Derivatives are key to numerous science, engineering, and machine learning applications. While existing tools generate derivatives of programs in a single language, modern parallel applications combine a set of frameworks and languages to leverage available performance and function in an evolving hardware landscape. We propose a scheme for differentiating arbitrary DAG-based parallelism that preserves scalability and efficiency, implemented into the LLVM-based Enzyme automatic differentiation framework. By integrating with a full-fledged compiler backend, Enzyme can differentiate numerous parallel frameworks and directly control code generation. Combined with its ability to differentiate any LLVM-based language, this flexibility permits Enzyme to leverage the compiler tool chain for parallel and differentiation-specitic optimizations. We differentiate nine distinct versions of the LULESH and miniBUDE applications, written in different programming languages (C++, Julia) and parallel frameworks (OpenMP, MPI, RAJA, Julia tasks, MPI.jl), demonstrating similar scalability to the original program. On benchmarks with 64 threads or nodes, we find a differentiation overhead of 3.4–6.8× on C++ and 5.4–12.5× on Julia.

Artificial intelligence (AI) and machine learning (ML) have been used in transforming our environment and the way people think, behave, and make decisions during the last few decades [1]. In the last two decades everyone connected to the Internet either an enterprise or individuals has become concerned about the security of his/their computational resources. Cybersecurity is responsible for protecting hardware and software resources from cyber attacks e.g. viruses, malware, intrusion, eavesdropping. Cyber attacks either come from black hackers or cyber warfare units. Artificial intelligence (AI) and machine learning (ML) have played an important role in developing efficient cyber security tools. This paper presents Latest Cyber Security Tools Based on Machine Learning which are: Windows defender ATP, DarckTrace, Cisco Network Analytic, IBM QRader, StringSifter, Sophos intercept X, SIME, NPL, and Symantec Targeted Attack Analytic.

A huge amount of stored and transferred data is expanding rapidly. Therefore, managing and securing the big volume of diverse applications should have a high priority. However, Structured Query Language Injection Attack (SQLIA) is one of the most common dangerous threats in the world. Therefore, a large number of approaches and models have been presented to mitigate, detect or prevent SQL injection attack but it is still alive. Most of old and current models are created based on static, dynamic, hybrid or machine learning techniques. However, SQL injection attack still represents the highest risk in the trend of web application security risks based on several recent studies in 2021. In this paper, we present a review of the latest research dealing with SQL injection attack and its types, and demonstrating several types of most recent and current techniques, models and approaches which are used in mitigating, detecting or preventing this type of dangerous attack. Then, we explain the weaknesses and highlight the critical points missing in these techniques. As a result, we still need more efforts to make a real, novel and comprehensive solution to be able to cover all kinds of malicious SQL commands. At the end, we provide significant guidelines to follow in order to mitigate such kind of attack, and we strongly believe that these tips will help developers, decision makers, researchers and even governments to innovate solutions in the future research to stop SQLIA.

Electrical substations in power grid act as the critical interface points for the transmission and distribution networks. Over the years, digital technology has been integrated into the substations for remote control and automation. As a result, substations are more prone to cyber attacks and exposed to digital vulnerabilities. One of the notable cyber attack vectors is the malicious command injection, which can lead to shutting down of substations and subsequently power outages as demonstrated in Ukraine Power Plant Attack in 2015. Prevailing measures based on cyber rules (e.g., firewalls and intrusion detection systems) are often inadequate to detect advanced and stealthy attacks that use legitimate-looking measurements or control messages to cause physical damage. Additionally, defenses that use physics-based approaches (e.g., power flow simulation, state estimation, etc.) to detect malicious commands suffer from high latency. Machine learning serves as a potential solution in detecting command injection attacks with high accuracy and low latency. However, sufficient datasets are not readily available to train and evaluate the machine learning models. In this paper, focusing on this particular challenge, we discuss various approaches for the generation of synthetic data that can be used to train the machine learning models. Further, we evaluate the models trained with the synthetic data against attack datasets that simulates malicious commands injections with different levels of sophistication. Our findings show that synthetic data generated with some level of power grid domain knowledge helps train robust machine learning models against different types of attacks.

