Part1 This is the research project. Research ProjectAssignment RubricDissertation TemplatePractical Connection Project (PPT was be posted under this link)Carefully examine the findings and recommendat

MACHINE LEARNING ENHANCEMENT OF PHYSICAL SECURITY 0

Part1 This is the research project. Research ProjectAssignment RubricDissertation TemplatePractical Connection Project (PPT was be posted under this link)Carefully examine the findings and recommendat 1

How Machine Learning Enhance Physical Security at Control Area Network Stations

Rajkumar Ashok-Kumar

Nidheesh Raju Baskaruni

Sudhir Reddy Muram

Ayyappa Kumar Sajja

Sai Nishanth Savrala

Rajinikar Biram

University of Cumberland’s

ISOL-634 Physical Security

Table of Content

Chapter one

Abstract ..................................................................................3

Introduction ............................................................................4

Background of the research ....................................................5

Purpose of the research ..........................................................6

Research Question..................................................................8

Research methodology.............................................................9

Literature review .....................................................................10

Results and discussion ........................................................... 12

Recommendation and conclusion ..........................................13

Definition of Terms………………………………………….14

References .............................................................................16

Abstract

The main objective of this research paper is to expound on the influences of the machine learning to physical security in various sectors in the society. The application of machine learning has diverse impacts on the security level in the society because of the incorporation of intelligent devices and algorithms helping in perfuming complex tasks. Physical security of the control area networks (CAN) are essential and require proper protection from misuse and machine learning assist in the promotion of that process. The machine learning operations can sometimes be the targets of criminals compromising their operations through the initiation of malicious attacks such as the introduction of stealthy algorithms stealing data from the machines. The introduction of the malicious stealthy data exploits and blind the genuine operating machines cheating their functions for the benefits of the criminals. However, mostly the application of the machine learning techniques assists in the mitigation of the infiltration of the criminals injecting their filthy programs into a proper functioning system (Lin, Yu, Yang, Xu, & Zhao, 2012). The intelligent algorithms assist in the extraction of relevant information from various physical data networks. The application also facilitates the automation of various tasks bringing the essential benefits based on new functionality, optimization of the resources and the personalization. The operations of machine learning successfully applied to various sectors, including computer and system security.

Keywords: Machine learning, Computer system, Artificial Intelligence, Sensors, Networks, Data, Information, security.

Introduction

The Professor gave us the freedom to select the topic that we were interested in, which triggered us to search online to find something related to physical security and which needed more further research. We all had different ideas, but our goal was simple, to have the latest technology involved, which is the main reason we went for Machine learning. Also, the fact that the reference paper that we ended up deciding had an enormous scope in continuing further research.

According to Moore (2018), the development of the interactive machine learning techniques has enhanced the security of the physical relationships of the control area networks (CAN). The author asserts that the machine learning techniques are in conjunction with physics models of turbulent flow to develop more sophisticated situations overcoming the model form uncertainty. Machine learning used in the development of the periodic traffic on the CAN bus for creation of the aspects of the combination of the co-occurrence of commands and the value of their specific data fields (Moore, 2018).

Moore further suggests that the protection of the networks is a challenging task because of the lack of publicly available translation in-vehicle network data to vehicle operation (Moore & Vann, 2019). The author further elucidates intrusion detection system (IDSs) of the controller area network (CANs) in the society citing their previous limitation to leveraging statistics aspects of the protocol standards (Lin, Yu, Yang, Xu, & Zhsaos, 2012). The machine learning refers to the changes within a system doing various tasks attributed to artificial intelligence (AI), and such operations include recognition, diagnosis, planning, prediction and the robotic controls (Moore & Vann, 2019).

Changes in the operation of the system enhance the functionality of the existing systems into a model of a new unique system with higher efficiency and functionality (Humphrey & Bello, 2012). The machine learning focuses on computer vision, natural language processing, hearing, and image processing and pattern recognition, including cognitive computing. Machine learning involved in physical security provides various machines with the abilities to collect data through senses similar to human senses (Pons & Serra, 2017).

Machine learning initiates the machine to learn without programming making then intelligent to perform complex tasks within the industries and organizations in the society (Mundhenk et al., 2015). There are various divisions of the machine learning techniques enhancing the application of the AI into the community, such as the supervised, unsupervised, and reinforcement learning techniques. The purpose of this research study is to explain how machine learning enhances physical security in the control area network stations.

