Showing papers by "National Taiwan University of Science and Technology published in 2018"

Identifying the Best Machine Learning Algorithms for Brain Tumor Segmentation, Progression Assessment, and Overall Survival Prediction in the BRATS Challenge

Abstract: Gliomas are the most common primary brain malignancies, with different degrees of aggressiveness, variable prognosis and various heterogeneous histologic sub-regions, i.e., peritumoral edematous/invaded tissue, necrotic core, active and non-enhancing core. This intrinsic heterogeneity is also portrayed in their radio-phenotype, as their sub-regions are depicted by varying intensity profiles disseminated across multi-parametric magnetic resonance imaging (mpMRI) scans, reflecting varying biological properties. Their heterogeneous shape, extent, and location are some of the factors that make these tumors difficult to resect, and in some cases inoperable. The amount of resected tumoris a factor also considered in longitudinal scans, when evaluating the apparent tumor for potential diagnosis of progression. Furthermore, there is mounting evidence that accurate segmentation of the various tumor sub-regions can offer the basis for quantitative image analysis towards prediction of patient overall survival. This study assesses thestate-of-the-art machine learning (ML) methods used for brain tumor image analysis in mpMRI scans, during the last seven instances of the International Brain Tumor Segmentation (BraTS) challenge, i.e., 2012-2018. Specifically, we focus on i) evaluating segmentations of the various glioma sub-regions in pre-operative mpMRI scans, ii) assessing potential tumor progression by virtue of longitudinal growth of tumor sub-regions, beyond use of the RECIST/RANO criteria, and iii) predicting the overall survival from pre-operative mpMRI scans of patients that underwent gross tota lresection. Finally, we investigate the challenge of identifying the best ML algorithms for each of these tasks, considering that apart from being diverse on each instance of the challenge, the multi-institutional mpMRI BraTS dataset has also been a continuously evolving/growing dataset.

A highly sensitive, self-powered triboelectric auditory sensor for social robotics and hearing aids

Abstract: The auditory system is the most efficient and straightforward communication strategy for connecting human beings and robots. Here, we designed a self-powered triboelectric auditory sensor (TAS) for constructing an electronic auditory system and an architecture for an external hearing aid in intelligent robotic applications. Based on newly developed triboelectric nanogenerator (TENG) technology, the TAS showed ultrahigh sensitivity (110 millivolts/decibel). A TAS with the broadband response from 100 to 5000 hertz was achieved by designing the annular or sectorial inner boundary architecture with systematic optimization. When incorporated with intelligent robotic devices, TAS demonstrated high-quality music recording and accurate voice recognition for realizing intelligent human-robot interaction. Furthermore, the tunable resonant frequency of TAS was achieved by adjusting the geometric design of inner boundary architecture, which could be used to amplify a specific sound wave naturally. On the basis of this unique property, we propose a hearing aid with the TENG technique, which can simplify the signal processing circuit and reduce the power consuming. This work expresses notable advantages of using TENG technology to build a new generation of auditory systems for meeting the challenges in social robotics.

In situ analytical techniques for battery interface analysis

Abstract: Lithium-ion batteries, simply known as lithium batteries, are distinct among high energy density charge-storage devices. The power delivery of batteries depends upon the electrochemical performances and the stability of the electrode, electrolytes and their interface. Interfacial phenomena of the electrode/electrolyte involve lithium dendrite formation, electrolyte degradation and gas evolution, and a semi-solid protective layer formation at the electrode–electrolyte interface, also known as the solid–electrolyte interface (SEI). The SEI protects electrodes from further exfoliation or corrosion and suppresses lithium dendrite formation, which are crucial needs for enhancing the cell performance. This review covers the compositional, structural and morphological aspects of SEI, both artificially and naturally formed, and metallic dendrites using in situ/in operando cells and various in situ analytical tools. Critical challenges and the historical legacy in the development of in situ/in operando electrochemical cells with some reports on state-of-the-art progress are particularly highlighted. The present compilation pinpoints the emerging research opportunities in advancing this field and concludes on the future directions and strategies for in situ/in operando analysis.