For a long time, online attacks were regarded to pose a severe threat to web - based applications, websites, and clients. It can bypass authentication methods, steal sensitive information from datasets and clients, and also gain ultimate authority of servers. A variety of ways for safeguarding online apps have been developed and used to deal the website risks. Based on the studies about the intersection of cybersecurity and machine learning, countermeasures for identifying typical web assaults have recently been presented (ML). In order to establish a better understanding on this essential topic, it is necessary to study ML methodologies, feature extraction techniques, evaluate datasets, and performance metrics utilised in a systematic manner. In this paper, we go through web security flaws like SQLi, XSS, malicious URLs, phishing attacks, path traversal, and CMDi in detail. We also go through the existing security methods for detecting these threats using machine learning approaches for URL classification. Finally, we discuss potential research opportunities for ML and DL-based techniques in this category, based on a thorough examination of existing solutions in the literature.

Most of the recent high-profile attacks targeting cyber-physical systems (CPS) started with lengthy reconnaissance periods that enabled attackers to gain in-depth understanding of the victim’s environment. To simulate these stealthy attacks, several covert channel tools have been published and proven effective in their ability to blend into existing CPS communication streams and have the capability for data exfiltration and command injection.In this paper, we report a novel machine learning feature engineering and data processing pipeline for the detection of covert channel attacks on CPS systems with real-time detection throughput. The system also operates at the network layer without requiring physical system domain-specific state modeling, such as voltage levels in a power generation system. We not only demonstrate the effectiveness of using TCP payload entropy as engineered features and the technique of grouping information into network flows, but also pitch the proposed detector against scenarios employing advanced evasion tactics, and still achieve above 99% detection performance.

With the rapid development of artificial intelligence (AI), many companies are moving towards automating their services using automated conversational agents. Dialogue-based conversational recommender agents, in particular, have gained much attention recently. The successful development of such systems in the case of natural language input is conditioned by the ability to understand the users’ utterances. Predicting the users’ intents allows the system to adjust its dialogue strategy and gradually upgrade its preference profile. Nevertheless, little work has investigated this problem so far. This paper proposes an LSTM-based Neural Network model and compares its performance to seven baseline Machine Learning (ML) classifiers. Experiments on a new publicly available dataset revealed The superiority of the LSTM model with 95% Accuracy and 94% F1-score on the full dataset despite the relatively small dataset size (9300 messages and 17 intents) and label imbalance.

Due to the migration megatrend, efficient and effective second-language acquisition is vital. One proposed solution involves AI-enabled conversational agents for person-centered interactive language practice. We present results from ongoing action research targeting quality assurance of proprietary generative dialog models trained for virtual job interviews. The action team elicited a set of 38 requirements for which we designed corresponding automated test cases for 15 of particular interest to the evolving solution. Our results show that six of the test case designs can detect meaningful differences between candidate models. While quality assurance of natural language processing applications is complex, we provide initial steps toward an automated framework for machine learning model selection in the context of an evolving conversational agent. Future work will focus on model selection in an MLOps setting.

Over the past two decades, several forms of non-intrusive technology have been deployed in cooperation with medical specialists in order to aid patients diagnosed with some form of mental, cognitive or psychological condition. Along with the availability and accessibility to applications offered by mobile devices, as well as the advancements in the field of Artificial Intelligence applications and Natural Language Processing, Conversational Agents have been developed with the objective of aiding medical specialists detecting those conditions in their early stages and monitoring their symptoms and effects on the cognitive state of the patient, as well as supporting the patient in their effort of mitigating those symptoms. Coupled with the recent advances in the the scientific field of machine and deep learning, we aim to explore the grade of applicability of such technologies into cognitive health support Conversational Agents, and their impact on the acceptability of such applications bytheir end users. Therefore, we conduct a systematic literature review, following a transparent and thorough process in order to search and analyze the bibliography of the past five years, focused on the implementation of Conversational Agents, supported by Artificial Intelligence technologies and in service of patients diagnosed with Mild Cognitive Impairment and its variants.

