Biblio

Deep learning technologies, which are the key components of state-of-the-art Artificial Intelligence (AI) services, have shown great success in providing human-level capabilities for a variety of tasks, such as visual analysis, speech recognition, and natural language processing and etc. Building a production-level deep learning model is a non-trivial task, which requires a large amount of training data, powerful computing resources, and human expertises. Therefore, illegitimate reproducing, distribution, and the derivation of proprietary deep learning models can lead to copyright infringement and economic harm to model creators. Therefore, it is essential to devise a technique to protect the intellectual property of deep learning models and enable external verification of the model ownership. In this paper, we generalize the "digital watermarking'' concept from multimedia ownership verification to deep neural network (DNNs) models. We investigate three DNN-applicable watermark generation algorithms, propose a watermark implanting approach to infuse watermark into deep learning models, and design a remote verification mechanism to determine the model ownership. By extending the intrinsic generalization and memorization capabilities of deep neural networks, we enable the models to learn specially crafted watermarks at training and activate with pre-specified predictions when observing the watermark patterns at inference. We evaluate our approach with two image recognition benchmark datasets. Our framework accurately (100$\backslash$%) and quickly verifies the ownership of all the remotely deployed deep learning models without affecting the model accuracy for normal input data. In addition, the embedded watermarks in DNN models are robust and resilient to different counter-watermark mechanisms, such as fine-tuning, parameter pruning, and model inversion attacks.

Macroscopic hysteresis loops and microscopic magnetic moment distributions have been determined by 3-D model for Co nanowire with various easy axis deviations from applied field. It is found that both the coercivity and the remanence decrease monotonously with the increase of easy axis deviation as well as the maximum magnetic product, indicating the large impact of the easy axis orientation on the magnetic performance. Moreover, the calculated angular distributions and the evolution of magnetic moments have been shown to explain the magnetic reversal process. It is demonstrated that the large demagnetization field in the two ends of the nanowire makes the occurrence of reversal domain nucleation easier, hence the magnetic reversal. In addition, the magnetic reversal was illustrated in terms of the analysis of the energy evolution.

A Stoner-Wohlfarth-type model is used to demonstrate the effect of the buildup of magnetic charges near the grain boundaries of low anisotropy polycrystalline materials, revealed by measuring the magnetization during positive-field warming after negative-field cooling. The remnant magnetization after negative-field cooling has two different contributions. The temperature-dependent component is modeled as an assembly of particles with thermal relaxation. The temperature-independent component is modeled as an assembly of particles overcoming variable phenomenological energy barriers corresponding to the change in susceptibility when the anisotropy constant changes its sign. The model is applicable to soft-magnetic materials where the buildup of the magnetic charges near the grain boundaries creates demagnetizing fields opposing, and comparable in magnitude to, the anisotropy field. The results of the model are in qualitative agreement with published data revealing the magneto-thermal characteristics of polycrystalline gadolinium.

Network security and data confidentiality of transmitted information are among the non-functional requirements of industrial wireless sensor networks (IWSNs) in addition to latency, reliability and energy efficiency requirements. Physical layer security techniques are promising solutions to assist cryptographic methods in the presence of an eavesdropper in IWSN setups. In this paper, we propose a physical layer security scheme, which is based on both insertion of an random error vector to forward error correction (FEC) codewords and transmission over decentralized relay nodes. Reed-Solomon and Golay codes are selected as FEC coding schemes and the security performance of the proposed model is evaluated with the aid of decoding error probability of an eavesdropper. The results show that security level is highly based on the location of the eavesdropper and secure communication can be achieved when some of channels between eavesdropper and relay nodes are significantly noisier.

As the Industrial Internet of Things (IIot) becomes more prevalent in critical application domains, ensuring security and resilience in the face of cyber-attacks is becoming an issue of paramount importance. Cyber-attacks against critical infrastructures, for example, against smart water-distribution and transportation systems, pose serious threats to public health and safety. Owing to the severity of these threats, a variety of security techniques are available. However, no single technique can address the whole spectrum of cyber-attacks that may be launched by a determined and resourceful attacker. In light of this, we consider a multi-pronged approach for designing secure and resilient IIoT systems, which integrates redundancy, diversity, and hardening techniques. We introduce a framework for quantifying cyber-security risks and optimizing IIoT design by determining security investments in redundancy, diversity, and hardening. To demonstrate the applicability of our framework, we present a case study in water-distribution systems. Our numerical evaluation shows that integrating redundancy, diversity, and hardening can lead to reduced security risk at the same cost.