Background of the research

Network control areas are sensitive because of meta-data they contain and the expensive, sophisticated equipment vulnerable to any threats (Liu et al., 2015). The physical security the network control areas need to enhance through the application of the intelligent devices capable of sensing any threats to keep the machines safe (Pattipati et al., 2018). The machine learning uses various techniques of operation such as face detection through a trained algorithm to detect different faces (Papernot, McDaniel, Sinha & Wellman, 2016). It also suitable in the credit card fraud detection performed using some concrete rules.

Speech recognition is another tactic of the application of machine learning where a machine is trained through the preset algorithms to identify individuals through their speech (Bakfan, Manos, Itay & Neulander, 2017). The application of the machine learning in the protection of the physical security of networks shared among the prominent organization such as Google, IBM, Intel, Apple and Twitter (Papernot, McDaniel, Sinha & Wellman, 2016). The current advancement of information technology and artificial intelligence increases the application of intelligent devices in most protected areas in society.

The machine learning assists in various applications in society, such as the control of the traffic light system. Machine learning combines with real-time demands to manipulate and change human lives regularly in the community (Staggs, 2013). It is used in most industries and the security areas to protect vital installations from the intruders who can interfere with the normal operations of the facilities. The next generation banking applies the use of machine learning to their services in different facilities to enhance the security within their physical security layers (Sinha et al., 2015). Healthcare also has incorporated the application of machine learning in its operations to intensify its services effectively.

Purpose of the research

The purpose of this research is to elaborate on how the machine learning technology enhances the physical security at control area network stations (Chhetri, 2016). It reflects the importance of machine learning in a combination of Artificial Intelligence (AI) in controlling of devices making intelligent to perform complex tasks without human interventions (Moore & Buckner, 2012). The control area network stations require high consolidation of security with restricted movement to protect the data and the equipment within the area (Kerekes et al., 2017). The amalgamation of the machine learning and the AI techniques elevates the physical security of such control network places through the application of sensible and intelligent devices to secure the area.

According to Jesus (2019), there is much emphasis on physical security, which chained towards civil society. The goal of this research is to elaborate on how machine learning technology enhances the physical security at control area network stations. It reflects the importance of machine learning in a combination of Artificial Intelligence (AI) in controlling of devices making intelligent to perform complex tasks without human interventions (Avatefiour, Hafeez, Tayyab & Malik, 2018). The control area network stations require high consolidation of security with restricted movement to protect the data and the equipment within the area (Tomlinson, Bryans & Shaikh, 2018). The amalgamation of the machine learning, and the AI techniques elevates the physical security of such control network places through the application of sensible and intelligent devices to secure the area.

Large datasets are the basis for efficient performances and improvement in the application of machine learning (Hussain, Hussain, Hassan & Hossain, 2019). The machine learning supported by other sectors in technology such as computer science, neural networks and deep learning for a better outcome. It also increases the level of security details involving physical security by using the latest techniques in smartphones, automobile, healthcare, which makes the physical security market lucrative through the adoption of Artificial Intelligence technologies.

Research question

To further help reduce the cost of security, we argue that hiring qualified and high-skilled cybersecurity personnel is essential to implement and customize opensource solutions to security threats. It is necessary to establish and obtain a full understanding of the services and network traffic that communicates in and out of the company's network (Savvas, 2012) and infrastructure, to be able to secure it appropriately. Currently, there is a gap that exists between SCADA and Cyber Security professionals. This dissertation also provided a simulated environment both virtualized and hardware based to help both better understand the mechanisms and protocols that operate the SCADA and ICS (Bruneau, 2010) industry as well as for SCADA and ICS professionals to understand the feasibility and process of a cyber-attack.

This dissertation's goal was to shorten that gap by providing materials and methods, simulations models and steps to increasing security (Rees, 2013) on the CAN. We addressed the research question concerning the use of opensource solutions to protect the CAN and establish a virtualized cyber learning arena to help defenders hone their defensive skills. The open-source solution is an effective way of control security (Wilson, 2005) cost for an organization but requires a highly skilled cybersecurity team to customize and configure the open-source solutions.

The methods in this dissertation are not perfect as more work needs to be the focus on improving the simulation model to scale more extensive operations. Additionally, more ICS forensic tools and robust security tools need to be developed such as honeypots for DNP3 (Open DNP3, 2011), BACnet (ASHRAE BACnet, 2014), and other ICS protocols in the industry. Machine learning algorithms will require more robust testing and experimentation to reduce false positives, and network visualization tools need to be more ICS (Bruneau, 2010) oriented. Most importantly, as devices developed for the ICS community, there should be encouragement and support for open-source projects to these tools to encourage increase security on the CAN further.

