Search

In our increasingly connected world, data comes from many different sources, in many different forms, and is noisy, complex, and structured. To confront modern data, we need to embrace the structure inherent in the data and in the predictions. Statistical relational learning (SRL) is a subfield of machine learning that provides an effective means of approaching this problem of structured prediction. SRL frameworks use weighted logical and arithmetic expressions to easily create probabilistic graphical models (PGMs) to jointly reason over interdependent data. However, despite being well suited for modern, interconnected data, SRL faces several challenges that keep it from becoming practical and widely used in the machine learning community. In this dissertation, I address four pillars of practicality for SRL systems: scalability, expressivity, model adaptability, and usability. My work in this dissertation uses and extends Probabilistic Soft Logic (PSL), a state-of-the-art open-source SRL framework.Scalability in SRL systems is essential for using large datasets and complex models. Because of the complex nature of interconnected data, models can easily outgrow available memory spaces. To address scalability for SRL, I developed methods that more efficiently and intelligently instantiate PGMs from templates and data. I also developed fixed-memory inference methods that can perform inference on very large models without requiring a proportional amount of memory.Expressivity allows SRL systems to represent many different problems and data patterns. Because SRL uses logical and arithmetic expressions to represent structured dependencies, SRL frameworks need to be able to express more than just what is represented by feature vectors. To address expressivity for SRL, I created a system to incorporate neural models with structured SRL inference, and expanded the interpretation of PSL weight hyperparameters to include additional types of distributions.Model adaptability is the ability of SRL frameworks to handle models that change. A changing model can be as simple as a model that has its hyperparameters updated, or as complex as a model that changes its structure over time. To address model adaptability for SRL, I developed new weight learning approaches for PSL, and created a system for generalized online inference in PSL.Usability make SRL frameworks easy for people to use. Because of the need to model structural dependencies, SRL frameworks are often harder to use when compared to more common machine learning libraries. To address usability for SRL, I have created a new SRL framework that removes the tight coupling between the different components of the SRL pipeline that is seen in other SRL frameworks and allows the the recreation of exiting SRL frameworks along with the creation of new SRL frameworks using the same common runtime. Additionally, I developed a visual model inspector for analyzing and debugging PSL models.

Using standard handheld interfaces for VR locomotion may not provide a believable self-motion experience and can contribute to unwanted side effects such as motion sickness, disorientation, or increased cognitive load. This paper demonstrates how using a seated leaning-based locomotion interface --HeadJoystick-- in VR ground-based navigation affects user experience, usability, and performance. In three within-subject studies, we compared controller (touchpad/thumbstick) with a more embodied interface ("HeadJoystick") where users moved their head and/or leaned in the direction of desired locomotion. In both conditions, users sat on a regular office chair and used it to control virtual rotations. In the first study, 24 participants used HeadJoystick versus Controller in three complementary tasks including reach-the-target, follow-the-path, and racing (dynamic obstacle avoidance). In the second study, 18 participants repeatedly used HeadJoystick versus Controller (8 one-minute trials each) in a reach-the-target task. To evaluate potential benefits of different brake mechanisms, in the third study 18 participants were asked to stop within each target area for one second. All three studies consistently showed advantages of HeadJoystick over Controller: we observed improved performance in all tasks, as well as higher user ratings for enjoyment, spatial presence, immersion, vection intensity, usability, ease of learning, ease of use, and rated potential for daily and long-term use, while reducing motion sickness and task load. Overall, our results suggest that leaning-based interfaces such as HeadJoystick provide an interesting and more embodied alternative to handheld interfaces in driving, reach-the-target, and follow-the-path tasks, and potentially a wider range of scenarios.