The Effect of Environmental Corporate Social Responsibility on Environmental Performance and Business Competitiveness: The Mediation of Green Information Technology Capital

Abstract: With the emergence of environmental sustainability and green business management, increasing demands have been made on businesses in the areas of environmental corporate social responsibility (ECSR). Furthermore, the influence of ECSR on green capital investment, environmental performance, and business competitiveness has also been the subject of attention from enterprises. However, in previous studies, the mediating role of green information technology (IT) capital in the relationship between ECSR, environmental performance, and business competitiveness, has not been investigated by researchers. In order to bridge this gap in the ECSR literature, this study aims to examine the influence of ECSR on green IT capital, and the consequent effect of green IT capital on environmental performance and business competitiveness. Data were collected from 358 companies from the top 1000 manufacturers in Taiwan. The results confirmed that ECSR has significant positive effects on green IT human capital, green IT structural capital, and green IT relational capital. Green IT structural capital and green IT relational capital have positive effects on environmental performance and business competitiveness, and environmental performance has a positive effect on business competitiveness. In addition, green IT structural capital and green IT relational capital have partial mediating effects on ECSR, environmental performance, and business competitiveness. The implications and suggestions for future research are discussed.

Impacts of an augmented reality-based flipped learning guiding approach on students’ scientific project performance and perceptions

Abstract: In recent years, flipped learning has received increasing emphasis; it engages students in deriving basic knowledge through instructional videos before the class, and hence more time is available for practicing, applying knowledge, or student-teacher interaction in class. Many scholars have pointed out that, with such a learning approach, teachers can design more effective in-class activities by guiding students to have higher order thinking as well as interactions with peers and teachers. In the meantime, researchers have also indicated that employing proper educational technologies or learning strategies could further improve students’ performance. Therefore, in this study, an Augmented Reality (AR)-based learning guiding mode is proposed for developing a flipped learning system. To examine the effectiveness of the proposed approach, an experiment was conducted in a natural science learning activity of an elementary school using the developed system. The participants were four classes of 111 fifth graders. Two classes were assigned to the experimental group, while the others were the control group. Those learning in the experimental group used the AR-based flipped learning mode, while those in the control group learned with the conventional flipped learning mode. From the experimental results, it was found that the AR-based flipped learning guiding approach not only benefited the students in terms of promoting their project performance, but also improved their learning motivation, critical thinking tendency, and group self-efficacy.

Some intuitionistic fuzzy Dombi Bonferroni mean operators and their application to multi-attribute group decision making

Abstract: The Bonferroni mean (BM) operator has the advantage of considering interrelationships between parameters, but it only can deal with crisp values. In recent years, many extended BM operators have been proposed to deal with fuzzy information. Dombi Bonferroni mean operators are special cases of general T-conorm and T-norm, which have the advantage of good flexibility with a general parameter. In this paper, we extend the BM operator based on the Dombi operations to propose the intuitionistic fuzzy Dombi Bonferroni mean (IFDBM) operator, the intuitionistic fuzzy weighted Dombi Bonferroni mean (IFWDBM) operator, the intuitionistic fuzzy Dombi geometric Bonferroni mean (IFDGBM) operator and the intuitionistic fuzzy weighted Dombi geometric Bonferroni mean (IFWDGBM) operator for dealing with the aggregation of intuitionistic fuzzy numbers (IFNs) and propose some multi-attribute group decision-making (MAGDM) methods. Firstly, we introduce the concept, the characteristics, the score function, the accuracy function and the operational rules of IFNs. Then, we propose the IFDBM operator, the IFWDBM operator, the IFDGBM operator and the IFWDGBM operator for aggregating IFNs. Then, we propose two MAGDM methods based on the proposed IFWDBM operator and the proposed IFWDGBM operator for dealing with MAGDM problems. Finally, we use an example to illustrate the MAGDM process of the proposed MAGDM methods. The proposed intuitionistic fuzzy Dombi Bonferroni mean operators are very useful to deal with MAGDM problems.

