Biblio

The problem of information security of critical information infrastructure objects in the conditions of openness is formulated. The concept of information infrastructure openness is analyzed. An approach to assessing the openness of an information system is presented. A set-theoretic model of information resources openness was developed. The formulation of the control problem over the degree of openness with restrictions on risk was carried out. An example of solving the problem of finding the coefficient of openness is presented.

This Innovative Practice Work-in-Progress paper presents a virtual, proactive, and collaborative learning paradigm that can engage learners with different backgrounds and enable effective retention and transfer of the multidisciplinary AI-cybersecurity knowledge. While progress has been made to better understand the trustworthiness and security of artificial intelligence (AI) techniques, little has been done to translate this knowledge to education and training. There is a critical need to foster a qualified cybersecurity workforce that understands the usefulness, limitations, and best practices of AI technologies in the cybersecurity domain. To address this import issue, in our proposed learning paradigm, we leverage multidisciplinary expertise in cybersecurity, AI, and statistics to systematically investigate two cohesive research and education goals. First, we develop an immersive learning environment that motivates the students to explore AI/machine learning (ML) development in the context of real-world cybersecurity scenarios by constructing learning models with tangible objects. Second, we design a proactive education paradigm with the use of hackathon activities based on game-based learning, lifelong learning, and social constructivism. The proposed paradigm will benefit a wide range of learners, especially underrepresented students. It will also help the general public understand the security implications of AI. In this paper, we describe our proposed learning paradigm and present our current progress of this ongoing research work. In the current stage, we focus on the first research and education goal and have been leveraging cost-effective Minecraft platform to develop an immersive learning environment where the learners are able to investigate the insights of the emerging AI/ML concepts by constructing related learning modules via interacting with tangible AI/ML building blocks.

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.

The number of people suffering from mental health issues like depression and anxiety have spiked enormously in recent times. Conversational agents like chatbots have emerged as an effective way for users to express their feelings and anxious thoughts and in turn obtain some empathetic reply that would relieve their anxiety. In our work, we construct two types of empathetic conversational agent models based on sequence-to-sequence modeling with and without attention mechanism. We implement the attention mechanism proposed by Bahdanau et al. for neural machine translation models. We train our model on the benchmark Facebook Empathetic Dialogue dataset and the BLEU scores are computed. Our empathetic conversational agent model incorporating attention mechanism generates better quality empathetic responses and is better in capturing human feelings and emotions in the conversation.

To conceive a CPS is a complex and multidisciplinary endeavour involving different stakeholders, potentially using a plethora of different languages to describe their views of the system at different levels of abstraction. Model-Driven Engineering comes, precisely, as a methodological approach to tackle the complexity of systems development with models as first-class citizens in the development process. The measure of realism of these models with respect to the real (sub)system is called fidelity. Usually, different models with different fidelity are then developed during the development process. Additionally, it is very common that the development process of CPS includes an incremental (and collaborative) use of simulations to study the behaviour emerging from the heterogeneous models of the system. Currently, the different models, with different fidelity, are managed in an ad hoc manner. Consequently, when a (Co)simulation is used to study a specific property of the system, the choice of the different models and their setup is made manually in a non-tractable way. In this paper we propose a structured new vision to CPS development, where the notion of simulation goal and multi-fidelity simulation unit are first-class citizens. The goal is to make a clear link between the system requirements, the system properties, the simulation goal and the multi-fidelity simulation unit. The outcome of this framework is a way to automatically determine the model at an adequate fidelity level suitable for answering a specific simulation goal.