Spreadsheet-based tools provide a simple yet effective way of calculating values, which makes them the number-one choice for building and formalizing simple models for budget planning and many other applications.A cell in a spreadsheet holds one specific value and gives a discrete, overprecise view of the underlying model. Therefore, spreadsheets are of limited use when investigating the immanent uncertainties of such models and answering what-if questions. Existing extensions typically require a complex modeling process that cannot be smoothly embedded in a tabular layout.In Fuzzy Spreadsheet, a cell can hold and display the distribution of values. This integrated uncertainty handling immediately conveys sensitivity and robustness information. The fuzzification of the cells enables calculations not only with precise values but with distributions, and probabilities. We conservatively added and carefully crafted visuals to maintain the look and feel of a traditional spreadsheet while facilitating what-if analyses.Given a user-specified reference cell, Fuzzy Spreadsheet automatically extracts and visualizes contextually relevant information, such as impact, uncertainty, and degree of neighborhood, for the selected and related cells.To evaluate its usability and the perceived mental effort required, we conducted a user study.The results show that our approach outperforms traditional spreadsheets in terms of answer correctness, response time, and perceived mental effort for almost all tasks tested.

Secure and usable user authentication is the first line of defense against cyber attacks on smart end-user devices. Advanced hacking techniques pose severe threats to the traditional authentication systems based on the password/PIN/fingerprint. We propose MagAuth, a secure and usable two-factor authentication scheme with commercial off-the-shelf (COTS) wrist wearables with magnetic strap bands to enhance the security and usability of password-based authentication for mobile touchscreen devices. In MagAuth, a user enrolls a self-chosen unlock pattern or touch gesture into his touchscreen device by performing it with the same hand the magnetic wrist wearable is on. The chosen unlock pattern or touch gesture serves as the first authentication factor, and the user's behavioral features manifested in the magnetic field changes during his finger movement correspond to the second factor. The user can unlock his touchscreen device only when both authentication factors can be validated. Comprehensive user experiments confirm the high security and usability of MagAuth. In particular, MagAuth achieves an average true-positive rate up to 96.3% and a false-positive rate no larger than 8.4%. Moreover, we show that MagAuth is highly resilient to various attacks.

As smart devices are increasingly getting deployed in distinct scenarios it is important to examine how the various demands of these practical uses will affect the dynamics of protection. This study provides an outline of the connection among the various security threats, specifications, implementations, and network safety as well as an overview. In addition, a few of the device communication services are often overviewed, evaluating the protection mechanisms. The Internet of Things (IoT) has been a focal point in the previous few years. On analysis, here are several kinds of problems and difficulties with the tremendous ability of the IoT. For IoT technologies, applications, and networks, cybersecurity is one of the crucial challenges. This study discusses the work advancement of IoT to examine every main part of IoT, and finds how some safety problems and concerns have to be recognized and discusses them momentarily. To secure information privacy, professional conduct, honesty, encryption, intrusion detection, and capability to recognize as well as versatility, interoperability, and usability, reliable and usable IoT protection is needed to be brought into account. In terms of certain realities, new IoT approaches from the scientific, educational and industrial sectors are presented and addressed by analyzing a few of the current study in the IoT field Depending on the results of this report, it is important to develop and implement suitable IoT applications that can ensure integrity, security, and honesty in interconnected conditions.

In multiple coordinated views (MCVs), visualizations across views update their content in response to users' interactions in other views. Interactive systems provide direct manipulation to create coordination between views, but are restricted to limited types of predefined templates. By contrast, textual specification languages enable flexible coordination but expose technical burden. To bridge the gap, we contribute Nebula, a grammar based on natural language for coordinating visualizations in MCVs. The grammar design is informed by a novel framework based on a systematic review of 176 coordinations from existing theories and applications, which describes coordination by demonstration, i.e., how coordination is performed by users. With the framework, Nebula specification formalizes coordination as a composition of user- and coordination-triggered interactions in origin and destination views, respectively, along with potential data transformation between the interactions. We evaluate Nebula by demonstrating its expressiveness with a gallery of diverse examples and analyzing its usability on cognitive dimensions.

Android overlay enables one app to draw over other apps by creating an extra View layer atop the host View, which nevertheless can be exploited by malicious apps (malware) to attack users. To combat this threat, prior countermeasures concentrate on restricting the capabilities of overlays at the OS level while sacrificing overlays usability; recently, the overlay mechanism has been substantially updated to prevent a variety of attacks, which however can still be evaded by considerable adversaries. To address these shortcomings, a more pragmatic approach is to enable early detection of overlay-based malware during the app market review process, so that all the capabilities of overlays can stay unchanged. For this purpose, in this paper we first conduct a large-scale comparative study of overlay characteristics in benign and malicious apps, and then implement the OverlayChecker system to automatically detect overlay-based malware for one of the worlds largest Android app stores. In particular, we have made systematic efforts in feature engineering, UI exploration, emulation architecture, and run-time environment, thus maintaining high detection accuracy (97% precision and 97% recall) and short per-app scan time (1.7 minutes) with only two commodity servers, under an intensive workload of 10K newly submitted apps per day.