Research methodology

The Qualitative research methodology is the most suitable data collection, analysis and interpretation technique for this research. It focuses on obtaining data through open-ended and conversational communication (Tyree, 2018). It assists in the collection of data in a non-numeric format helping the researchers to explore different methods of decision making for detailed insight. Qualitative data collection method will help the researchers to collect an enormous amount of data collected through, interviews, focus groups, written notes or videotaping. The methodology of this research study is the qualitative technique for the data collection, analysis and interpretation (Schlüter& Böck, 2013). The research technique adopts a descriptive design and applies qualitative approaches to win data collection and analysis.

Qualitative method will be the approach because it will allow for an in-depth collection of data in terms of opinions; attitudes and feelings of the participants identified. The primary and secondary data will be collected (Moore, 2018). Primary data will be obtained from interactions with participants in the field through the research instruments such as questionnaires, interviews, and focus group discussion guides as well as observation sheets. Secondary data is gathered from library books, journals, thesis, and newspapers dissertations. Data analysis will start right in the field to help to avoid loss of vital information (Parvez, 2017). Content and document review analysis will be carried out to test the applicability of the information from secondary data sources to this study. Analyzed data will be reported in the form of narratives with the first-hand quotation from primary sources.

Proposition

The research for paper is considered as an aspect of developing the already present structure and workflow. There are substantial evidence and events that have shown the desired intent of the hackers and criminals who will penetrate through the CAN and misuse the data. To further continue the research, there is a need for enhancing the usage of technologies and bring in place the machine learning that will be the steel base for the operation of CAN. Hence, the focus is towards the machine learning and artificial intelligence and the usage of it in CAN.

Literature review

The machine learning is vital in the physical security of the control area network station because of its protection and high-security level (Narayanan, Mittal & Joshi, 2016). It enhances the level of the security within the area through the codes systems and the initiation of the artificial intelligence techniques incorporated by the machine learning (Hussain, Hussain, Hassan & Hossain, 2019). Machine learning enhances the application of Artificial intelligence techniques within the physical security of making many organizations get efficient and trusted security through technology (Song, Xiong & Chen, 2016). Physical security of at a control area network station is vital and protecting it is a priority. It entails the protection of the critical data, network software, and other confidential information within the station.

It is also about securing the network equipment, facilities, and company assets through proper management (Tomlinson, Bryans & Shaikh, 2018). Physical security faces various risks such as the attack by nature including power fluctuation, flood, and fire and malicious party such as the terrorism, vandalism, theft (Hermans, Denil, De Meulenaere & Anthonis, 2009). Different organizations face different kinds of physical security threats but the machine learning enhances the level of the organizational security detail elevating their performance in their areas of operations (Krakovna & Doshi-Velez, 2016). Machine learning employs the Artificial intelligence (AI) technology in the physical security of vital installation in the society by incorporating algorithms to perform some intelligent tasks based on the software and hardware components(Singla & Sharma,2019).

The core object of machine learning is the availability of data to be manipulated to produce a given solution (El Masri, Artail & Akkary, 2017). Large datasets are the basis for efficient performances and improvement in the application of machine learning. The machine learning is supported by other sectors in technology such as computer science, neural networks and deep learning for a better outcome (Hines, Bankston & Aldrich, 2015). It also increases the level of security details involving physical security by using latest techniques in smartphones, automobile, healthcare, which makes the physical security market lucrative through the adoption of the Artificial Intelligence technologies(Fugiglando, Santi, Milardo, Abida & Ratti,2017,). Machine learning enhanced through the implementation of the intelligent devices through the cloud or the hybrid combination of two approaches with every tactic evolving with its advantages and disadvantages (Levi, Allouche & Kontorovich, 2018). A stronger machine learning initiation in the physical security application ensure offsetting possible dangers within the security applications.

The machine learning will increase to the changes in the system performing various security tasks attributed to artificial intelligence (AI), and such operations include recognition, diagnosis, planning, prediction and the robotic controls(Zimmer & Rothman, 2009). Changes in the process of the system enhance the functionality of the existing systems into a model of a new unique system with higher efficiency and functionality (Sacha et al., 2016). The machine learning focuses on computer vision, natural language processing, hearing, and image processing and pattern recognition, including cognitive computing (Littman, 2012). Machine learning involved in physical security provides various machines with the abilities to collect data through senses similar to human senses.