Fueled by the emergence of IoT-enabled medical sensors and big data analytics, nations all over the world are widely adopting digitalization of healthcare systems. This is certainly a positive trend for improving the entire spectrum of quality of care, but this convenience is also posing a huge challenge on the security of healthcare data. For ensuring privacy and protection of healthcare data, access control is regarded as one of the first-line-of-defense mechanisms. As none of the traditional enterprise access control models can completely cater to the need of the healthcare domain which includes a myriad of contexts, in this paper, we present a context-policy-based access control scheme. Our scheme relies on the eTRON cybersecurity architecture for tamper-resistance and cryptographic functions, and leverages a context-specific blend of classical discretionary and role-based access models for incorporation into legacy systems. Moreover, our scheme adheres to key recommendations of prominent statutory and technical guidelines including HIPAA and HL7. The protocols involved in the proposed access control system have been delineated, and a proof-of-concept implementation has been carried out - along with a comparison with other systems, which clearly suggests that our approach is more responsive to different contexts for protecting healthcare data.

The Internet of Things (IoT) has been one of the biggest revelations of the last decade. These cyber-physical systems seamlessly integrate and improve the activities in our daily lives. Hence, creating a wide application for it in several domains, such as smart buildings and cities. However, the integration of IoT also comes with privacy challenges. The privacy challenges result from the ability of these devices to pervasively collect personal data about individuals through sensors in ways that could be unknown to them. A number of research efforts have evaluated privacy policy awareness and enforcement as key components for addressing these privacy challenges. This paper provides a framework for understanding contextualized privacy policy within the IoT domain. This will enable IoT privacy researchers to better understand IoT privacy policies and their modeling.

Intent classification is an important and relevant area of research in artificial intelligence and machine learning, with applications ranging from marketing and product design to intelligent communication. This paper explores the performance of various models and techniques for short text intent classification in the context of chatbots. The problem was explored for use within the mental wellness and therapy chatbot application, Wysa, to give improved responses to free-text user input. The authors looked at classifying text samples in-to 4 categories - assertions, refutations, clarifiers and transitions. For this, the suitability of the following techniques was evaluated: count vectors, TF-IDF, sentence embeddings and n-grams, as well as modifications of the same. Each technique was used to train a number of state-of-the-art classifiers, and the results have been compiled and presented. This is the first documented implementation of Arora's modification to sentence embeddings for real world use. It also introduces a technique to generate custom stop words that gave a significant gain in performance (10 percentage points). The best pipeline, using these techniques together, gave an accuracy of 95 percent.

The following topics are dealt with: authorisation; data privacy; mobile computing; security of data; cryptography; Internet of Things; message authentication; invasive software; Android (operating system); vectors.

The DES is a symmetric cryptosystem which encrypts data in blocks of 64 bits using 48 bit keys in 16 rounds. It comprises a key schedule, encryption and decryption components. The key schedule, in particular, uses three static component units, the PC-1, PC-2 and rotation tables. However, can these three static components of the key schedule be altered? The DES development team never explained most of these component units. Understanding the DES key schedule is, thus, hard. In addition, reproducing the DES model with unknown component units is challenging, making it hard to adapt and bring implementation of the DES model closer to novice developers' context. We propose an alternative approach for re-implementing the DES key schedule using, rather, dynamic instead of static tables. We investigate the design features of the DES key schedule and implement the same. We then propose a re-engineering view towards a more white-box design. Precisely, generation of the PC-1, rotation and PC-2 tables is revisited to random dynamic tables created at run time. In our views, randomly generated component units eliminate the feared concerns regarding perpetrators' possible knowledge of the internal structures of the static component units. Comparison of the performances of the hybrid DES key schedule to that of the original DES key schedule shows closely related outcomes, connoting the hybrid version as a good alternative to the original model. Memory usage and CPU time were measured. The hybrid insignificantly out-performs the original DES key schedule. This outcome may inspire further researches on possible alterations to other DES component units as well, bringing about completely white-box designs to the DES model.

This paper has firstly introduced big data services and cloud computing model based on different process forms, and analyzed the authentication technology and security services of the existing big data to understand their processing characteristics. Operation principles and complexity of the big data services and cloud computing have also been studied, and summary about their suitable environment and pros and cons have been made. Based on the Cloud Computing, the author has put forward the Model of Big Data Cloud Computing based on Extended Subjective Logic (MBDCC-ESL), which has introduced Jφsang's subjective logic to test the data credibility and expanded it to solve the problem of the trustworthiness of big data in the cloud computing environment. Simulation results show that the model works pretty well.

