Generating Semantic Adversarial Examples via Feature Manipulation in Latent Space.

The susceptibility of deep neural networks (DNNs) to adversarial intrusions, exemplified by adversarial examples, is well-documented. Conventional attacks implement unstructured, pixel-wise perturbations to mislead classifiers, which often results in a noticeable departure from natural samples and lacks human-perceptible interpretability. In this work, we present an adversarial attack strategy that implements fine-granularity, semantic-meaning-oriented structural perturbations. Our proposed methodology manipulates the semantic attributes of images through the use of disentangled latent codes. We engineer adversarial perturbations by manipulating either a single latent code or a combination thereof. To this end, we propose two unsupervised semantic manipulation strategies: one based on vector-disentangled representation and the other on feature map-disentangled representation, taking into consideration the complexity of the latent codes and the smoothness of the reconstructed images. Our empirical evaluations, conducted extensively on real-world image data, showcase the potency of our attacks, particularly against black-box classifiers. Furthermore, we establish the existence of a universal semantic adversarial example that is agnostic to specific images.

Evaluation of Federated Learning in Phishing Email Detection.

The use of artificial intelligence (AI) to detect phishing emails is primarily dependent on large-scale centralized datasets, which has opened it up to a myriad of privacy, trust, and legal issues. Moreover, organizations have been loath to share emails, given the risk of leaking commercially sensitive information. Consequently, it has been difficult to obtain sufficient emails to train a global AI model efficiently. Accordingly, privacy-preserving distributed and collaborative machine learning, particularly federated learning (FL), is a desideratum. As it is already prevalent in the healthcare sector, questions remain regarding the effectiveness and efficacy of FL-based phishing detection within the context of multi-organization collaborations. To the best of our knowledge, the work herein was the first to investigate the use of FL in phishing email detection. This study focused on building upon a deep neural network model, particularly recurrent convolutional neural network (RNN) and bidirectional encoder representations from transformers (BERT), for phishing email detection. We analyzed the FL-entangled learning performance in various settings, including (i) a balanced and asymmetrical data distribution among organizations and (ii) scalability. Our results corroborated the comparable performance statistics of FL in phishing email detection to centralized learning for balanced datasets and low organizational counts. Moreover, we observed a variation in performance when increasing the organizational counts. For a fixed total email dataset, the global RNN-based model had a 1.8% accuracy decrease when the organizational counts were increased from 2 to 10. In contrast, BERT accuracy increased by 0.6% when increasing organizational counts from 2 to 5. However, if we increased the overall email dataset by introducing new organizations in the FL framework, the organizational level performance improved by achieving a faster convergence speed. In addition, FL suffered in its overall global model performance due to highly unstable outputs if the email dataset distribution was highly asymmetric.

Can the Policy of Increasing Retirement Age Raise Pension Revenue in China-A Case Study of Anhui Province.

With gradual progress in the medical field and the rising living standard of people, the life expectancy of people is gradually increasing. Unfortunately, this positive development contributes significantly to the aging of societies and creates huge challenges for pension systems. In order to mitigate the pressure on its pension system in the coming years, China is considering increasing the retirement age, just like many other countries. Based on the wage data of urban employees, pension revenue and expenditure data of employees in Anhui Province over the years, we constructed a model to predict average wages and forecast the revenue of the urban pension system from 2022 to 2032. We predicted the pension revenues by simulating an adjusted retirement age under two different schemes. The results of the study showed that the policies of appropriately increasing the retirement age can raise pension revenue. Compared with a one-step retirement age change scheme, a rolling retirement age change scheme that increases the retirement age by several months each year was found to be more suitable for the healthy development of the pension system.

Cyber Code Intelligence for Android Malware Detection.

Evolving Android malware poses a severe security threat to mobile users, and machine-learning (ML)-based defense techniques attract active research. Due to the lack of knowledge, many zero-day families' malware may remain undetected until the classifier gains specialized knowledge. The most existing ML-based methods will take a long time to learn new malware families in the latest malware family landscape. Existing ML-based Android malware detection and classification methods struggle with the fast evolution of the malware landscape, particularly in terms of the emergence of zero-day malware families and limited representation of single-view features. In this article, a new multiview feature intelligence (MFI) framework is developed to learn the representation of a targeted capability from known malware families for recognizing unknown and evolving malware with the same capability. The new framework performs reverse engineering to extract multiview heterogeneous features, including semantic string features, API call graph features, and smali opcode sequential features. It can learn the representation of a targeted capability from known malware families through a series of processes of feature analysis, selection, aggregation, and encoding, to detect unknown Android malware with shared target capability. We create a new dataset with ground-truth information regarding capability. Many experiments are conducted on the new dataset to evaluate the performance and effectiveness of the new method. The results demonstrate that the new method outperforms three state-of-the-art methods, including: 1) Drebin; 2) MaMaDroid; and 3) N -opcode, when detecting unknown Android malware with targeted capabilities.