The Internet of Things (IoT) has been growing over the past few years due to its flexibility and ease of use in real-time applications. The IoT’s foremost task is ensuring that there is proper communication between different types of applications and devices, and that the application layer protocols fulfill this necessity. However, as the number of applications grows, it is necessary to modify or enhance the application layer protocols according to specific IoT applications, allowing specific issues to be addressed, such as dynamic adaption to network conditions and interoperability. Recently, several IoT application layer protocols have been enhanced and modified according to application requirements. However, no existing survey articles have focused on these protocols. In this article, we survey traditional and recent advances in IoT application layer protocols, as well as relevant real-time applications and their adapted application layer protocols for improving performance. As changing the nature of protocols for each application is unrealistic, machine learning offers means of making protocols intelligent and able to adapt dynamically. In this context, we focus on providing open challenges to drive IoT application layer protocols in such a direction.

The increased availability of quantitative historical datasets has provided new research opportunities for multiple disciplines in social science. In this article, we work closely with the constructors of a new dataset, CGED-Q (China Government Employee Database-Qing), that records the career trajectories of over 340,000 government officials in the Qing bureaucracy in China from 1760 to 1912. We use these data to study career mobility from a historical perspective and understand social mobility and inequality. However, existing statistical approaches are inadequate for analyzing career mobility in this historical dataset with its fine-grained attributes and long time span, since they are mostly hypothesis-driven and require substantial effort. We propose CareerLens, an interactive visual analytics system for assisting experts in exploring, understanding, and reasoning from historical career data. With CareerLens, experts examine mobility patterns in three levels-of-detail, namely, the macro-level providing a summary of overall mobility, the meso-level extracting latent group mobility patterns, and the micro-level revealing social relationships of individuals. We demonstrate the effectiveness and usability of CareerLens through two case studies and receive encouraging feedback from follow-up interviews with domain experts.

In this paper, we provide the design and implementation of UOIT, a text entry method optimized for smartwatches. UOIT uses only one page where a user can see and tap directly for entry without any additional actions, such as zoom-in/zoom-out and swipes, which are required in the existing entry methods. To fully utilize the constrained screen space and to address the "fat finger" problem, we use a technique called "drawing-like typing", which reduces the 26 small alphabetic keys into 13 large keys with a dual input property. To evaluate the performance of UOIT, we conducted two user studies while varying the learning period. In the short-term experiments (i.e., two days), we observed a fast learning curve of users when using the UIOT keyboard. Moreover, with the long-term experiments (i.e., a month), we show that users can type as fast as QWERTY keyboard but with much less errors. Moreover, UOIT outperforms the state-of-the-art keyboard in both speed and error rate.

Brain-computer interface (BCI) performance is often impacted due to the inherent non-stationarity in the recorded EEG signals coupled with a high variability across subjects. This study proposes a novel method using Logistic Regression with Tangent Space-based Transfer Learning (LR-TSTL) for motor imagery (MI)-based BCI classification problems. The single-trial covariance matrix (CM) features computed from the EEG signals are transformed into a Riemannian geometry frame and tangent space features are computed by considering the lower triangular matrix. These are then further classified using the logistic regression model to improve classification accuracy. The performance of LR-TSTL is tested on healthy subjects’ dataset as well as on stroke patients’ dataset. As compared to existing within-subject learning approaches the proposed method gave an equivalent or better performance in terms of average classification accuracy (78.95±11.68%), while applied as leave-one-out cross-subject learning for healthy subjects. Interestingly, for the patient dataset LR-TSTL significantly (p