This is a full paper for innovate practice. Building a private cloud or using a public cloud is now feasible at many institutions. This paper presents the innovative design of cloudbased labs and programming assignments for a networking course and a cybersecurity course, and our experiences of innovatively using the private cloud at our institution to support these learning activities. It is shown by the instructor's observations and student survey data that our approach benefits learning and teaching. This approach makes it possible and secure to develop some learning activities that otherwise would not be allowed on physical servers. It enables the instructor to support students' desire of developing programs in their preferred programming languages. It allows students to debug and test their programs on the same platform to be used by the instructor for testing and grading. The instructor does not need to spend extra time administrating the computing environments. A majority (88% or more) of the students agree that working on those learning activities in the private cloud not only helps them achieve the course learning objectives, but also prepares them for their future careers.

This Innovative Practice Work in Progress paper makes the case for using concept inventories in cybersecurity education and presents an example of the development of a concept inventory in the field of secure programming. The secure programming concept inventory is being developed by a team of researchers from four universities. We used a Delphi study to define the content area to be covered by the concept inventory. Participants in the Delphi study included ten experts from academia, government, and industry. Based on the results, we constructed a concept map of secure programming concepts. We then compared this concept map to the Joint Task Force on Cybersecurity Education Curriculum 2017 guidelines to ensure complete coverage of secure programming concepts. Our mapping indicates a substantial match between the concept map and those guidelines.

This Innovate Practice Work in Progress paper is about education on Cybersecurity, which is essential in training of innovative talents in the era of the Internet. Besides knowledge and skills, it is important as well to enhance the students' awareness of cybersecurity in daily life. Considering that contactless smart cards are common and widely used in various areas, one basic and two advanced contactless smart card experiments were designed innovatively and assigned to junior students in 3-people groups in an introductory cybersecurity summer course. The experimental principles, facilities, contents and arrangement are introduced successively. Classroom tests were managed before and after the experiments, and a box and whisker plot is used to describe the distributions of the scores in both tests. The experimental output and student feedback implied the learning objectives were achieved through the problem-based, active and group learning experience during the experiments.

This Work-In-Progress Paper for the Innovative Practice Category presents a novel experiment in active learning of cybersecurity. We introduced a new workshop on hacking for an existing science-popularizing program at our university. The workshop participants, 28 teenagers, played a cybersecurity game designed for training undergraduates and professionals in penetration testing. Unlike in learning environments that are simplified for young learners, the game features a realistic virtual network infrastructure. This allows exploring security tools in an authentic scenario, which is complemented by a background story. Our research aim is to examine how young players approach using cybersecurity tools by interacting with the professional game. A preliminary analysis of the game session showed several challenges that the workshop participants faced. Nevertheless, they reported learning about security tools and exploits, and 61% of them reported wanting to learn more about cybersecurity after the workshop. Our results support the notion that young learners should be allowed more hands-on experience with security topics, both in formal education and informal extracurricular events.

This Research Work in Progress paper presents a study on improving student learning performance in a virtual hands-on lab system in cybersecurity education. As the demand for cybersecurity-trained professionals rapidly increasing, virtual hands-on lab systems have been introduced into cybersecurity education as a tool to enhance students' learning. To improve learning in a virtual hands-on lab system, instructors need to understand: what learning activities are associated with students' learning performance in this system? What relationship exists between different learning activities? What instructors can do to improve learning outcomes in this system? However, few of these questions has been studied for using virtual hands-on lab in cybersecurity education. In this research, we present our recent findings by identifying that two learning activities are positively associated with students' learning performance. Notably, the learning activity of reading lab materials (p \textbackslashtextless; 0:01) plays a more significant role in hands-on learning than the learning activity of working on lab tasks (p \textbackslashtextless; 0:05) in cybersecurity education.In addition, a student, who spends longer time on reading lab materials, may work longer time on lab tasks (p \textbackslashtextless; 0:01).