The collected data is then processed through the computational intelligence technique and the machine learning methods to conduct predictions and making decisions similar to human beings (Fox, Sudderth, Jordan & Willsky, 2011). The objectives and importance of machine learning is the ability to enhance predictions and perform the clustering through the extraction of the associated rules to make proper decisions from a particular dataset. Machine learning initiates the machine to learn without programming making then intelligent to perform complex tasks within the industries and organizations in the society (Agusti & Antoni, 2009). There are various divisions of the machine learning techniques enhancing the application of the AI into the community, such as the supervised, unsupervised, and reinforcement learning techniques.

Results and discussion

Following an intensive data collection, analysis and interpretation from different respondent about the research topic. It was discovered that machine learning is essential in the enhancement of the security details of formal installations (Calderon & Bloom, 2015). The process is possible through the use of intelligent devices and algorithms to code the complex operations for elevating different tasks. The application of the latest technology has also increased the ability of the development of more sophisticated systems to perform various pre-determined services (Clancy & Goergen, 2008). The incorporation of the machine learning operations focuses on the senor-human based activity recognition and depend on the learning and the data sets collected within the context of a particular action driven with predefined algorithmic codes (Filev, Chinnam, Tseng & Baruah, 2010). The rapid enhancement of the machine learning to physical security is expedited artificial intelligence based technologies in the video surveillance systems and the computer-assisted diagnosis scans.

The real-time crime centres also inculcate the use of machine learning to boost their physical security details because of the predictive analysis tools uses by the detectives for easy data mining (Katanić & Fertalj, 2017). There is massive development of the algorithm efficiency within the video surveillance analytic market used by the developers of the machine learning devices to maximize the output efficiency (Sankavaram, Kodali Pattipati, & Singh, 2015). The independent software vendors always train the algorithms in the cloud using the Amazon Web Services (AWS) for expansive operations (Ivanov, 2017).

Recommendation and conclusion

According to the importance and the advancement of Machine learning techniques that have been achieved in the recent past. I encourage the implementation and the use of technology to expand the level of system operations in different sectors of society. I would recommend the technological studies in the community to provide people with a proper understanding of the machine learning devices and their specific operation. I recommend government intervention in the improvement of the use of machine learning operations within various sectors in society to help improve the technological standards. The government should engage in sponsoring technical education by introducing the Computer Science and Engineering Colleges to enhance the study of Science related courses.

In conclusion, the use of machine learning has improved the level of performance of different electronic devices and security in society. Machine learning applications within the physical security make most machines have the ability to gather data through senses similar to human senses. The collected data are processed through the computational intelligence method and the machine learning tactics to conduct the predictions and making decisions related to ordinary people. The objectives and importance of machine learning is the ability to enhance predictions and perform the clustering through the extraction of the associated rules to make proper decisions from a dataset. Machine learning initiates the machine to learn without programming making then intelligent to perform complex tasks within the industries and organizations in the society. There are various divisions of the machine learning techniques enhancing the application of the AI into the community, such as the supervised, unsupervised, and reinforcement learning techniques. The primary objective and purpose of this research are to elucidate on the operations of the machine learning and how it enhances the physical security at control area network stations.

Definition of Terms:

Physical Security – Physical Security is the protection of personal hardware, software, networks and data from physical actions or events that can cause damages in various levels (Rouse, 2016).

Machine Learning – Machine learnings algorithm will allow the computers to think and take decisions on their own (Maini, 2017).

Control Area Network – Serial Communication Bus initially developed for the automotive industry to replace complicated two-wire bus (Corrigan, 2016).

Artificial Intelligence – Artificial Intelligence is tending to make the computer think like a human and make the best decision possible (Maini, 2017)

Datasets – Dataset can be defined as a set of data that is meaningful (Paruchuri, 2019)

Robust testing – Robust testing can be defined as the process of verifying the robustness of a method (Mwaura, 2019)

Natural Language Processing – Natural Language Processing is the technology which is used to aid computers to understand the natural language of human (Garbade, 2018)

Image Processing – It is the method to convert image into digital form (Mary, 2019)

Cognitive Computing – Cognitive computing is the process to simulate human thought processes in a computerized model (Marr, 2016)

Pattern Recognition – It is the process of recognizing patterns by using machine learning algorithms (Ansari, 2018)

Algorithms - Algorithms (Jason.b,2010) are often grouped by similarity in terms of their function (how they work). For example, tree-based methods and neural network inspired ways.