Trends in mobile technology-supported collaborative learning: A systematic review of journal publications from 2007 to 2016

Abstract: This study reviewed the literature on mobile technology-supported collaborative learning from 2007 to 2016. Several issues, such as the distributions and research methods, learning devices and learning environments, participants, research issues, application domains, grouping methods and collaborative learning strategies, are addressed. In addition, the relationship between the learning strategies and measurement issues are investigated. The review found that the amount of research on mobile collaborative learning increased and the connection between new mobile technology and collaborative learning activities became tighter. College students received the greatest emphasis, but more focus should be put on junior and elementary school students. Few studies were conducted on teachers and adults. In the most recent five years, the research was focused on improving learners' performance in science, especially social science, and in natural scenarios outside of the classroom, but less emphasis was put on developing learners’ skills and higher order skills. There was little research focusing on different selection methods of group members and the teaching effects of grouping design. Most research adopted conceptualized collaborative learning strategies. Furthermore, some studies proposed that the collaborative learning activities conducted in mobile learning environments should be designed carefully to guide students to experience more effective collaborative constructivist learning. Based on the findings, in-depth discussion and suggestions for future studies are given.

Why rankings of biomedical image analysis competitions should be interpreted with care

Abstract: International challenges have become the standard for validation of biomedical image analysis methods. Given their scientific impact, it is surprising that a critical analysis of common practices related to the organization of challenges has not yet been performed. In this paper, we present a comprehensive analysis of biomedical image analysis challenges conducted up to now. We demonstrate the importance of challenges and show that the lack of quality control has critical consequences. First, reproducibility and interpretation of the results is often hampered as only a fraction of relevant information is typically provided. Second, the rank of an algorithm is generally not robust to a number of variables such as the test data used for validation, the ranking scheme applied and the observers that make the reference annotations. To overcome these problems, we recommend best practice guidelines and define open research questions to be addressed in the future.

A Blockchain Connected Gateway for BLE-Based Devices in the Internet of Things

Abstract: Recently, the popularity of the Internet of Things (IoT) has led to a rapid development and significant advancement of ubiquitous applications seamlessly integrated within our daily life. Owing to the accompanying growth of the importance of privacy, a great deal of attention has focused on the issues of secure management and robust access control of IoT devices. In this paper, we propose the design of a blockchain connected gateway which adaptively and securely maintains user privacy preferences for IoT devices in the blockchain network. Individual privacy leakage can be prevented because the gateway effectively protects users’ sensitive data from being accessed without their consent. A robust digital signature mechanism is proposed for the purposes of authentication and secure management of privacy preferences. Furthermore, we adopt the blockchain network as the underlying architecture of data processing and maintenance to resolve privacy disputes.

Polyethylene oxide film coating enhances lithium cycling efficiency of an anode-free lithium-metal battery

Abstract: The practical implementation of an anode-free lithium-metal battery with promising high capacity is hampered by dendrite formation and low coulombic efficiency. Most notably, these challenges stem from non-uniform lithium plating and unstable SEI layer formation on the bare copper electrode. Herein, we revealed the homogeneous deposition of lithium and effective suppression of dendrite formation using a copper electrode coated with a polyethylene oxide (PEO) film in an electrolyte comprising 1 M LiTFSI, DME/DOL (1/1, v/v) and 2 wt% LiNO3. More importantly, the PEO film coating promoted the formation of a thin and robust SEI layer film by hosting lithium and regulating the inevitable reaction of lithium with the electrolyte. The modified electrode exhibited stable cycling of lithium with an average coulombic efficiency of ∼100% over 200 cycles and low voltage hysteresis (∼30 mV) at a current density of 0.5 mA cm-2. Moreover, we tested the anode-free battery experimentally by integrating it with an LiFePO4 cathode into a full-cell configuration (Cu@PEO/LiFePO4). The new cell demonstrated stable cycling with an average coulombic efficiency of 98.6% and capacity retention of 30% in the 200th cycle at a rate of 0.2C. These impressive enhancements in cycle life and capacity retention result from the synergy of the PEO film coating, high electrode-electrolyte interface compatibility, stable polar oligomer formation from the reduction of 1,3-dioxolane and the generation of SEI-stabilizing nitrite and nitride upon lithium nitrate reduction. Our result opens up a new route to realize anode-free batteries by modifying the copper anode with PEO to achieve ever more demanding yet safe interfacial chemistry and control of dendrite formation.