A Comprehensive Survey on Community Detection With Deep Learning.

Detecting a community in a network is a matter of discerning the distinct features and connections of a group of members that are different from those in other communities. The ability to do this is of great significance in network analysis. However, beyond the classic spectral clustering and statistical inference methods, there have been significant developments with deep learning techniques for community detection in recent years--particularly when it comes to handling high-dimensional network data. Hence, a comprehensive review of the latest progress in community detection through deep learning is timely. To frame the survey, we have devised a new taxonomy covering different state-of-the-art methods, including deep learning models based on deep neural networks (DNNs), deep nonnegative matrix factorization, and deep sparse filtering. The main category, i.e., DNNs, is further divided into convolutional networks, graph attention networks, generative adversarial networks, and autoencoders. The popular benchmark datasets, evaluation metrics, and open-source implementations to address experimentation settings are also summarized. This is followed by a discussion on the practical applications of community detection in various domains. The survey concludes with suggestions of challenging topics that would make for fruitful future research directions in this fast-growing deep learning field.

User, Usage and Usability: Redefining Human Centric Cyber Security.

The effectiveness of cyber security measures are often questioned in the wake of hard hitting security events. Despite much work being done in the field of cyber security, most of the focus seems to be concentrated on system usage. In this paper, we survey advancements made in the development and design of the human centric cyber security domain. We explore the increasing complexity of cyber security with a wider perspective, defining user, usage and usability (3U's) as three essential components for cyber security consideration, and classify developmental efforts through existing research works based on the human centric security design, implementation and deployment of these components. Particularly, the focus is on studies that specifically illustrate the shift in paradigm from functional and usage centred cyber security, to user centred cyber security by considering the human aspects of users. The aim of this survey is to provide both users and system designers with insights into the workings and applications of human centric cyber security.

Daedalus: Breaking Nonmaximum Suppression in Object Detection via Adversarial Examples.

This article demonstrates that nonmaximum suppression (NMS), which is commonly used in object detection (OD) tasks to filter redundant detection results, is no longer secure. Considering that NMS has been an integral part of OD systems, thwarting the functionality of NMS can result in unexpected or even lethal consequences for such systems. In this article, an adversarial example attack that triggers malfunctioning of NMS in OD models is proposed. The attack, namely, Daedalus, compresses the dimensions of detection boxes to evade NMS. As a result, the final detection output contains extremely dense false positives. This can be fatal for many OD applications, such as autonomous vehicles and surveillance systems. The attack can be generalized to different OD models, such that the attack cripples various OD applications. Furthermore, a way of crafting robust adversarial examples is developed by using an ensemble of popular detection models as the substitutes. Considering the pervasive nature of model reuse in real-world OD scenarios, Daedalus examples crafted based on an ensemble of substitutes can launch attacks without knowing the parameters of the victim models. The experimental results demonstrate that the attack effectively stops NMS from filtering redundant bounding boxes. As the evaluation results suggest, Daedalus increases the false positive rate in detection results to 99.9% and reduces the mean average precision scores to 0, while maintaining a low cost of distortion on the original inputs. It also demonstrates that the attack can be practically launched against real-world OD systems via printed posters.

Can Cultural Intelligence Affect Employee's Innovative Behavior? Evidence From Chinese Migrant Workers in South Korea.

This empirical study explores the effect of cultural intelligence (CQ) on migrant workers' innovative behavior, as well as the mediating role of knowledge sharing on the CQ-innovative behavior relationship. Besides, it also examines the extent to which the mediating process is moderated by climate for inclusion. Using survey data collected from Chinese migrant workers and their supervisors working in South Korea (n = 386), migrant workers' CQ is found to positively impact their innovative behavior through enhanced knowledge sharing. However, it is observed that this indirect relationship is significant, only for migrant workers in a strong climate for inclusion. Thus, these findings reveal the underlying mediation and moderation mechanism and consequently unveil the important role of migrant workers' CQ in shaping their behavior. This study provides insightful and practical implications to a multicultural organization, where culturally diverse migrant workers work together.