The Internet of Things (IoT) is enabling smart houses, where multiple users with complex social relationships interact with smart devices. This requires sophisticated access control specification and enforcement models, that are currently lacking. In this paper, we introduce the extended generalized role based access control (EGRBAC) model for smart home IoT. We provide a formal definition for EGRBAC and illustrate its features with a use case. A proof-of-concept demonstration utilizing AWS-IoT Greengrass is discussed in the appendix. EGRBAC is a first step in developing a comprehensive family of access control models for smart home IoT.

A common technique for building ICT systems is to build them as successive layers of bstraction: for instance, the Instruction Set Architecture (ISA) is an abstraction of the hardware, and compilers or interpreters build higher level abstractions on top of the ISA.The functionality of an ICT application can often be understood by considering only a single level of abstraction. For instance the source code of the application defines the functionality using the level of abstraction of the source programming language. Functionality can be well understood by just studying this source code.Many important security issues in ICT system however are cross-layer issues: they can not be understood by considering the system at a single level of abstraction, but they require understanding how multiple levels of abstraction are implemented. Attacks may rely on, or exploit, implementation details of one or more layers below the source code level of abstraction.The purpose of this paper is to illustrate this cross-layer nature of security by discussing old and new examples of cross-layer security issues, and by providing a classification of these issues.

Cross-site scripting (XSS) is one of the main threats of Web applications, which has great harm. How to effectively detect and defend against XSS attacks has become more and more important. Due to the malicious obfuscation of attack codes and the gradual increase in number, the traditional XSS detection methods have some defects such as poor recognition of malicious attack codes, inadequate feature extraction and low efficiency. Therefore, we present a novel approach to detect XSS attacks based on the attention mechanism of Long Short-Term Memory (LSTM) recurrent neural network. First of all, the data need to be preprocessed, we used decoding technology to restore the XSS codes to the unencoded state for improving the readability of the code, then we used word2vec to extract XSS payload features and map them to feature vectors. And then, we improved the LSTM model by adding attention mechanism, the LSTM-Attention detection model was designed to train and test the data. We used the ability of LSTM model to extract context-related features for deep learning, the added attention mechanism made the model extract more effective features. Finally, we used the classifier to classify the abstract features. Experimental results show that the proposed XSS detection model based on LSTM-Attention achieves a precision rate of 99.3% and a recall rate of 98.2% in the actually collected dataset. Compared with traditional machine learning methods and other deep learning methods, this method can more effectively identify XSS attacks.

The evolving of context-aware applications are becoming more readily available as a major driver of the growth of future connected smart, autonomous environments. However, with the increasing of security risks in critical shared massive data capabilities and the increasing regulation requirements on privacy, there is a significant need for new paradigms to manage security and privacy compliances. These challenges call for context-aware and fine-grained security policies to be enforced in such dynamic environments in order to achieve efficient real-time authorization between applications and connected devices. We propose in this work a novel solution that aims to provide context-aware security model for Android applications. Specifically, our proposition provides automated context-aware access control model and leverages Attribute-Based Encryption (ABE) to secure data communications. Thorough experiments have been performed and the evaluation results demonstrate that the proposed solution provides an effective lightweight adaptable context-aware encryption model.

Although sequence-to-sequence attentional neural machine translation (NMT) has achieved great progress recently, it is confronted with two challenges: learning optimal model parameters for long parallel sentences and well exploiting different scopes of contexts. In this paper, partially inspired by the idea of segmenting a long sentence into short clauses, each of which can be easily translated by NMT, we propose a hierarchy-to-sequence attentional NMT model to handle these two challenges. Our encoder takes the segmented clause sequence as input and explores a hierarchical neural network structure to model words, clauses, and sentences at different levels, particularly with two layers of recurrent neural networks modeling semantic compositionality at the word and clause level. Correspondingly, the decoder sequentially translates segmented clauses and simultaneously applies two types of attention models to capture contexts of interclause and intraclause for translation prediction. In this way, we can not only improve parameter learning, but also well explore different scopes of contexts for translation. Experimental results on Chinese-English and English-German translation demonstrate the superiorities of the proposed model over the conventional NMT model.