A scoping review of research on digital game-based language learning

Abstract: The continuing attention to the educational value of digital games highlights the need for more focused literature reviews in order to identify critical gaps and opportunities in domain-specific areas. The current study thus set out to provide a scoping overview of empirical evidence on the use and impacts of digital games in language education from 2007 to 2016, as a means to advance the emerging research on digital game-based language learning (DGBLL). A total of 50 selected studies were systematically analyzed, revealing the following findings: (1) Most of the selected DGBLL studies adopted mixed methods to examine the educational use of digital games; (2) Immersive games, notably massively multiplayer online role-playing games, were the most common genre in the current DGBLL literature; (3) Most of the games for language learning were custom-built by DGBLL researchers; (4) Personal computers were the most common platforms for playing games to support language learning; (5) Most of the DGBLL studies adopted games to facilitate the learning of English as a second or foreign language; (6) Most of the research on DGBLL investigated learners with mixed levels of language proficiency; (7) University students were the most frequently selected samples in the existing DGBLL literature; and (8) The majority of DGBLL studies featured positive outcomes in regard to student learning, with the most frequently reported ones being related to affective or psychological states, closely followed by language acquisition. Taken as a whole, these findings reflect the diverse nature of this field and suggest the overall feasibility of using digital games for promoting the language and literacy learning of both native and non-native speakers in various aspects. Several promising but under-researched areas were also identified in this review, along with discussions on their implications for future investigations.

Global Asymptotic Model-Free Trajectory-Independent Tracking Control of an Uncertain Marine Vehicle: An Adaptive Universe-Based Fuzzy Control Approach

Abstract: Motivated by the challenging difficulty in tracking an uncertain marine vehicle (MV) with unknown dynamics and disturbances to any unmeasurable/unknown trajectory, which is unresolved, an adaptive universe-based fuzzy control (AUFC) scheme with retractable fuzzy partitioning (RFP) in global universe of discourse (UoD) is created to achieve global asymptotic model-free trajectory-independent tracking. By defining an error surface and intensively exploring the MV structure, tracking error dynamics are sufficiently trimmed via separating external unknowns including trajectory dynamics and disturbances from internal nonlinearities dependent on tracking errors. An innovative retractable fuzzy approximator (RFA) using the RFP is developed to estimate internal nonlinearities and does not require a priori knowledge on the UoD, thereby contributing to a globally adaptive approximation based control approach in conjunction with Lyapunov synthesis. Together with RFA residuals, external unknowns are globally dominated by adaptive universal compensators driven by tracking error surface. Eventually, tracking errors and their derivatives globally asymptotically converge to the origin and all other signals of the closed-loop system are bounded. Simulation studies demonstrate superior performance of the proposed AUFC scheme in terms of both tracking and approximation.

Trends and research issues of mobile learning studies in nursing education: A review of academic publications from 1971 to 2016