Datacenter - A data centre is a facility that centralizes an organization's IT operations and equipment, as well as where it stores, manages, and disseminates its data (Barroso, 2019)

Data Analytics - Data analytics (Brett, 2018) is the process of aggregating, parsing, and visualizing of data generated by software from a wide variety of sources.

Vital: The definition of fundamental (Lief, 2018) is something that is essential or necessary or a person who is lively and full of life.

Data Clusters – The set of clusters which defines a data is Data clusters (Barroso, 2019)

References

Agusti, S., & Antoni, M. B. (2009). Advances in artificial intelligence for privacy protection and security (Vol. 1). World Scientific.

Bai, H., Cai, S., Ye, N., Hsu, D., & Lee, W. S. (2015, May). Intention-aware online POMDP planning for autonomous driving in a crowd. In 2015 IEEE international conference on robotics and automation (icra) (pp. 454-460). IEEE

Bakfan, S., Manos, H., Itay, P. E. R. I., & Neulander, O. (2017). U.S. Patent No. 9,767,624. Washington, DC: U.S. Patent and Trademark Office.

Calderon, C. P., & Bloom, K. (2015). Inferring latent states and refining force estimates via hierarchical Dirichlet process modeling in single particle tracking experiments. PloS one, 10(9), e0137633.

Jason, B.: A Tour of Machine Learning Algorithms, https://machinelearningmastery.com/a-tour-

of-machine-learning-algorithms/.

Chhetri, S. R. (2016). Novel side-channel attack model for cyber-physical additive manufacturing systems (Doctoral dissertation, University of California, Irvine).

Barroso, L.A.,  Urs Hoelzle: An Introduction to the Design of Warehouse-Scale Machines, https://dl.acm.org/citation.cfm?id=1643608.

Brett, N.: Machine Data Analytics, https://www.sumologic.com/machine-data-analytics/.

Clancy, T. C., & Goergen, N. (2008, May). Security in cognitive radio networks: Threats and mitigation. In 2008 3rd International Conference on Cognitive Radio Oriented Wireless Networks and Communications (CrownCom 2008)(pp. 1-8). IEEE.

El Masri, A. E. B., Artail, H., & Akkary, H. (2017, November). Toward self-policing: Detecting drunk driving behaviors through sampling CAN bus data. In 2017 International Conference on Electrical and Computing Technologies and Applications (ICECTA) (pp. 1-5). IEEE.

Filev, D. P., Chinnam, R. B., Tseng, F., & Baruah, P. (2010). An industrial strength novelty detection framework for autonomous equipment monitoring and diagnostics. IEEE Transactions on Industrial Informatics, 6(4), 767-779.

Fox, E., Sudderth, E. B., Jordan, M. I., & Willsky, A. S. (2011). Bayesian nonparametric inference of switching dynamic linear models. IEEE Transactions on Signal Processing, 59(4), 1569-1585.

Fugiglando, U., Santi, P., Milardo, S., Abida, K., & Ratti, C. (2017, October). Characterizing the driver dna through can bus data analysis. In Proceedings of the 2Nd ACM International Workshop on Smart, Autonomous, and Connected Vehicular Systems and Services (pp. 37-41). ACM.

Hartmann, T., Moawad, A., Fouquet, F., & Le Traon, Y. (2019). The next evolution of mde: a seamless integration of machine learning into domain modeling. Software & Systems Modeling, 18(2), 1285-1304.

Hermans, T., Denil, J., De Meulenaere, P., & Anthonis, J. (2009). Decoding of data on a CAN powertrain network. In 16th Annual Symposium on Communication and Vehicular Technology in the Benelux (1) (Vol. 6).

Hines, K. E., Bankston, J. R., & Aldrich, R. W. (2015). Analyzing single-molecule time series via nonparametric Bayesian inference. Biophysical Journal, 108(3), 540-556.

Holzinger, A. (2016). Interactive machine learning for health informatics: when do we need the human-in-the-loop? Brain Informatics, 3(2), 119-131.

Humphrey, E. J., & Bello, J. P. (2012, December). Rethinking automatic chord recognition with convolutional neural networks. In 2012 11th International Conference on Machine Learning and Applications (Vol. 2, pp. 357-362). IEEE.