The emergence of smart cars has revolutionized the automotive industry. Today's vehicles are equipped with different types of electronic control units (ECUs) that enable autonomous functionalities like self-driving, self-parking, lane keeping, and collision avoidance. The ECUs are connected to each other through an in-vehicle network, named Controller Area Network. In this talk, we will present the different cyber attacks that target autonomous vehicles and explain how an intrusion detection system (IDS) using machine learning can play a role in securing the Controller Area Network. We will also discuss the main research contributions for the security of autonomous vehicles. Specifically, we will describe our IDS, named Histogram-based Intrusion Detection and Filtering framework. Next, we will talk about the machine learning explainability issue that limits the acceptability of machine learning in autonomous vehicles, and how it can be addressed using our novel intrusion detection system based on rule extraction methods from Deep Neural Networks.

Modern connected vehicles are equipped with a large number of sensors, which enable a wide range of services that can improve overall traffic safety and efficiency. However, remote access to connected vehicles also introduces new security issues affecting both inter and intra-vehicle communications. In fact, existing intra-vehicle communication systems, such as Controller Area Network (CAN), lack security features, such as encryption and secure authentication for Electronic Control Units (ECUs). Instead, Original Equipment Manufacturers (OEMs) seek security through obscurity by keeping secret the proprietary format with which they encode the information. Recently, it has been shown that the reuse of CAN frame IDs can be exploited to perform CAN bus reverse engineering without physical access to the vehicle, thus raising further security concerns in a connected environment. This work investigates whether anonymizing the frames of each newly released vehicle is sufficient to prevent CAN bus reverse engineering based on frame ID matching. The results show that, by adopting Machine Learning techniques, anonymized CAN frames can still be fingerprinted and identified in an unknown vehicle with an accuracy of up to 80 %.

Intrusion detection for Controller Area Network (CAN) protocol requires modern methods in order to compete with other electrical architectures. Fingerprint Intrusion Detection Systems (IDS) provide a promising new approach to solve this problem. By characterizing network traffic from known ECUs, hazardous messages can be discriminated. In this article, a modified version of Fingerprint IDS is employed utilizing both step response and spectral characterization of network traffic via neural network training. With the addition of feature set reduction and hyperparameter tuning, this method accomplishes a 99.4% detection rate of trusted ECU traffic.

The exponential growth of IoT-type systems has led to a reconsideration of the field of database management systems in terms of storing and handling high-volume data. Recently, many real-time Database Management Systems(DBMS) have been developed to address issues such as security, managing concurrent access to stored data, and optimizing data query performance. This paper studies methods that allow to reduce the temporal validity range for common DBMS. The primary purpose of IoT edge devices is to generate data and make it available for machine learning or statistical algorithms. This is achieved inside the Knowledge Discovery in Databases process. In order to visualize and obtain critical Data Mining results, all the device-generated data must be made available as fast as possible for selection, preprocessing and data transformation. In this research we investigate if IoT edge devices can be used with common DBMS proper configured in order to access data fast instead of working with Real Time DBMS. We will study what kind of transactions are needed in large IoT ecosystems and we will analyze the techniques of controlling concurrent access to common resources (stored data). For this purpose, we built a series of applications that are able to simulate concurrent writing operations to a common DBMS in order to investigate the performance of concurrent access to database resources. Another important procedure that will be tested with the developed applications will be to increase the availability of data for users and data mining applications. This will be achieved by using field indexing.

False data injection cyber-attack detection models on smart grid operation have been much explored recently, considering analytical physics-based and data-driven solutions. Recently, a hybrid data-driven physics-based model framework for monitoring the smart grid is developed. However, the framework has not been implemented in real-time environment yet. In this paper, the framework of the hybrid model is developed within a real-time simulation environment. OPAL-RT real-time simulator is used to enable Hardware-in-the-Loop testing of the framework. IEEE 9-bus system is considered as a testing grid for gaining insight. The process of building the framework and the challenges faced during development are presented. The performance of the framework is investigated under various false data injection attacks.

One of the major concerns in the real-time monitoring systems in a smart grid is the Cyber security threat. The false data injection attack is emerging as a major form of attack in Cyber-Physical Systems (CPS). A False data Injection Attack (FDIA) can lead to severe issues like insufficient generation, physical damage to the grid, power flow imbalance as well as economical loss. The recent advancements in machine learning algorithms have helped solve the drawbacks of using classical detection techniques for such attacks. In this article, we propose to use Autoencoders (AE’s) as a novel Machine Learning approach to detect FDI attacks without any major modifications. The performance of the method is validated through the analysis of the simulation results. The algorithm achieves optimal accuracy owing to the unsupervised nature of the algorithm.