Benefiting from the rapid development of communication technology and Internet of Things (IoT) devices, crowdsensing is on the rise. Sensor data from IoT devices can be requested for data analysis and utilization, however, the collected data of an object from multiple devices are usually different. Therefore, how to extract the most reliable data from numerous data has become an important topic, and truth discovery receives great attention. These collected data often contain personal sensitive information, if users’ privacy cannot be protected, many users are unwilling to contribute their data, and the usability of the published data will be greatly reduced. In this article, a robust privacy-preserving truth discovery scheme is proposed to simultaneously achieve the reliability and privacy of data. Specifically, the data are collected and encrypted before it is sent from the user. Compared with the existing works, there are two additional benefits, trusted third party and noncolluding platforms are not necessary anymore, hence the robustness is improved and single-point failure bottlenecks are eliminated. Besides, the proposed RPPTD is secure against many known attacks in open wireless networks, and the human-factor-aware differential aggregation attack. Finally, the performance evaluation indicates that our scheme is efficient and suitable for the practical environment.

Deep Learning (DL) techniques allow ones to train models from a dataset to solve tasks. DL has attracted much interest given its fancy performance and potential market value, while security issues are amongst the most colossal concerns. However, the DL models may be prone to the membership inference attack, where an attacker determines whether a given sample is from the training dataset. Efforts have been made to hinder the attack but unfortunately, they may lead to a major overhead or impaired usability. In this paper, we propose and implement DAMIA, leveraging Domain Adaptation (DA) as a defense aginist membership inference attacks. Our observation is that during the training process, DA obfuscates the dataset to be protected using another related dataset, and derives a model that underlyingly extracts the features from both datasets. Seeing that the model is obfuscated, membership inference fails, while the extracted features provide supports for usability. Extensive experiments have been conducted to validates our intuition. The model trained by DAMIA has a negligible footprint to the usability. Our experiment also excludes factors that may hinder the performance of DAMIA, providing a potential guideline to vendors and researchers to benefit from our solution in a timely manner.

Browser extensions are widely used nowadays to customize users' browsers with more functionalities, meanwhile introduce potential risks due to escalated privileges. Existing security mechanisms, such as Same Origin Policy and Content Security Policy, do not apply to browser extensions that can read and write on web applications at any time. In spite of the state-of-art industrial efforts that rely on centralized management to inspect and detect malicious behaviors massively, the detection-based method cannot analyze fast-evolving behaviors of malicious browser extensions. To this end, we adopt a novel approach to protect users from malicious browser extensions, where we consider the problem of malicious extensions on the side of web applications. From a high level point of view, web developers are allowed to specify sensitive parts in a web application by using our provided software enclave. With our proposed WebEnclave extension installed, when users visit a web application, sensitive information required for the web application to work normally is sealed into an isolated world locally that malicious extensions cannot access. Extensive evaluation of our built prototype shows it can effectively protect user secrets from malicious extensions with negligible performance overhead and usability inconvenience. We also publish source codes for public use.

Context based API recommendation is an important way to help developers find the needed APIs effectively and efficiently. For effective API recommendation, we need not only a joint view of both structural and textual code information, but also a holistic view of correlated API usage in control and data flow graph as a whole. Unfortunately, existing API recommendation methods exploit structural or textual code information separately. In this work, we propose a novel API recommendation approach called APIRec-CST (API Recommendation by Combining Structural and Textual code information). APIRec-CST is a deep learning model that combines the API usage with the text information in the source code based on an API Context Graph Network and a Code Token Network that simultaneously learn structural and textual features for API recommendation. We apply APIRec-CST to train a model for JDK library based on 1,914 open-source Java projects and evaluate the accuracy and MRR (Mean Reciprocal Rank) of API recommendation with another 6 open-source projects. The results show that our approach achieves respectively a top-1, top-5, top-10 accuracy and MRR of 60.3%, 81.5%, 87.7% and 69.4%, and significantly outperforms an existing graph-based statistical approach and a tree-based deep learning approach for API recommendation. A further analysis shows that textual code information makes sense and improves the accuracy and MRR. The sensitivity analysis shows that the top-k accuracy and MRR of APIRec-CST are insensitive to the number of APIs to be recommended in a hole. We also conduct a user study in which two groups of students are asked to finish 6 programming tasks with or without our APIRec-CST plugin. The results show that APIRec-CST can help the students to finish the tasks faster and more accurately and the feedback on the usability is overwhelmingly positive.