This research-to-practice work-in-progress addresses a new approach to cybersecurity education. The cyber security skills shortage is reaching prevalent proportions. The consensus in the STEM community is that the problem begins at k-12 schools with too few students interested in STEM subjects. One way to ensure a larger pipeline in cybersecurity is to train more high school teachers to not only teach cybersecurity in their schools or integrate cybersecurity concepts in their classrooms but also to promote IT security as an attractive career path. The proposed research will result in developing a unique and novel curriculum and scalable program in the area of cybersecurity and a set of powerful tools for a fun learning experience in cybersecurity education. In this project, we are focusing on the potential to advance research agendas in cybersecurity and train the future generation with cybersecurity skills and answer fundamental research questions that still exist in the blended learning methodologies for cybersecurity education and assessment. Leadership and entrepreneurship skills are also added to the mix to prepare students for real-world problems. Delivery methods, timing, format, pacing and outcomes alignment will all be assessed to provide a baseline for future research and additional synergy and integration with existing cybersecurity programs to expand or leverage for new cybersecurity and STEM educational research. This is a new model for cybersecurity education, leadership, and entrepreneurship and there is a possibility of a significant leap towards a more advanced cybersecurity educational methodology using this model. The project will also provide a prototype for innovation coupled with character-building and ethical leadership.

This innovative practice paper considers the heightening awareness of the need for cybersecurity programs in light of several well publicized cyber-attacks in recent years. An examination of the academic job market reveals that a significant number of institutions are looking to hire new faculty in the area of cybersecurity. Additionally, a growing number of universities are starting to offer courses, certifications and degrees in cybersecurity. Other recent activity includes the development of a model cybersecurity curriculum and the creation of a program accreditation criteria for cybersecurity through ABET. This sudden and significant growth in demand for cybersecurity expertise has some similarities to the significant demand for networking faculty that Computer Science programs experienced in the late 1980s as a result of the rise of the Internet. This paper examines the resources necessary to respond to the demand for cybersecurity courses and programs and draws some parallels and distinctions to the demand for networking faculty over 25 years ago. Faculty and administration are faced with a plethora of questions to answer as they approach this problem: What degree and courses to offer, what certifications to consider, which curriculum to incorporate and how to deliver the material (online, faceto-face, or something in-between)? However, the most pressing question in today's fiscal climate in higher education is: what resources will it take to deliver a cybersecurity program?

This Innovate Practice full paper presents a cloud-based personalized learning lab platform. Personalized learning is gaining popularity in online computer science education due to its characteristics of pacing the learning progress and adapting the instructional approach to each individual learner from a diverse background. Among various instructional methods in computer science education, hands-on labs have unique requirements of understanding learner's behavior and assessing learner's performance for personalization. However, it is rarely addressed in existing research. In this paper, we propose a personalized learning platform called ThoTh Lab specifically designed for computer science hands-on labs in a cloud environment. ThoTh Lab can identify the learning style from student activities and adapt learning material accordingly. With the awareness of student learning styles, instructors are able to use techniques more suitable for the specific student, and hence, improve the speed and quality of the learning process. With that in mind, ThoTh Lab also provides student performance prediction, which allows the instructors to change the learning progress and take other measurements to help the students timely. For example, instructors may provide more detailed instructions to help slow starters, while assigning more challenging labs to those quick learners in the same class. To evaluate ThoTh Lab, we conducted an experiment and collected data from an upper-division cybersecurity class for undergraduate students at Arizona State University in the US. The results show that ThoTh Lab can identify learning style with reasonable accuracy. By leveraging the personalized lab platform for a senior level cybersecurity course, our lab-use study also shows that the presented solution improves students engagement with better understanding of lab assignments, spending more effort on hands-on projects, and thus greatly enhancing learning outcomes.

This Research to Practice Full Paper presents a new methodology in cybersecurity education. In the context of the cybersecurity profession, the `isolation problem' refers to the observed isolation of different knowledge units, as well as the isolation of technical and business perspectives. Due to limitations in existing cybersecurity education, professionals entering the field are often trapped in microscopic perspectives, and struggle to extend their findings to grasp the big picture in a target network scenario. Guided by a previous developed and published framework named “cross-layer situation knowledge reference model” (SKRM), which delivers comprehensive level big picture situation awareness, our new methodology targets at developing suites of teaching modules to address the above issues. The modules, featuring interactive hands-on labs that emulate real-world multiple-step attacks, will help students form a knowledge network instead of isolated conceptual knowledge units. Students will not just be required to leverage various techniques/tools to analyze breakpoints and complete individual modules; they will be required to connect logically the outputs of these techniques/tools to infer the ground truth and gain big picture awareness of the cyber situation. The modules will be able to be used separately or as a whole in a typical network security course.