In the context of Advanced Persistent Threat (APT) attacks, this paper introduces a model, called Nuke, which tries to provide a more operational reading of the attackers' lifecycle in a compromised network. It allows to consider the notions of regression; and repetitiveness of final objectives achievement. By confronting this model with examples of recent attacks (Equifax data breach and TV5Monde sabotage), we emphasize the importance of the attack chronology in the Cyber Threat Intelligence (CTI) reports, as well as the Tactics, Techniques and Procedures (TTP) used by the attacker during his progression.

This paper presents a computational model for managing an Embodied Conversational Agent's first impressions of warmth and competence towards the user. These impressions are important to manage because they can impact users' perception of the agent and their willingness to continue the interaction with the agent. The model aims at detecting user's impression of the agent and producing appropriate agent's verbal and nonverbal behaviours in order to maintain a positive impression of warmth and competence. User's impressions are recognized using a machine learning approach with facial expressions (action units) which are important indicators of users' affective states and intentions. The agent adapts in real-time its verbal and nonverbal behaviour, with a reinforcement learning algorithm that takes user's impressions as reward to select the most appropriate combination of verbal and non-verbal behaviour to perform. A user study to test the model in a contextualized interaction with users is also presented. Our hypotheses are that users' ratings differs when the agents adapts its behaviour according to our reinforcement learning algorithm, compared to when the agent does not adapt its behaviour to user's reactions (i.e., when it randomly selects its behaviours). The study shows a general tendency for the agent to perform better when using our model than in the random condition. Significant results shows that user's ratings about agent's warmth are influenced by their a-priori about virtual characters, as well as that users' judged the agent as more competent when it adapted its behaviour compared to random condition.

With cybersecurity situation more and more complex, data-driven security has become indispensable. Numerous cybersecurity data exists in textual sources and data analysis is difficult for both security analyst and the machine. To convert the textual information into structured data for further automatic analysis, we extract cybersecurity-related entities and propose a self-attention-based neural network model for the named entity recognition in cybersecurity. Considering the single word feature not enough for identifying the entity, we introduce CNN to extract character feature which is then concatenated into the word feature. Then we add the self-attention mechanism based on the existing BiLSTM-CRF model. Finally, we evaluate the proposed model on the labelled dataset and obtain a better performance than the previous entity extraction model.

Regulatory efforts such as the General Data Protection Regulation (GDPR) embody a notion of privacy risk that is centered around the fundamental rights of data subjects. This is, however, a fundamentally different notion of privacy risk than the one commonly used in threat modeling which is largely agnostic of involved data subjects. This mismatch hampers the applicability of privacy threat modeling approaches such as LINDDUN in a Data Protection by Design (DPbD) context. In this paper, we present a data subject-aware privacy risk assessment model in specific support of privacy threat modeling activities. This model allows the threat modeler to draw upon a more holistic understanding of privacy risk while assessing the relevance of specific privacy threats to the system under design. Additionally, we propose a number of improvements to privacy threat modeling, such as enriching Data Flow Diagram (DFD) system models with appropriate risk inputs (e.g., information on data types and involved data subjects). Incorporation of these risk inputs in DFDs, in combination with a risk estimation approach using Monte Carlo simulations, leads to a more comprehensive assessment of privacy risk. The proposed risk model has been integrated in threat modeling tool prototype and validated in the context of a realistic eHealth application.