Dual-Mediation Paths Linking Corporate Social Responsibility to Employee's Job Performance: A Multilevel Approach.

This study attempts to examine the direct impact of corporate social responsibility (CSR) initiatives on employees' job performance and the indirect relationships between CSR initiatives on employees' job performance via industrial relations climate and psychological contract fulfillment. Data were collected from 764 supervisor-subordinate dyads and 271 middle managers from 85 companies. Using a multilevel approach, the results showed that organizational-level CSR was positively related to employees' job performance. Moreover, the industrial relations climate and psychological contract fulfillment played mediating effects between CSR initiatives and job performance. This study provides novel theoretical evidence for why and how CSR initiatives improve job performance. Theoretical and practical implications for implementing CSR initiatives are discussed.

Impact of At-Home Telemonitoring on Health Services Expenditure and Hospital Admissions in Patients With Chronic Conditions: Before and After Control Intervention Analysis.

BACKGROUND: Telemonitoring is becoming increasingly important for the management of patients with chronic conditions, especially in countries with large distances such as Australia. However, despite large national investments in health information technology, little policy work has been undertaken in Australia in deploying telehealth in the home as a solution to the increasing demands and costs of managing chronic disease. OBJECTIVE: The objective of this trial was to evaluate the impact of introducing at-home telemonitoring to patients living with chronic conditions on health care expenditure, number of admissions to hospital, and length of stay (LOS). METHODS: A before and after control intervention analysis model was adopted whereby at each location patients were selected from a list of eligible patients living with a range of chronic conditions. Each test patient was case matched with at least one control patient. Test patients were supplied with a telehealth vital signs monitor and were remotely managed by a trained clinical care coordinator, while control patients continued to receive usual care. A total of 100 test patients and 137 control patients were analyzed. Primary health care benefits provided to Australian patients were investigated for the trial cohort. Time series data were analyzed using linear regression and analysis of covariance for a period of 3 years before the intervention and 1 year after. RESULTS: There were no significant differences between test and control patients at baseline. Test patients were monitored for an average of 276 days with 75% of patients monitored for more than 6 months. Test patients 1 year after the start of their intervention showed a 46.3% reduction in rate of predicted medical expenditure, a 25.5% reduction in the rate of predicted pharmaceutical expenditure, a 53.2% reduction in the rate of predicted unscheduled admission to hospital, a 67.9% reduction in the predicted rate of LOS when admitted to hospital, and a reduction in mortality of between 41.3% and 44.5% relative to control patients. Control patients did not demonstrate any significant change in their predicted trajectory for any of the above variables. CONCLUSIONS: At-home telemonitoring of chronically ill patients showed a statistically robust positive impact increasing over time on health care expenditure, number of admissions to hospital, and LOS as well as a reduction in mortality. TRIAL REGISTRATION: Retrospectively registered with the Australian and New Zealand Clinical Trial Registry ACTRN12613000635763; https://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=364030 (Archived by WebCite at http://www.webcitation.org/6sxqjkJHW).

Facilitating Secure Sharing of Personal Health Data in the Cloud.

BACKGROUND: Internet-based applications are providing new ways of promoting health and reducing the cost of care. Although data can be kept encrypted in servers, the user does not have the ability to decide whom the data are shared with. Technically this is linked to the problem of who owns the data encryption keys required to decrypt the data. Currently, cloud service providers, rather than users, have full rights to the key. In practical terms this makes the users lose full control over their data. Trust and uptake of these applications can be increased by allowing patients to feel in control of their data, generally stored in cloud-based services. OBJECTIVE: This paper addresses this security challenge by providing the user a way of controlling encryption keys independently of the cloud service provider. We provide a secure and usable system that enables a patient to share health information with doctors and specialists. METHODS: We contribute a secure protocol for patients to share their data with doctors and others on the cloud while keeping complete ownership. We developed a simple, stereotypical health application and carried out security tests, performance tests, and usability tests with both students and doctors (N=15). RESULTS: We developed the health application as an app for Android mobile phones. We carried out the usability tests on potential participants and medical professionals. Of 20 participants, 14 (70%) either agreed or strongly agreed that they felt safer using our system. Using mixed methods, we show that participants agreed that privacy and security of health data are important and that our system addresses these issues. CONCLUSIONS: We presented a security protocol that enables patients to securely share their eHealth data with doctors and nurses and developed a secure and usable system that enables patients to share mental health information with doctors.