Deep neural networks (DNNs) are known vulnerable to adversarial attacks. That is, adversarial examples, obtained by adding delicately crafted distortions onto original legal inputs, can mislead a DNN to classify them as any target labels. In a successful adversarial attack, the targeted mis-classification should be achieved with the minimal distortion added. In the literature, the added distortions are usually measured by \$L\_0\$, \$L\_1\$, \$L\_2\$, and \$L\_$\backslash$infty \$ norms, namely, L\_0, L\_1, L\_2, and L\_$ınfty$ attacks, respectively. However, there lacks a versatile framework for all types of adversarial attacks. This work for the first time unifies the methods of generating adversarial examples by leveraging ADMM (Alternating Direction Method of Multipliers), an operator splitting optimization approach, such that \$L\_0\$, \$L\_1\$, \$L\_2\$, and \$L\_$\backslash$infty \$ attacks can be effectively implemented by this general framework with little modifications. Comparing with the state-of-the-art attacks in each category, our ADMM-based attacks are so far the strongest, achieving both the 100% attack success rate and the minimal distortion.

The traditional security system requires a lot of manpower, and the wireless security system has been developed to reduce costs. However, for wireless systems, stability and reliability are important system indicators. In order to effectively improve these two indicators, we have imported butterfly network coding algorithm into the wireless sensing network. Because this algorithm enables each node to play multiple roles, such as routing, encoding, decoding, sending and receiving, it can also improve the throughput of network transmission, and effectively improve the stability and reliability of the wireless security system. This paper used the Wi-Fi module to implement the butterfly network coding algorithm, and is actually installed in the building. The basis for transmission and reception of all nodes in the network is received signal strength indication (RSSI). On the other hand, this is an IoT system for security monitoring.

As the era of cloud technology arises, more and more people are beginning to migrate their applications and personal data to the cloud. This makes web-based applications an attractive target for cyber-attacks. As a result, web-based applications now need more protections than ever. However, current anomaly-based web attack detection approaches face the difficulties like unsatisfying accuracy and lack of generalization. And the rule-based web attack detection can hardly fight unknown attacks and is relatively easy to bypass. Therefore, we propose a novel deep learning approach to detect anomalous requests. Our approach is to first train two Recurrent Neural Networks (RNNs) with the complicated recurrent unit (LSTM unit or GRU unit) to learn the normal request patterns using only normal requests unsupervisedly and then supervisedly train a neural network classifier which takes the output of RNNs as the input to discriminate between anomalous and normal requests. We tested our model on two datasets and the results showed that our model was competitive with the state-of-the-art. Our approach frees us from feature selection. Also to the best of our knowledge, this is the first time that the RNN is applied on anomaly-based web attack detection systems.

Following the recent adoption of deep neural networks (DNN) accross a wide range of applications, adversarial attacks against these models have proven to be an indisputable threat. Adversarial samples are crafted with a deliberate intention of undermining a system. In the case of DNNs, the lack of better understanding of their working has prevented the development of efficient defenses. In this paper, we propose a new defense method based on practical observations which is easy to integrate into models and performs better than state-of-the-art defenses. Our proposed solution is meant to reinforce the structure of a DNN, making its prediction more stable and less likely to be fooled by adversarial samples. We conduct an extensive experimental study proving the efficiency of our method against multiple attacks, comparing it to numerous defenses, both in white-box and black-box setups. Additionally, the implementation of our method brings almost no overhead to the training procedure, while maintaining the prediction performance of the original model on clean samples.

In this paper we presents the notion of key-rotatable and security-updatable homomorphic encryption (KR-SU-HE) scheme, which is a class of public-key homomorphic encryption in which the keys and the security of any ciphertext can be rotated and updated while still keeping the underlying plaintext intact and unrevealed. We formalise syntax and security notions for KR-SU-HE schemes and then build a concrete scheme based on the Learning With Errors assumption. We then perform testing implementation to show that our proposed scheme is efficiently practical.