This case study explores concepts and methodologies in user interface (UI) and user experience (UX) design with a view to increasing information retention and memorisation through the inclusion of human-centred design principles. It focuses on the participants’ individual context, mental state and abilities. In emergencies, visual perception and situation awareness may be restricted due to the impact of sensory symptoms (panic, tunnel vision or limited motor skills), thus calling for a linear course of action. This applied research project, ‘Floodscape’, is a mobile application designed to educate its user about possible inundation zones resulting from a tsunami. Ongoing user engagement through interactive simulation is the prime focus of the initial (dormant) state of the app. In case of an actual tsunami the app registers the event and adapts its UI accordingly. Crucial life-saving wayfinding information will then be displayed in a contextual manner considering contrast, typography, limited user attention and ease of use.

Medical personnel usually write and design documents that inform physicians or patients about procedures or therapies. Document design, however, requires skills that are not normally applied, resulting in information that is often not used properly. This article describes a project developed by the Alberta Colorectal Cancer Screening Program. The goal was to help patients better prepare for their colonoscopies. The process started with an analysis of the existing documents, and the development of performance specifications based on the literature on legibility, reading comprehension, memorization and use of information, plain language, visual perception, page layout, and image use. The project included an iterative process of prototyping and testing that resulted in 23 design criteria. Each iteration was tested with users to ensure ease of use, completeness of information, and accuracy and clarity to facilitate adoption. The project helped reduce practice variation regarding bowel preparation in the province of Alberta, Canada. This project illustrates how information design can help healthcare organizations provide patient-centred care. Information design helps patients engage in their own caring process, by providing information that people can use, understand and apply. After 15 months of use, the document has been downloaded more than 48,000 times, suggesting a good physician reception.

Federated learning enables collaborative deep learning over multiple clients without sharing their local data, and it becomes increasingly popular due to the good balance between data privacy and model usability. Generally, it faces the heavy communication overhead when a large number of clients are involved and the low convergence rate incurred by non-IID data. However, few existing solutions cannot simultaneously address the communication and statistic challenges. In this paper, we propose a computation and communication efficient federated learning via adaptive sampling. By capturing different data distribution among clients, we utilize the concept of self-paced learning to adaptively adjust thresholds to filter training data for each client and also to select suitable clients to be involved in each global learning round. We prove its correctness through theoretical analysis and also evaluate its performance through experimental evaluations on real-world datasets. Detailed experimental results show that it can effectively reduce the communication cost while achieving the good trade-off between accuracy and efficiency.

Most robotic systems designed for mass manufacturing are optimized for a specific type of product. They generally lack the ability to adapt to low-volume customized products. In this article, we present a system based on a modular design for manufacturing personalized medical stent graft implants. The concept is based on learning-by-demonstration by integrating real-time 3-D vision, multirobot collaboration, and personalization to guide the robots to learn and execute tasks continuously with adaptation to different implant geometry. The system is optimized to generate customized and collision-free paths for efficient object manipulation and task completion. We show that the system is generalizable to different stent graft designs and the proposed multirobot system can seamlessly work together with high efficiency without collisions. The results have also suggested its usability for other manipulation tasks, especially for flexible production of customized products where bimanual or multirobot cooperation is required.

Facial authentication has become more and more popular on personal devices. Due to the ease of use, it has great potential to be widely deployed for web-service authentication in the near future whereby people can easily log on to online accounts from different devices without memorizing lengthy passwords. However, the growing number of attacks on machine learning especially the Deep Neural Networks (DNN) which is commonly used for facial recognition, imposes big challenges on the successful roll-out of such web-service face authentication. Although there have been studies on defending some machine learning attacks, we are not aware of any specific effort devoted to the web-service facial authentication setting. In this paper, we first demonstrate a new data poisoning attack which does not require to have any knowledge of the server-side and just needs a handful of malicious photo injections to enable an attacker to easily impersonate the victim in the existing facial authentication systems. We then propose a novel defensive approach called DEFEAT that leverages deep learning techniques to automatically detect such attacks. We have conducted extensive experiments on real datasets and our experimental results show that our defensive approach achieves more than 90% detection accuracy.

