social worth, people try to use this market by artificially increasing the likes on their posts. There is a lack of research in the current literature on Instagram which is one of the growing OSNs. Our work focuses on detecting valid and fake like of posts with the application of learning model taking into consideration several popular factors. We developed an automated learning model to detect fake liking behavior on the Instagram post. The learned model can accurately differentiate between the legitimate and fake liker with an accuracy of 97% with ensemble-based learning model and also autoencoder is used to detect bots activity.
With recent advances in computer and sensor technologies in the last few decades, the use of robo... more With recent advances in computer and sensor technologies in the last few decades, the use of robots for various applications has increased enormously. The reliability of robots depends on the minimization of component failures and downtime. To improve the reliability, periodic monitoring of components and their behavior are essential to inference component fatigue and potential breakdowns. Since fully autonomous robots are very expensive, telepresence robots are affordable for mass scale deployment and can be controlled by a trained human operator like avatars. To increase the efficiency and to reduce the downtime of telepresence robot service, it is essential to observe the various commands performed on the robot and to analyze the samples of component status over a long period. We propose an efficient data driven model with a collection of frequent time-stamped data from various components of a telepresence robot and predict potential failure warnings. The collected historical dat...
Social networks have become very popular in recent years due to the exponential growth of the per... more Social networks have become very popular in recent years due to the exponential growth of the personalized computing devices. Twitter, Facebook, and LinkedIn are the evidence of massive online social networks. Product reviews through these social media are now widely used by individuals and organizations for their decision making. The trustworthiness of these reviews is unreliable when businesses generate fraudulent positive reviews for themselves or negative reviews for their competitors. For reviews to reflect genuine user experiences and opinions, such spam reviews need to be detected. Prior works focused on fake reviews and individual fake reviewers. However, to target a particular product, fake reviewers work collaboratively in groups and control the sentiment of the product. Even more challenging is there might be one real person creating multiple fake IDs to write several reviews for the same product. This paper focuses on finding such group of fake reviewers. The proposed me...
Phishing websites are fraudulent sites that impersonate a trusted party to gain access to sensiti... more Phishing websites are fraudulent sites that impersonate a trusted party to gain access to sensitive information of an individual person or organization. Traditionally, phishing website detection is done through the usage of blacklist databases. However, due to the current, rapid development of global networking and communication technologies, there are numerous websites and it has become difficult to classify based on traditional methods since new websites are created every second. In this paper, we are proposing a real-time, anti-phishing system. In the first step, we extract the lexical and host-based properties of a website. In the second step, we combine URL (Uniform Resource Locator) features, NLP and host-based properties to train the machine learning and deep learning models. Our detection model is able to detect phishing URLs with a detection rate of 94.89%.
Host based intrusion detection systems monitor operations for significant deviations from normal ... more Host based intrusion detection systems monitor operations for significant deviations from normal and healthy behavior. Anomalies are patterns in data that do not conform to the expected normal behavior. System call analysis has been conclusively established as the best method to reveal details about the program behavior. Therefore, attackers create new exploits that makes major impact at the system call level. In this research, we developed an enhanced and optimized deep learning LSTM (Long Short Term Memory) network, for anomaly detection, trained on sequences of system calls. Our model detects any anomalous behavior in the system calls with 80% accuracy.
Social networks have become very popular in recent years due to the exponential growth of the per... more Social networks have become very popular in recent years due to the exponential growth of the personalized computing devices. Twitter, Facebook, and LinkedIn are the evidence of massive online social networks. Product reviews through these social media are now widely used by individuals and organizations for their decision making. The trustworthiness of these reviews is unreliable when businesses generate fraudulent positive reviews for themselves or negative reviews for their competitors. For reviews to reflect genuine user experiences and opinions, such spam reviews need to be detected. Prior works focused on fake reviews and individual fake reviewers. However, to target a particular product, fake reviewers work collaboratively in groups and control the sentiment of the product. Even more challenging is there might be one real person creating multiple fake IDs to write several reviews for the same product. This paper focuses on finding such group of fake reviewers. The proposed method first uses a deep-walk approach to find behavioral representations for each fake reviewer account. Subsequently, we use several supervised and unsupervised machine learning algorithms to identify the groups of fake reviewers.