Abstract: In the past decades, the issues related to mobile learning have been widely discussed around the globe; however, the development and trends of applying mobile technologies in nursing education still lack systematic analysis. In this study, a meta review of the studies published in academic journals from 1971 to 2016 was conducted to analyze the application domains, subjects, adopted learning strategies, investigated research issues and findings of mobile technology-supported nursing education. From the review results, it was found that the use of mobile technologies in nursing education and training have made great progress in the past decades. In addition to the changes in mobile technologies and the increasing number of mobile learning studies in nursing education, the subjects and research issues have also become more diverse in recent years. It was also found that mobile learning has mainly been applied to the training of basic nursing concepts and skills as well as to long-term care and obstetrics and gynecology, while few or even no studies are related to other nursing education domains. In addition, several widely adopted mobile learning strategies, such as inquiry-based learning, contextual mobile learning, snychronous sharing, Mindtools, project-based learning and peer assessment, have seldom been adopted in mobile nursing education. This also reflects the fact that most of these studies focused on skills training and basic knowledge comprehension, while few were conducted in the domains aimed at fostering learners' higher order thinking competences, such as problem solving or critical thinking. On the other hand, it was found that the number of studies using an experimental design has increased in recent years; moreover, most studies reported the learners’ cognitive performance and perceptions, while their learning behaviors were seldom analyzed. Accordingly, the research trends and potential research issues of mobile nursing education are proposed as a reference for researchers, instructors and policy makers.

RouteNet: routability prediction for mixed-size designs using convolutional neural network

Abstract: Early routability prediction helps designers and tools perform preventive measures so that design rule violations can be avoided in a proactive manner. However, it is a huge challenge to have a predictor that is both accurate and fast. In this work, we study how to leverage convolutional neural network to address this challenge. The proposed method, called RouteNet, can either evaluate the overall routability of cell placement solutions without global routing or predict the locations of DRC (Design Rule Checking) hotspots. In both cases, large macros in mixed-size designs are taken into consideration. Experiments on benchmark circuits show that RouteNet can forecast overall routability with accuracy similar to that of global router while using substantially less runtime. For DRC hotspot prediction, RouteNet improves accuracy by 50% compared to global routing. It also significantly outperforms other machine learning approaches such as support vector machine and logistic regression.

Synergy between Ceria Oxygen Vacancies and Cu Nanoparticles Facilitates the Catalytic Conversion of CO2 to CO under Mild Conditions

Abstract: The performance of supported metal catalysts can depend on many factors, including metal particle size and dispersion and metal–support interactions, and differentiation of these effects is challen...

Forward Forecast of Stock Price Using Sliding-Window Metaheuristic-Optimized Machine-Learning Regression

Abstract: Time series forecasting has been widely used to determine the future prices of stock, and the analysis and modeling of finance time series importantly guide investors’ decisions and trades. In addition, in a dynamic environment such as the stock market, the nonlinearity of the time series is pronounced, immediately affecting the efficacy of stock price forecasts. Thus, this paper proposes an intelligent time series prediction system that uses sliding-window metaheuristic optimization for the purpose of predicting the stock prices of Taiwan construction companies one step ahead. It may be of great interest to home brokers who do not possess sufficient knowledge to invest in such companies. The system has a graphical user interface and functions as a stand-alone application. The developed hybrid system exhibited outstanding prediction performance and it improves overall profit for investment performance. The proposed model is a promising predictive technique for highly nonlinear time series, whose patterns are difficult to capture by traditional models.

A Universal Method to Engineer Metal Oxide-Metal-Carbon Interface for Highly Efficient Oxygen Reduction.

Abstract: Oxygen is the most abundant element in the Earth’s crust. The oxygen reduction reaction (ORR) is also the most important reaction in life processes and energy converting/storage systems. Developing techniques toward high-efficiency ORR remains highly desired and a challenge. Here, we report a N-doped carbon (NC) encapsulated CeO2/Co interfacial hollow structure (CeO2–Co–NC) via a generalized strategy for largely increased oxygen species adsorption and improved ORR activities. First, the metallic Co nanoparticles not only provide high conductivity but also serve as electron donors to largely create oxygen vacancies in CeO2. Second, the outer carbon layer can effectively protect cobalt from oxidation and dissociation in alkaline media and as well imparts its higher ORR activity. In the meanwhile, the electronic interactions between CeO2 and Co in the CeO2/Co interface are unveiled theoretically by density functional theory calculations to justify the increased oxygen absorption for ORR activity improvement....