Hussain, F., Hussain, R., Hassan, S. A., & Hossain, E. (2019). Machine Learning in IoT Security: Current Solutions and Future Challenges. arXiv preprint arXiv:1904.05735.

Ivanov, R. S. (2017). Context-Aware Sensor Fusion For Securing Cyber-Physical Systems.

Katanić, N., & Fertalj, K. (2017). Improving physical security with machine learning and sensor-based human activity recognition. WSEAS transactions on information science and applications, 14, 1.

Kerekes, R. A., Karnowski, T. P., Kuhn, M., Moore, M. R., Stinson, B., Tokola, R., ... & Vann, J. M. (2017, June). Vehicle classification and identification using multi-modal sensing and signal learning. In 2017 IEEE 85th Vehicular Technology Conference (VTC Spring) (pp. 1-5). IEEE.

Krakovna, V., & Doshi-Velez, F. (2016). Increasing the interpretability of recurrent neural networks using hidden markov models. arXiv preprint arXiv:1606.05320.

Levi, M., Allouche, Y., & Kontorovich, A. (2018, June). Advanced analytics for connected car cybersecurity. In 2018 IEEE 87th Vehicular Technology Conference (VTC Spring)(pp. 1-7). IEEE.

Lin, J., Yu, W., Yang, X., Xu, G., & Zhao, W. (2012, April). On false data injection attacks against distributed energy routing in smart grid. In Proceedings of the 2012 IEEE/ACM Third International Conference on Cyber-Physical Systems (pp. 183-192). IEEE Computer Society.

Littman, M. L. (2012, August). Inducing partially observable Markov decision processes. In International Conference on Grammatical Inference (pp. 145-148).

Liu, J., Zhu, L., Wang, Y., Liang, X., Hyyppä, J., Chu, T. ... & Chen, R. (2015). Reciprocal estimation of pedestrian location and motion state toward a smartphone geo-context computing solution. Micromachines, 6(6), 699-717.

Liu, J., Zhu, L., Wang, Y., Liang, X., Hyyppä, J., Chu, T., ... & Chen, R. (2015). Reciprocal estimation of pedestrian location and motion state toward a smartphone geo-context computing solution. Micromachines, 6(6), 699-717.

Moore, M. R. (2018). Machine learning for cyber-physical system protection of can bus enabled vehicles using driver state discovery (Doctoral dissertation, Tennessee Technological University)

Moore, M. R. (2018). Machine learning for cyber-physical system protection of can bus enabled vehicles using driver state discovery (Doctoral dissertation, Tennessee Technological University).

Moore, M. R., & Buckner, M. A. (2012). Learning-from-signals on edge devices. IEEE Instrumentation & Measurement Magazine, 15(2), 40-44.

Moore, M. R., & Vann, J. M. (2019, January). Anomaly Detection of Cyber Physical Network Data Using 2D Images. In 2019 IEEE International Conference on Consumer Electronics (ICCE) (pp. 1-5). IEEE.

Moore, M. R., & Vann, J. M. (2019, January). Anomaly Detection of Cyber Physical Network Data Using 2D Images. In 2019 IEEE International Conference on Consumer Electronics (ICCE) (pp. 1-5). IEEE.

Mundhenk, P., Steinhorst, S., Lukasiewycz, M., Fahmy, S. A., & Chakraborty, S. (2015, June). Security analysis of automotive architectures using probabilistic model checking. In Proceedings of the 52nd Annual Design Automation Conference (p. 38). ACM.

Narayanan, S. N., Mittal, S., & Joshi, A. (2016, September). Using semantic technologies to mine vehicular context for security. In 2016 IEEE 37th Sarnoff Symposium (pp. 124-129). IEEE.

Papernot, N., McDaniel, P., Sinha, A., & Wellman, M. (2016). Towards the science of security and privacy in machine learning. arXiv preprint arXiv:1611.03814.

Parvez, B. (2017). Embedded Vision Machine Learning on Embedded Devices for Image classification in Industrial Internet of things.

Pattipati, K., Kodali, A., Luo, J., Choi, K., Singh, S., Sankavaram, C., ... & Qiao, L. (2008). An integrated diagnostic process for automotive systems. In Computational intelligence in automotive applications (pp. 191-218). Springer, Berlin, Heidelberg.

Pons, J., & Serra, X. (2017, March). Designing efficient architectures for modeling temporal features with convolutional neural networks. In 2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (pp. 2472-2476). IEEE.