In the current world, the applications of anomaly detection range from fraud detection to diagnos... more In the current world, the applications of anomaly detection range from fraud detection to diagnosis in the medical area. Most of the current methodologies are applicable only when a particular dataset pertains to certain assumptions and a distinct domain. Such assumptions require prior knowledge of the dataset. The training development cycle time to find the best single model is time-consuming and challenging. Unsupervised anomaly detection methods do not use the target label for training. However, they result in high false positive rates. In this paper, we address the problem of the ensemble anomaly detection approach that generalizes well across multiple domains. We design a multi- level hybrid approach. At the first level, we train several weak classifiers (weak one class classifiers). Next, we utilize deep learning-based AutoEncoder to reduce the dimension of the dataset. These are the two sets of hybrid features. Next, different one-class classifiers have their strength and lim...
Machine Learning for Societal Improvement, Modernization, and Progress
Cybersecurity attacks are rising both in rate and complexity over time. More development and cons... more Cybersecurity attacks are rising both in rate and complexity over time. More development and constant improvement in defensive approaches are needed to secure the operational systems against such attacks. Several malicious attacks pose severe security threats to organizations and users in today's internet age. It is vital to train enhanced malware classification systems to capture the variation in the malware type that belongs to the same family type. In this chapter, the author addresses the malware detection issue using a learning-based approach. First, the author explains various machine learning and deep learning algorithms to solve the problem. Next, the author provides practical implementation by proposing a deep learning-based framework on the open-source benchmark dataset on API calls. The dataset contains API calls during normal and malware-infected processes. The proposed framework trains a hybrid model of convolution neural network followed by long short-term memory t...
Computer Science & Engineering: An International Journal
Information spread through online social media or sites has increased drastically with the swift ... more Information spread through online social media or sites has increased drastically with the swift growth of the Internet. Unverified or fake news reaches numerous users without concern about the trustworthiness of the info. Such fake news is created for political or commercial interests to mislead the users. In current society, the spread of misinformation is a big challenge. Hence, we propose a deep learning-based Long Short Term Memory (LSTM) classifier for fake news classification. Textual content is the primary unit in the fake news scenario. Therefore, natural language processing-based feature extraction is used to generate language-driven features. Experimental results show that NLP-based featured extraction with LSTM model achieves a higher accuracy rate in discernible less time.
International Journal of Computer Science and Information Security (IJSCIS), 2020
The utility companies are heavily investing in the production of renewable energy to meet the cur... more The utility companies are heavily investing in the production of renewable energy to meet the current demand for energy consumption by the commercial industries and the households. Renewable energy is derived from the earth's natural sources. The most common renewable energy resources are Solar, Biogas, Geothermal, Hydroelectric, Wind, Solar Thermal, etc. Conventional energy sources are making a negative impact on the environment and the economy; hence it's a need of an hour to have a blend of sources of energy with a majority percentage of sustainable energy sources. Renewable energy is an alternative to conventional energy, and it tends to be much less harmful to the environment. The prediction of energy generation from sustainable sources is required to gauge the amount of energy produced in the future to meets higher energy consumption needs. In this paper, we provide an approach for predicting the geothermal energy production using a LSTM (Long Short Term-Memory) Sequence to Sequence Encoder-Decoder Neural Network architecture. Keywords— Renewable Energy, Geothermal, Prediction, Deep Learning, Recurrent neural networks, Long Short-Term Memory.
2019 International Conference on Computational Science and Computational Intelligence (CSCI), 2019
With the advent of technology, sophisticated malware presents a significant threat to computer se... more With the advent of technology, sophisticated malware presents a significant threat to computer security. In this work, we propose anomaly detection techniques that learn three different behaviors of windows system-call sequences. We apply Long-Short-Term-Memory (LSTM) for temporal behavior, Cosine Similarity for frequency distribution behavior, and Jaccard Similarity for commonality behavior. The proposed framework monitors the processes in a hypervisor-based environment to detect compromised virtual machines. System call sequences of normal processes and malware-infected processes were extracted with memory forensic techniques. Our proposed anomaly detection techniques were able to learn the above three behavior of the system call sequences with 99% accuracy.