Vehicular networks are susceptible to variety of attacks such as denial of service (DoS) attack, sybil attack and false alert generation attack. Different cryptographic methods have been proposed to protect vehicular networks from these kind of attacks. However, cryptographic methods have been found to be less effective to protect from insider attacks which are generated within the vehicular network system. Misbehavior detection system is found to be more effective to detect and prevent insider attacks. In this paper, we propose a machine learning based misbehavior detection system which is trained using datasets generated through extensive simulation based on realistic vehicular network environment. The simulation results demonstrate that our proposed scheme outperforms previous methods in terms of accurately identifying various misbehavior.

Sharing personal data with other people or organizations over the web has become a common phenomena of our modern life. This type of sharing is usually managed by access control mechanisms that include access control model and policies. However, these models are designed from the organizational perspective and do not provide sufficient flexibility and control to the individuals. Therefore, individuals often cannot control sharing of their personal data based on their personal context. In addition, the existing context-aware access control models usually check contextual condition once at the beginning of the access and do not evaluate the context during an on-going access. Moreover, individuals do not have control to define how often they want to evaluate the context condition for an ongoing access. Wearable devices such as Fitbit and Apple Smart Watch have recently become increasingly popular. This has made it possible to gather an individual's real-time contextual information (e.g., location, blood-pressure etc.) which can be used to enforce continuous authorization to the individual's data resources. In this paper, we introduce a novel data sharing policy model for continuous authorization in subject-driven data sharing. A software prototype has been implemented employing a wearable device to demonstrate continuous authorization. Our continuous authorization framework provides more control to the individuals by enabling revocation of on-going access to shared data if the specified context condition becomes invalid.

The heart of the Internet of Things (IoT) is data. IoT services processes data from sensors that interface their physical surroundings, and from other software such as Internet weather databases. They produce data to control physical environments via actuators, and offer data to other services. More recently, service-centric designs for managing the IoT have been proposed. Data-centric or name-based communication architectures complement these developments very well. Especially for edge-based or site-local installations, data-centric Internet architectures can be implemented already today, as they do not require any changes at the core. We present the Virtual State Layer (VSL), a site-local data-centric architecture for the IoT. Special features of our solution are full separation of logic and data in IoT services, offering the data-centric VSL interface directly to developers, which significantly reduces the overall system complexity, explicit data modeling, a semantically-rich data item lookup, stream connections between services, and security-by-design. We evaluate our solution regarding usability, performance, scalability, resilience, energy efficiency, and security.

The Internet of Things (IoT) provides connectivity between heterogeneous devices in different applications, such as smart wildlife, supply chain and traffic management. Trust management system (TMS) assesses the trustworthiness of service with respect to its quality. Under different context information, a service provider may be trusted in one context but not in another. The existing context-aware trust models usually store trust values under different contexts and search the closest (to a given context) record to evaluate the trustworthiness of a service. However, it is not suitable for distributed resource-constrained IoT devices which have small memory and low power. Reputation systems are applied in many trust models where trustor obtains recommendations from others. In context-based trust evaluation, it requires interactive queries to find relevant information from remote devices. The communication overhead and energy consumption are issues in low power networks like 6LoWPAN. In this paper, we propose a new context-aware trust model for lightweight IoT devices. The proposed model provides a trustworthiness overview of a service provider without storing past behavior records, that is, constant size storage. The proposed model allows a trustor to decide the significance of context items. This could result in distinctive decisions under the same trustworthiness record. We also show the performance of the proposed model under different attacks.

Machine learning (ML) models are often trained using private datasets that are very expensive to collect, or highly sensitive, using large amounts of computing power. The models are commonly exposed either through online APIs, or used in hardware devices deployed in the field or given to the end users. This provides an incentive for adversaries to steal these ML models as a proxy for gathering datasets. While API-based model exfiltration has been studied before, the theft and protection of machine learning models on hardware devices have not been explored as of now. In this work, we examine this important aspect of the design and deployment of ML models. We illustrate how an attacker may acquire either the model or the model architecture through memory probing, side-channels, or crafted input attacks, and propose (1) power-efficient obfuscation as an alternative to encryption, and (2) timing side-channel countermeasures.