Crowdsourcing has attracted widespread attention in recent years and developed into various applications. An indispensable service of crowdsourcing systems is task recommendation, which means tasks should be accurately recommended to the workers with aligned interests. However, existing systems rely on their separate servers to conduct recommendation services, resulting in computing resources locked inside each isolated system. Moreover, due to the wide attacking surfaces of traditional centralized servers setting, existing systems are subject to single points of failure or malicious data breaches. Therefore, failure to address these inherent limitations properly will hinder the wide adoption of crowdsourcing. In this paper, we propose and implement FedCrowd, the first federated and privacy-preserving crowdsourcing platform by using blockchain technology. Our main idea is to employ the smart contract as a trusted platform for systems to release encrypted tasks, and carefully craft matching protocols to enable efficient task recommendations in the ciphertext domain. Our task-matching protocols are highly customized for the decentralized settings, where users can securely perform keyword and range-based queries over federated task indexes without sharing secret keys. We formally analyze the security strengths and complete the prototype implementation on Ethereum. Experiment results demonstrate the feasibility and usability of the FedCrowd platform.

Argumentation Maps support participants in geographically referenced debates as they occur, for example, as part of urban planning processes. In a quasi-naturalistic case study, 11 student participants discussed planning issues on the University of Toronto downtown campus. The analysis of this case study focuses on general usability aspects of an Argumentation Map prototype, such as cost of entry, efficiency, interactivity, and connectivity. By applying usability analysis methods from the field of human-computer interaction, we evaluate the learnability, memorability, and user satisfaction with this tool’s functionality. Our findings indicate that the participants were generally satisfied, but we include specific suggestions for improving the functionality of Argumentation Maps, e.g., with respect to map navigation, display of discussion contributions, and online status of participants. On a more general level, this case study contributes to the methods spectrum of research into participatory spatial decision support systems as an example of user testing in a realistic decision-making context.

Modality translation grants diagnostic value to wearable devices by translating signals collected from low-power sensors to their highly-interpretable counterparts that are more familiar to healthcare providers. For instance, bio-impedance (Bio-Z) is a conveniently collected modality for measuring physiological parameters but is not highly interpretable. Thus, translating it to a well-known modality such as electrocardiogram (ECG) improves the usability of Bio-Z in wearables. Deep learning solutions are well-suited for this task given complex relationships between modalities generated by distinct processes. However, current algorithms usually train a single model for all users that results in ignoring cross-user variations. Retraining for new users usually requires collecting abundant labeled data, which is challenging in healthcare applications. In this paper, we build a modality translation framework to translate Bio-Z to ECG by learning personalized user information without training several independent architectures. Furthermore, our framework is able to adapt to new users in testing using very few samples. We design a meta-learning framework that contains shared and user-specific parameters to account for user differences while learning from the similarity amongst user signals. In this model, a meta-learner approximated by a neural network learns how to learn user-specific parameters and can efficiently update them in testing. Our experiments show that the proposed model reduces the percentage root mean square difference (PRD) by 41% compared to training a single model for all users and by 36% compared to training independent models for each user. When adapting the model to new users, our model outperforms fine-tuning a pre-trained model through back-propagation by 40% using as few as two new samples in testing.

As the world becomes increasingly interconnected, emerging and innovative sensing technologies are shaping the future of agriculture, with a special focus on sustainabilityrelated issues. In this context, we envision the possibility to exploit Social Internet of Things for sensing of environmental conditions (solar radiation, humidity, air temperature, soil moisture) and communications, deep learning for plant disease detection, and crowdsourcing for images collection and classification, engaging farmers and community garden owners and experts. Through data fusion and deep learning, the designed system can exploit the collected data and predict when a plant would (or not) get a disease, with a specific degree of precision, with the final purpose to render agriculture more sustainable. We here present the architecture, the deep learning model, and the responsive web app. Finally, some experimental evaluations and usability/engagement tests are reported and discussed, together with final remarks, limitations, and future work.