Design of a multi-site multi-state clinical trial of home monitoring of chronic disease in the community in Australia.

BACKGROUND: Telehealth services based on at-home monitoring of vital signs and the administration of clinical questionnaires are being increasingly used to manage chronic disease in the community, but few statistically robust studies are available in Australia to evaluate a wide range of health and socio-economic outcomes. The objectives of this study are to use robust statistical methods to research the impact of at home telemonitoring on health care outcomes, acceptability of telemonitoring to patients, carers and clinicians and to identify workplace cultural factors and capacity for organisational change management that will impact on large scale national deployment of telehealth services. Additionally, to develop advanced modelling and data analytics tools to risk stratify patients on a daily basis to automatically identify exacerbations of their chronic conditions. METHODS/DESIGN: A clinical trial is proposed at five locations in five states and territories along the Eastern Seaboard of Australia. Each site will have 25 Test patients and 50 case matched control patients. All participants will be selected based on clinical criteria of at least two hospitalisations in the previous year or four or more admissions over the last five years for a range of one or more chronic conditions. Control patients are matched according to age, sex, major diagnosis and their Socio-Economic Indexes for Areas (SEIFA). The Trial Design is an Intervention control study based on the Before-After-Control-Impact (BACI) design. DISCUSSION: Our preliminary data indicates that most outcome variables before and after the intervention are not stationary, and accordingly we model this behaviour using linear mixed-effects (lme) models which can flexibly model within-group correlation often present in longitudinal data with repeated measures. We expect reduced incidence of unscheduled hospitalisation as well as improvement in the management of chronically ill patients, leading to better and more cost effective care. Advanced data analytics together with clinical decision support will allow telehealth to be deployed in very large numbers nationally without placing an excessive workload on the monitoring facility or the patient's own clinicians. TRIAL REGISTRATION: Registered with Australian New Zealand Clinical Trial Registry on 1st April 2013. Trial ID: ACTRN12613000635763.

Barriers for delivering telehealth in rural australia: a review based on Australian trials and studies.

BACKGROUND: Australians in rural and remote areas live with far poorer health outcomes than those in urban areas. Telehealth services have emerged as a promising solution to narrow this health gap, as they improve the level and diversity of health services delivery to rural and remote Australian communities. Although the benefits of telehealth services are well studied and understood, the uptake has been very slow. MATERIALS AND METHODS: To understand the underpinning issues, we conducted a literature review on barriers to telehealth adoption in rural and remote Australian communities, based on the published works of Australian clinical trials and studies. RESULTS: This article presents our findings using a comprehensive barrier matrix. This matrix is composed of four stakeholders (governments, technology developers and providers, health professionals, and patients) and five different categorizations of barriers (regulatory, financial, cultural, technological, and workforce). We explain each cell of the matrix (four stakeholders×five categories) and map the reported work into the matrix. CONCLUSIONS: Several exemplary barrier cases are also described to give more insights into the complexity and dilemma of adopting telehealth services. Finally, we outline recent technological advancements that have a great potential to overcome some of the identified barriers.

A framework for telehealth program evaluation.

Evaluating telehealth programs is a challenging task, yet it is the most sensible first step when embarking on a telehealth study. How can we frame and report on telehealth studies? What are the health services elements to select based on the application needs? What are the appropriate terms to use to refer to such elements? Various frameworks have been proposed in the literature to answer these questions, and each framework is defined by a set of properties covering different aspects of telehealth systems. The most common properties include application, technology, and functionality. With the proliferation of telehealth, it is important not only to understand these properties, but also to define new properties to account for a wider range of context of use and evaluation outcomes. This article presents a comprehensive framework for delivery design, implementation, and evaluation of telehealth services. We first survey existing frameworks proposed in the literature and then present our proposed comprehensive multidimensional framework for telehealth. Six key dimensions of the proposed framework include health domains, health services, delivery technologies, communication infrastructure, environment setting, and socioeconomic analysis. We define a set of example properties for each dimension. We then demonstrate how we have used our framework to evaluate telehealth programs in rural and remote Australia. A few major international studies have been also mapped to demonstrate the feasibility of the framework. The key characteristics of the framework are as follows: (a) loosely coupled and hence easy to use, (b) provides a basis for describing a wide range of telehealth programs, and (c) extensible to future developments and needs.