Sacha, D., Sedlmair, M., Zhang, L., Lee, J. A., Weiskopf, D., North, S., & Keim, D. (2016, August). Human-centered machine learning through interactive visualization. ESANN.

Sankavaram, C., Kodali, A., Pattipati, K. R., & Singh, S. (2015). Incremental classifiers for data-driven fault diagnosis applied to automotive systems. IEEE access, 3, 407-419.

Schlüter, J., & Böck, S. (2013, September). Musical onset detection with convolutional neural networks. In 6th international workshop on machine learning and music (MML), Prague, Czech Republic.

Singla, A., & Sharma, A. (2019). Physical Access System Security of IoT Devices using Machine Learning Techniques. Available at SSRN 3356785.

Sinha, A., Nguyen, T. H., Kar, D., Brown, M., Tambe, M., & Jiang, A. X. (2015). From physical security to cybersecurity. Journal of Cybersecurity, 1(1), 19-35

Song, W., Xiong, G., & Chen, H. (2016). Intention-aware autonomous driving decision-making in an uncontrolled intersection. Mathematical Problems in Engineering, 2016.

Staggs, J. (2013). How to hack your mini cooper: reverse engineering can messages on passenger automobiles. Institute for Information Security.

Tomlinson, A., Bryans, J., & Shaikh, S. A. (2018, September). Towards viable intrusion detection methods for the automotive controller area network. In 2nd ACM Computer Science in Cars Symposium.

Tyree, Z. (2018). Characterizing the Geometry of a Random Point Cloud (Doctoral dissertation, Florida Atlantic University).

Zimmer, V. J., & Rothman, M. A. (2009). U.S. Patent No. 7,496,961. Washington, DC: U.S. Patent and Trademark Office.

Hurst, C. (2017, November 02). Machine Learning for Physical Security - an Interview with

Stabilitas Data Scientist Dr. Retrieved July 14, 2019, from

https://medium.com/stabilitas/machine-learning-for-physical-security-an-interview-with-stabilitas-data-scientist-dr-9f17568c34d

Jesus, A. D. (2019, May 17). AI for Physical Security – 4 Current Applications. Retrieved July

14, 2019, from https://emerj.com/ai-sector-overviews/ai-for-physical-security/

Corrigan, S. (2016). Road vehicles. Controller area network (CAN). Introduction to the

Controller Area Network, 2-2. doi:10.3403/bsiso11898.

Ansari, S. (2018, May 04). Pattern Recognition: Introduction. Retrieved from

https://www.geeksforgeeks.org/pattern-recognition-introduction/

Mary, R. (n.d.). Introduction to Image Processing. Retrieved July 14, 2019, from

https://www.engineersgarage.com/articles/image-processing-tutorial-applications

Marr, B. (2016, March 23). What Everyone Should Know About Cognitive Computing.

Retrieved July 14, 2019, from https://www.forbes.com/sites/bernardmarr/2016/03/23/what-everyone-should-know-about-cognitive-computing/#2e62d12f5088

Garbade, M. (2018, October 15). A Simple Introduction to Natural Language Processing.

Retrieved from https://becominghuman.ai/a-simple-introduction-to-natural-language-processing-ea66a1747b32

Mwaura, W. (2019, April 30). Towards a robust testing future: Unit Vs Integration Tests.

Retrieved July 14, 2019, from https://medium.com/the-andela-way/towards-a-robust-testing-future-unit-vs-integration-tests-700f7a1f6a7f

Paruchuri, V., & Dataquest. (2019, June 19). Free Data Sets for Data Science Projects. Retrieved

July 14, 2019, from https://www.dataquest.io/blog/free-datasets-for-projects/

Parikh, B. (n.d.). CAN Protocol - Understanding the Controller Area Network Protocol.

Retrieved July 14, 2019, from https://www.engineersgarage.com/article/what-is-controller-area-network

Maini, V. (2018, May 28). A Beginner's Guide to AI/ML - Machine Learning for Humans.

Retrieved July 14, 2019, from https://medium.com/machine-learning-for-humans/why-machine-learning-matters-6164faf1df12

Rouse, M. (n.d.). What is physical security? - Definition from WhatIs.com. Retrieved July 14,

2019, from https://searchsecurity.techtarget.com/definition/physical-security

Lief, P. (2018): Synonyms of vital, https://www.thesaurus.com/browse/vital.