Background: Hemoglobin is an essential biomolecule for the transportation of oxygen, therefore, its assessment is also important to be done frequently in numerous clinical practices. Traditional invasive techniques have concomitant shortcomings, such as time delay, the onset of infections, and discomfort, which necessitate a non-invasive hemoglobin estimation solution to get rid of these constraints in health informatics. Currently, various techniques are underway in the allied domain, and scanty products are also feasible in the market. However, due to the low satisfaction rate, invasive solutions are still assumed as the gold standard. Recently introduced technologies effectively evolved as optical spectroscopy and digital photographic concepts on different sensing spots, e.g., fingertip, palpebral conjunctiva, bulbar conjunctiva, and fingernail. Productive sensors develop more than eight wavelengths to compute hemoglobin concentration and four wavelengths to display only Hb-index (trending of hemoglobin) either in disposable adhesive or reusable cliptype sensor’s configuration. Objective: This study aims at an optimistic optical spectroscopic technique to measure hemoglobin concentration and conditional usability of non-invasive blood parameters’ diagnostics at point-ofcare. Methods: Two distinguishable light emitting sources (810 nm and 1300 nm) are utilized at isosbestic points with a single photodetector (800-1700 nm). With this purpose, reusable finger probe assembly is facilitated in transmittance mode based on the newly offered sliding mechanism to block ambient light. Results: Investigation with proposed design presents correlation coefficients between reference hemoglobin and every individual feature, a multivariate linear regression model for highly correlated independent features. Moreover, principal component analytical model with multivariate linear regression offers mean bias of 0.036 and -0.316 g/dL, precision of 0.878 and 0.838 and limits of agreement from -1.685 to 1.758 g/dL and -1.790 to 1.474 g/dL for 18 and 21 principal components, respectively. Conclusion: The encouraging readouts emphasize favorable precision; therefore, it is proposed that the sensing system is amenable to assess hemoglobin in settings with limited resources and strengthening future routes for the point of care applications.

When it comes to complex machine learning models, commonly referred to as black boxes, understanding the underlying decision making process is crucial for domains such as healthcare and financial services, as well as when they are used in connection with safety critical systems such as autonomous vehicles. As a result, interest in explainable artificial intelligence (xAI) tools and techniques has increased in recent years. However, the user experience (UX) effectiveness of existing xAI frameworks, especially concerning algorithms that work with data as opposed to images, is still an open research question. In order to address this gap, we examine the UX effectiveness of the Local Interpretable Model-Agnostic Explanations (LIME) xAI framework, one of the most popular model agnostic frameworks found in the literature, with a specific focus on its performance in terms of making tabular models more interpretable. In particular, we apply several state of the art machine learning algorithms on a tabular dataset, and demonstrate how LIME can be used to supplement conventional performance assessment methods. Based on this experience, we evaluate the understandability of the output produced by LIME both via a usability study, involving participants who are not familiar with LIME, and its overall usability via a custom made assessment framework, called Model Usability Evaluation (MUsE), which is derived from the International Organisation for Standardisation 9241-11:2018 standard.

Malfunctions are a major challenge in partially automated driving. During such malfunctions, the driver must be able to adequately take over vehicle guidance without being requested to intervene. This may be particularly difficult in urban areas due to their high complexity and information density. Augmented Reality Head-Up Displays (ARHUDs) may have the potential to support the driver during the monitoring task by providing driving-related information at its required location in the primary field of view. The effects of an ARHUD compared to a Baseline concept in case of malfunctions were investigated in a driving simulation experiment with 52 participants. In a partially automated urban drive, participants experienced a longitudinal and a lateral malfunction in permuted order. The concepts -ARHUD or Baseline- were presented as a between-subject factor. The results showed significantly shorter take-over times when using the ARHUD, resulting in fewer crashes. For those who were able to avoid the crash, no differences in the take-over quality between both concepts were found. There was one difference in visual attention: the attention ratio on the instrument cluster was lower for the ARHUD. In addition, the ARHUD revealed a significantly higher trust and usability rating. However, there were no differences in acceptance and subjective workload between the two concepts. The results showed that the ARHUD has more potential to prevent crashes in the event of malfunctions compared to the Baseline. Nevertheless, the high number of crashes, regardless of the concept, showed the importance of a fallback level for partially automated urban driving.