Two-dimensional electronic transport and surface electron accumulation in MoS2.

Abstract: Because the surface-to-volume ratio of quasi-two-dimensional materials is extremely high, understanding their surface characteristics is crucial for practically controlling their intrinsic properties and fabricating p-type and n-type layered semiconductors. Van der Waals crystals are expected to have an inert surface because of the absence of dangling bonds. However, here we show that the surface of high-quality synthesized molybdenum disulfide (MoS2) is a major n-doping source. The surface electron concentration of MoS2 is nearly four orders of magnitude higher than that of its inner bulk. Substantial thickness-dependent conductivity in MoS2 nanoflakes was observed. The transfer length method suggested the current transport in MoS2 following a two-dimensional behavior rather than the conventional three-dimensional mode. Scanning tunneling microscopy and angle-resolved photoemission spectroscopy measurements confirmed the presence of surface electron accumulation in this layered material. Notably, the in situ-cleaved surface exhibited a nearly intrinsic state without electron accumulation.

Path Tracking of an Autonomous Ground Vehicle With Different Payloads by Hierarchical Improved Fuzzy Dynamic Sliding-Mode Control

Abstract: A hierarchically improved fuzzy dynamical sliding-model control (HIFDSMC) is presented to address the autonomous ground vehicle (AGV) path tracking problem. The proposed controller has two portions: one is the virtual desired input (VDI), and the second is the path tracking control (PTC). In addition to the equivalent control in VDI and PTC, an improved fuzzy dynamical sliding-mode control (IFDSMC) is designed to deal with the system uncertainties, e.g., different payloads. Contributions of this paper include the following four parts: 1) Based on the nominal system response, the fuzzy rules and scaling factors of the IFDSMCs in the VDI and PTC are easily chosen. In contrast, a conventional fuzzy logic control approach requires more trial-and-error tuning to obtain a satisfactory performance. 2) The proposed HIFDSMC possesses the tuning mechanism (the coefficients of two sliding surfaces, the scaling factors in indirect and direct modes, and the fine tuning in fuzzy table) such that the uncertainties are tackled without a larger computational burden. 3) The stability of the closed-loop system is verified by the Lyapunov stability with hierarchical concept. 4) Different payloads not at the mass center of the AGV (e.g., greater than 25% in the total weight of the AGV) are tackled by the IFDSMCs to obtain a satisfactory performance. Finally, the application to the assembly line of the AGV with different payloads to track the circular and piecewise straight-line paths by the proposed HIFDSMC is compared with the hierarchical fuzzy decentralized PTC.

Ultrathin Sulfur-Doped Graphitic Carbon Nitride Nanosheets As Metal-Free Catalyst for Electrochemical Sensing and Catalytic Removal of 4-Nitrophenol

Abstract: Graphene-like sulfur-containing graphitic carbon nitride (S-GCN) nanosheets were successfully prepared and thoroughly characterized. A simple synthetic method by a thermal condensation approach was...

A New Controller for Bidirectional Wireless Power Transfer Systems

Abstract: Wireless power transfer (WPT) is a promising technology for supplying power to various applications with no physical contacts. Recently, bidirectional WPT (BD-WPT) systems are gaining popularity for grid-to-vehicle and vehicle-to-grid applications which essentially require power transfer in both directions. However, BD-WPT systems are complex in nature, and require sophisticated control strategies to provide synchronization as well as to regulate the power flow between two sides. This paper proposes a new controller that uses measured active power (P) and reactive power (Q) at the resonant network of BD-WPT systems to regulate the power flow in both directions while providing synchronization between two sides without a dedicated communication interface for controlling power transfer. The controller, located on the pickup side, is applicable to BD-WPT systems with either single or multiple loads and ensures that the volt-ampere rating of the pickup converter is at lowest for battery charging, by keeping the reactive power at the resonant tank intake by the load side minimum. Experimental results obtained from a 1-kW prototype system show good agreement with simulated results, validating that the proposed controller can be used to regulate the power flow in BD-WPT systems.