Online social networks (OSN) are one of the favorite places where people share posts like their p... more Online social networks (OSN) are one of the favorite places where people share posts like their photos, videos, and text to gain popularity. On the other hand, the marketing industry tries to gain the popularity of their advertisement using such OSNs. Popularity of a particular post depends on the number of likes received by that post. To increase one’s social worth, people try to use this market by artificially increasing the likes on their posts. There is a lack of research in the current literature on Instagram which is one of the growing OSNs. Our work focuses on detecting valid and fake like of posts with the application of learning model taking into consideration several popular factors. We developed an automated learning model to detect fake liking behavior on the Instagram post. The learned model can accurately differentiate between the legitimate and fake liker with an accuracy of 97% with ensemble-based learning model and also autoencoder is used to detect bots activity.
With recent advances in computer and sensor technologies in the last few decades, the use of robo... more With recent advances in computer and sensor technologies in the last few decades, the use of robots for various applications has increased enormously. The reliability of robots depends on the minimization of component failures and downtime. To improve the reliability, periodic monitoring of components and their behavior are essential to inference component fatigue and potential breakdowns. Since fully autonomous robots are very expensive, telepresence robots are affordable for mass scale deployment and can be controlled by a trained human operator like avatars. To increase the efficiency and to reduce the downtime of telepresence robot service, it is essential to observe the various commands performed on the robot and to analyze the samples of component status over a long period. We propose an efficient data driven model with a collection of frequent time-stamped data from various components of a telepresence robot and predict potential failure warnings. The collected historical dat...
Social networks have become very popular in recent years due to the exponential growth of the per... more Social networks have become very popular in recent years due to the exponential growth of the personalized computing devices. Twitter, Facebook, and LinkedIn are the evidence of massive online social networks. Product reviews through these social media are now widely used by individuals and organizations for their decision making. The trustworthiness of these reviews is unreliable when businesses generate fraudulent positive reviews for themselves or negative reviews for their competitors. For reviews to reflect genuine user experiences and opinions, such spam reviews need to be detected. Prior works focused on fake reviews and individual fake reviewers. However, to target a particular product, fake reviewers work collaboratively in groups and control the sentiment of the product. Even more challenging is there might be one real person creating multiple fake IDs to write several reviews for the same product. This paper focuses on finding such group of fake reviewers. The proposed me...
Phishing websites are fraudulent sites that impersonate a trusted party to gain access to sensiti... more Phishing websites are fraudulent sites that impersonate a trusted party to gain access to sensitive information of an individual person or organization. Traditionally, phishing website detection is done through the usage of blacklist databases. However, due to the current, rapid development of global networking and communication technologies, there are numerous websites and it has become difficult to classify based on traditional methods since new websites are created every second. In this paper, we are proposing a real-time, anti-phishing system. In the first step, we extract the lexical and host-based properties of a website. In the second step, we combine URL (Uniform Resource Locator) features, NLP and host-based properties to train the machine learning and deep learning models. Our detection model is able to detect phishing URLs with a detection rate of 94.89%.
Host based intrusion detection systems monitor operations for significant deviations from normal ... more Host based intrusion detection systems monitor operations for significant deviations from normal and healthy behavior. Anomalies are patterns in data that do not conform to the expected normal behavior. System call analysis has been conclusively established as the best method to reveal details about the program behavior. Therefore, attackers create new exploits that makes major impact at the system call level. In this research, we developed an enhanced and optimized deep learning LSTM (Long Short Term Memory) network, for anomaly detection, trained on sequences of system calls. Our model detects any anomalous behavior in the system calls with 80% accuracy.
Social networks have become very popular in recent years due to the exponential growth of the per... more Social networks have become very popular in recent years due to the exponential growth of the personalized computing devices. Twitter, Facebook, and LinkedIn are the evidence of massive online social networks. Product reviews through these social media are now widely used by individuals and organizations for their decision making. The trustworthiness of these reviews is unreliable when businesses generate fraudulent positive reviews for themselves or negative reviews for their competitors. For reviews to reflect genuine user experiences and opinions, such spam reviews need to be detected. Prior works focused on fake reviews and individual fake reviewers. However, to target a particular product, fake reviewers work collaboratively in groups and control the sentiment of the product. Even more challenging is there might be one real person creating multiple fake IDs to write several reviews for the same product. This paper focuses on finding such group of fake reviewers. The proposed method first uses a deep-walk approach to find behavioral representations for each fake reviewer account. Subsequently, we use several supervised and unsupervised machine learning algorithms to identify the groups of fake reviewers.
Uploads