Motivated by Industry 4.0, 5G-enabled unmanned aerial vehicles (UAVs; also known as drones) are widely applied in various industries. However, the open nature of 5G networks threatens the safe sharing of data. In particular, privacy leakage can lead to serious losses for users. As a new machine learning paradigm, federated learning (FL) avoids privacy leakage by allowing data models to be shared instead of raw data. Unfortunately, the traditional FL framework is strongly dependent on a centralized aggregation server, which will cause the system to crash if the server is compromised. Unauthorized participants may launch poisoning attacks, thereby reducing the usability of models. In addition, communication barriers hinder collaboration among a large number of cross-domain devices for learning. To address the above issues, a blockchain-empowered decentralized horizontal federated learning (BE-DHFL) framework is proposed. The authentication of cross-domain UAVs is accomplished through multisignature smart contracts. Global model updates are computed by using these smart contracts instead of a centralized server. Extensive experimental results show that the proposed scheme achieves high efficiency of cross-domain authentication and good accuracy.

In this paper, we study the problem of quality of service (QoS)-aware cost management of sensor-cloud comprising of multiple sensor-cloud service providers (SCSPs) and sensor-owners. The rapid adaptation of the wireless sensor network (WSN) technology led to the conceptualization of the sensor-cloud infrastructure which primarily aims to reduce the complexities associated with operating WSN-based applications by rendering Sensors-as-a-Service (Se-aaS). However, the oligopolistic market scenario of sensor-cloud involving multiple SCSPs and sensor-owners significantly impacts its profitability and QoS. Thus, there is a need to explore the dynamics of this market competition elaborately in order to maintain the usability of sensor-cloud. Existing literature focusing on sensor-cloud fails to address these issues. Hence, in this work, we analyze the interactions between the SCSPs and the sensor-owners in the Se-aaS market using a game-theoretic approach. We propose a QoS-aware dynamic cost management scheme, named QUEST, to determine the optimal strategies of the various actors in sensor-cloud market. Through simulations, we observe that using QUEST, the price paid by end-users decreases by 10.31-20.43% and the profit of sensor-owners increases by 66.83-89.94%. Moreover, QUEST ensures the optimal distribution of service-requests among the SCSPs and the sensor-owners, and the satisfaction factor of the end-users.

Vision-based intelligent systems such as driver assistance systems and transportation systems should take into account weather conditions. The presence of haze in images can be a critical threat to driving scenarios. Haze density measures the visibility and usability of hazy images captured in real-world conditions. The prediction of haze density can be valuable in various vision-based intelligent systems, especially in those systems deployed in outdoor environments. Haze density prediction is a challenging task since the haze and many scene contents have a lot in common in appearance. Existing methods generally utilize different priors and design complex handcrafted features to predict the visibility or haze density of the image. In this article, we propose a novel end-to-end convolutional neural network (CNN) based method to predict haze density, named as HazDesNet. Our HazDesNet takes a hazy image as input and predicts a pixel-level haze density map. The density map is then refined and smoothed, and the average of the refined map is calculated as the global haze density of the image. To verify the performance of HazDesNet, a subjective human study is performed to build a Human Perceptual Haze Density (HPHD) database, which includes 500 real-world hazy images and 100 synthetic hazy images, and the corresponding human-rated perceptual haze density scores. Experimental results show that our method achieves the best haze density prediction performance on our built HPHD database and existing databases. Besides the global quantitative results, our HazDesNet is capable of predicting a continuous, stable, fine, and high-resolution haze density map. We will make the database and code publicly available at https://github.com/JiaheZhang/HazDesNet.

This paper presents a method for the theoretical analysis of the radiative-wireless power transmission (R-WPT) in lossy environments. We identified the efficiency bounds for specific receiver antennas when the available transmission area was limited. In addition, the optimal transmission currents were obtained and used to determine the proper location and polarization of the transmitting array. It was found that the fields emitted from the receiver in the transmitting mode are an important factor for determining the efficiency bound. To understand the R-WPT with lossy media better, the proposed theoretical framework was applied to two examples. The efficiency bounds for the examples were compared with the results obtained by using previous works to assess the usability of the proposed theory. This study could serve as a basis for the development of R-WPT systems for practical microwave transmission applications with lossy objects.