The Internet of Things is a developing technology that converts physical objects into virtual objects connected to the internet using wired and wireless network architecture. Use of cross-layer techniques in the internet of things is primarily driven by the high heterogeneity of hardware and software capabilities. Although traditional layered architecture has been effective for a while, cross-layer protocols have the potential to greatly improve a number of wireless network characteristics, including bandwidth and energy usage. Also, one of the main concerns with the internet of things is security, and machine learning (ML) techniques are thought to be the most cuttingedge and viable approach. This has led to a plethora of new research directions for tackling IoT's growing security issues. In the proposed study, a number of cross-layer approaches based on machine learning techniques that have been offered in the past to address issues and challenges brought on by the variety of IoT are in-depth examined. Additionally, the main issues are mentioned and analyzed, including those related to scalability, interoperability, security, privacy, mobility, and energy utilization.

In the deep nano-scale regime, reliability has emerged as one of the major design issues for high-density integrated systems. Among others, key reliability-related issues are soft errors, high temperature, and aging effects (e.g., NBTI-Negative Bias Temperature Instability), which jeopardize the correct applications' execution. Tremendous amount of research effort has been invested at individual system layers. Moreover, in the era of growing cyber-security threats, modern computing systems experience a wide range of security threats at different layers of the software and hardware stacks. However, considering the escalating reliability and security costs, designing a highly reliable and secure system would require engaging multiple system layers (i.e. both hardware and software) to achieve cost-effective robustness. This talk provides an overview of important reliability issues, prominent state-of-the-art techniques, and various hardwaresoftware collaborative reliability modeling and optimization techniques developed at our lab, with a focus on the recent works on ML-based reliability techniques. Afterwards, this talk will also discuss how advanced ML techniques can be leveraged to devise new types of hardware security attacks, for instance on logic locked circuits. Towards the end of the talk, I will also give a quick pitch on the reliability and security challenges for the embedded machine learning (ML) on resource/energy-constrained devices subjected to unpredictable and harsh scenarios.

In the Smart Grid paradigm, this critical infrastructure operation is increasingly exposed to cyber-threats due to the increased dependency on communication networks. An adversary can launch an attack on a power grid operation through False Data Injection into system measurements and/or through attacks on the communication network, such as flooding the communication channels with unnecessary data or intercepting messages. A cross-layered strategy that combines power grid data, communication grid monitoring and Machine Learning-based processing is a promising solution for detecting cyber-threats. In this paper, an implementation of an integrated solution of a cross-layer framework is presented. The advantage of such a framework is the augmentation of valuable data that enhances the detection of anomalies in the operation of power grid. IEEE 118-bus system is built in Simulink to provide a power grid testing environment and communication network data is emulated using SimComponents. The performance of the framework is investigated under various FDI and communication attacks.

The value and size of information exchanged through dark-web pages are remarkable. Recently Many researches showed values and interests in using machine-learning methods to extract security-related useful knowledge from those dark-web pages. In this scope, our goals in this research focus on evaluating best prediction models while analyzing traffic level data coming from the dark web. Results and analysis showed that feature selection played an important role when trying to identify the best models. Sometimes the right combination of features would increase the model’s accuracy. For some feature set and classifier combinations, the Src Port and Dst Port both proved to be important features. When available, they were always selected over most other features. When absent, it resulted in many other features being selected to compensate for the information they provided. The Protocol feature was never selected as a feature, regardless of whether Src Port and Dst Port were available.

Cyber threats can cause severe damage to computing infrastructure and systems as well as data breaches that make sensitive data vulnerable to attackers and adversaries. It is therefore imperative to discover those threats and stop them before bad actors penetrating into the information systems.Threats hunting algorithms based on machine learning have shown great advantage over classical methods. Reinforcement learning models are getting more accurate for identifying not only signature-based but also behavior-based threats. Quantum mechanics brings a new dimension in improving classification speed with exponential advantage. The accuracy of the AI/ML algorithms could be affected by many factors, from algorithm, data, to prejudicial, or even intentional. As a result, AI/ML applications need to be non-biased and trustworthy.In this research, we developed a machine learning-based cyber threat detection and assessment tool. It uses two-stage (both unsupervised and supervised learning) analyzing method on 822,226 log data recorded from a web server on AWS cloud. The results show the algorithm has the ability to identify the threats with high confidence.