Proxy Re-encryption (PRE) is a useful cryptographic structure who enables a semi-trusted proxy to convert a ciphertext for Alice into a ciphertext for Bob without seeing the corresponding plaintext. Although there are many PRE schemes in recent years, few of them are set up based on lattice. Not only this, these lattice-based PRE schemes are all more complicated than the traditional PRE schemes. In this paper, through the study of the common lattice problems such as the Small integer solution (SIS) and the Learning with Errors (LWE), we analyze the feasibility of efficient lattice-based PRE scheme combined with the previous results. Finally, we propose an efficient lattice-based PRE scheme L-PRE without losing the hardness of lattice problems.

In order to improve the efficiency of the existing homomorphic encryption method, based on the DGHV scheme, an improved fully homomorphic scheme over the integer is proposed. Under the premise of ensuring data owner and user data security, the scheme supports the addition and multiplication operations of ciphertext, and ensures faster execution efficiency and meets the security requirements of cloud computing. Security analysis shows that our scheme is safe. Performance assessment demonstrates that our scheme can more efficiently implement data than DGHV scheme.

Searchable Symmetric Encryption is a mechanism that facilitates search over encrypted data that are outsourced to an untrusted server. SSE schemes are practical as they trade nicely security for efficiency. However, the supported functionalities are mainly limited to single keyword queries. In this paper, we present a new efficient SSE scheme, called REX, that supports range queries. REX is a no interactive (single round) and response-hiding scheme. It has optimal communication and search computation complexity, while it is much more secure than traditional Order Preserving Encryption based range SSE schemes.

Real-time editing tools like Google Docs, Microsoft Office Online, or Etherpad have changed the way of collaboration. Many of these tools are based on Operational Transforms (OT), which guarantee that the views of different clients onto a document remain consistent over time. Usually, documents and operations are exposed to the server in plaintext – and thus to administrators, governments, and potentially cyber criminals. Therefore, it is highly desirable to work collaboratively on encrypted documents. Previous implementations do not unleash the full potential of this idea: They either require large storage, network, and computation overhead, are not real-time collaborative, or do not take the structure of the document into account. The latter simplifies the approach since only OT algorithms for byte sequences are required, but the resulting ciphertexts are almost four times the size of the corresponding plaintexts. We present SECRET, the first secure, efficient, and collaborative real-time editor. In contrast to all previous works, SECRET is the first tool that (1.) allows the encryption of whole documents or arbitrary sub-parts thereof, (2.) uses a novel combination of tree-based OT with a structure preserving encryption, and (3.) requires only a modern browser without any extra software installation or browser extension. We evaluate our implementation and show that its encryption overhead is three times smaller in comparison to all previous approaches. SECRET can even be used by multiple users in a low-bandwidth scenario. The source code of SECRET is published on GitHub as an open-source project:https://github.com/RUB-NDS/SECRET/

IoT applications often utilize the cloud to store and provide ubiquitous access to collected data. This naturally facilitates data sharing with third-party services and other users, but bears privacy risks, due to data breaches or unauthorized trades with user data. To address these concerns, we present Pilatus, a data protection platform where the cloud stores only encrypted data, yet is still able to process certain queries (e.g., range, sum). More importantly, Pilatus features a novel encrypted data sharing scheme based on re-encryption, with revocation capabilities and in situ key-update. Our solution includes a suite of novel techniques that enable efficient partially homomorphic encryption, decryption, and sharing. We present performance optimizations that render these cryptographic tools practical for mobile platforms. We implement a prototype of Pilatus and evaluate it thoroughly. Our optimizations achieve a performance gain within one order of magnitude compared to state-of-the-art realizations; mobile devices can decrypt hundreds of data points in a few hundred milliseconds. Moreover, we discuss practical considerations through two example mobile applications (Fitbit and Ava) that run Pilatus on real-world data.

In this paper, we focus on fraud detection on a signed graph with only a small set of labeled training data. We propose a novel framework that combines deep neural networks and spectral graph analysis. In particular, we use the node projection (called as spectral coordinate) in the low dimensional spectral space of the graph's adjacency matrix as the input of deep neural networks. Spectral coordinates in the spectral space capture the most useful topology information of the network. Due to the small dimension of spectral coordinates (compared with the dimension of the adjacency matrix derived from a graph), training deep neural networks becomes feasible. We develop and evaluate two neural networks, deep autoencoder and convolutional neural network, in our fraud detection framework. Experimental results on a real signed graph show that our spectrum based deep neural networks are effective in fraud detection.