Steady advancement in Artificial Intelligence (AI) development over recent years has caused AI systems to become more readily adopted across industry and military use-cases globally. As powerful as these algorithms are, there are still gaping questions regarding their security and reliability. Beyond adversarial machine learning, software supply chain vulnerabilities and model backdoor injection exploits are emerging as potential threats to the physical safety of AI reliant CPS such as autonomous vehicles. In this work in progress paper, we introduce the concept of AI supply chain vulnerabilities with a provided proof of concept autonomous exploitation framework. We investigate the viability of algorithm backdoors and software third party library dependencies for applicability into modern AI attack kill chains. We leverage an autonomous vehicle case study for demonstrating the applicability of our offensive methodologies within a realistic AI CPS operating environment.

Missing values are an unavoidable problem for classification tasks of machine learning in medical data. With the rapid development of the medical system, large scale medical data is increasing. Missing values increase the difficulty of mining hidden but useful information in these medical datasets. Deletion and imputation methods are the most popular methods for dealing with missing values. Existing studies ignored to compare and discuss the deletion and imputation methods of missing values under the row missing rate and the total missing rate. Meanwhile, they rarely used experiment data sets that are mixed-type and large scale. In this work, medical data sets of various sizes and mixed-type are used. At the same time, performance differences of deletion and imputation methods are compared under the MCAR (Missing Completely At Random) mechanism in the baseline task using LR (Linear Regression) and SVM (Support Vector Machine) classifiers for classification with the same row and total missing rates. Experimental results show that under the MCAR missing mechanism, the performance of two types of processing methods is related to the size of datasets and missing rates. As the increasing of missing rate, the performance of two types for processing missing values decreases, but the deletion method decreases faster, and the imputation methods based on machine learning have more stable and better classification performance on average. In addition, small data sets are easily affected by processing methods of missing values.

Model compression is one of the most preferred techniques for efficiently deploying deep neural networks (DNNs) on resource- constrained Internet of Things (IoT) platforms. However, the simply compressed model is often vulnerable to adversarial attacks, leading to a conflict between robustness and efficiency, especially for IoT devices exposed to complex real-world scenarios. We, for the first time, address this problem by developing a novel framework dubbed Magical-Decomposition to simultaneously enhance both robustness and efficiency for hardware. By leveraging a hardware-friendly model compression method called singular value decomposition, the defending algorithm can be supported by most of the existing DNN hardware accelerators. To step further, by using a recently developed DNN interpretation tool, the underlying scheme of how the adversarial accuracy can be increased in the compressed model is highlighted clearly. Ablation studies and extensive experiments under various attacks/models/datasets consistently validate the effectiveness and scalability of the proposed framework.

Cloud provides access to shared pool of resources like storage, networking, and processing. Distributed denial of service attacks are dangerous for Cloud services because they mainly target the availability of resources. It is important to detect and prevent a DDoS attack for the continuity of Cloud services. In this review, we analyze the different mechanisms of detection and prevention of the DDoS attacks in Clouds. We identify the major DDoS attacks in Clouds and compare the frequently-used strategies to detect, prevent, and mitigate those attacks that will help the future researchers in this area.

One of the major threats in the cyber security and networking world is a Distributed Denial of Service (DDoS) attack. With massive development in Science and Technology, the privacy and security of various organizations are concerned. Computer Intrusion and DDoS attacks have always been a significant issue in networked environments. DDoS attacks result in non-availability of services to the end-users. It interrupts regular traffic flow and causes a flood of flooded packets, causing the system to crash. This research presents a Machine Learning-based DDoS attack detection system to overcome this challenge. For the training and testing purpose, we have used the NSL-KDD Dataset. Logistic Regression Classifier, Support Vector Machine, K Nearest Neighbour, and Decision Tree Classifier are examples of machine learning algorithms which we have used to train our model. The accuracy gained are 90.4, 90.36, 89.15 and 82.28 respectively. We have added a feature called BOTNET Prevention, which scans for Phishing URLs and prevents a healthy device from being a part of the botnet.