Search

Hyperspectral imagery provides wealthy spectral information to make it suitable for many applications. However, for specific applications, extracting suitable bands from high-dimensional data is a tedious and difficult task. In the past, many methods have been developed to perform band selection for specific tasks such as target detection. However, there is very little work to consider and deal with the effects of suspected interfering targets. In this paper, a new method for band selection, called target-constrained interference-minimized band selection (TCIMBS) is developed for specific target detection. It can select a band set with strong characterization capabilities for desired targets and good suppression for undesired targets and background (BKG). Experimental results demonstrate that TCIMBS can improve the detection performance, and also achieve better performances in comparison with several state-of-the-art methods.

IoT devices must and will be more intelligent in the future. However, due to the lack of benchmarks representative to the diverse IoT applications, there are limited architecture performance studies on IoT devices. This paper presents IoTBench, the first benchmark suite targeting at IoT edge-device applications. This suite includes seven representative programs from three major IoT categories. We investigate the computational demand and execution efficiency of these benchmarks running on a popular IoT device platform. We also analyze and characterize the energy consumption of IoTBench using analytic approaches. Overall, IoTBench establishes a foundation for innovative architecture design of IoT edge devices.

In terms of the application of the Massive Open Online Courses (MOOCs) in the elementary schools, it requires both the online courses and actual classroom courses to make it work. Therefore, SPOCs (Small Private Online Courses) is more suitable than MOOCs for elementary school students. However, the elementary school students are not mature enough regarding their mental age; in addition, their autonomous learning ability is insufficient to handle all courses; also, in actual classroom teaching, their in-class discussion participation is low. Therefore, this study hopes to design a improved SPOCs teaching method to solve the problems of the current SPOC teaching method might encounter. This paper aims to explore the impact of different teaching methods (traditional lecture teaching method, conventional SPOC teaching method and improved SPOCs teaching method) on students' learning outcome of butterfly's ecological knowledge, the degree of classroom discussion and the viewing rate of pre-class films. The paper would provide a solid reference for elementary school Massive Open Online Course teaching or SPOCs (Small Private Online Course) teaching. For the experiment of this study, the number of valid samples were 75 elementary Grade 5 students from 3 different classes in Tainan. The three classes were randomly divided into three groups: one experimental group and two control groups. The research method was a quasi-experimental study, and the butterfly ecological knowledge teaching was carried out for five weeks. “Butterfly Ecology Knowledge Learning Platform” was implemented during the experimental teaching process. All students from the three groups received the “Butterfly Ecological Knowledge Learning Test” in order to recognize student's learning outcomes. The experiment results show the learning outcome of students who received “improved SPOC teaching method” in butterfly ecological knowledge is significantly better than those who received “traditional lecture method” and the “conventional SPOCs teaching method”.

With the advance of information technology, video on the web (such as, Youtube) had been a common and convenient channel for learning by students. But the characteristics of transient information in videos are easy to impose a high cognitive load to students, especially for elementary school students. Because video is not a single text or a single image, which has rich colors and complex scenarios. Video requires the students to watch it attentively, otherwise it is possible to misunderstand the meaning of the context and lose clues involved in the scenarios. Due to that the frames in the video are continuously broadcasted one-by-one, about 20–30 frames per second, students not only need to quickly engage the possible messages between the previous frame with the next one, but also need to integrate the messages with their existing knowledge. Obviously, this a difficult job for students with limited short-term memory, which leads to a cognitive load problem and reduces learning effectiveness. To solve this problem, Video-segmentation was known as an instructional technique that advocates dividing the learning video into meaningful segments to reduce cognitive load. However, all the related researches showed that Video-segmentation instructional technique can raise the learning achievement only for inexperienced learners. Oppositely, it will be not only no effectiveness but also have negative consequences for experienced learners, which is called the Expertise Reversal Effect. Therefore, this paper propose a new novel approach to adaptively divide the video based on learners' prior knowledge and grade to meaningful segments for each student. Such an approach could reduce learners' cognitive load and improve the learning achievement. The results show that there indeed exists Expertise Reversal Effect of segmentation in terms of scores of higher-prior knowledge students in the control group. In experimental group, scores of post-test are significantly superior to scores of pre-test for mid-prior and lower-prior knowledge students.

English is the most important international communication tool in the word. How to teach students to learn English in Non-English speaking countries is always a major task for education. Today, learning for fun and learning from life experience would be a better policy for learning English, especially for children. For the reason that, video is a good media suitable for learning languages, which can record real life situations and can be easily clicked-through via a browser on the Internet, i.e., Youtube. But, when the student watches an English video with caption, the student would be interfered by the pop-up caption since the words in caption are familiar. On the other hand, the caption is closed for watching a video, which may results in heavy cognitive load of learning for the student when most of the words in caption are unfamiliar. Therefore, in this paper, we proposed a new mechanism, called Personal Word-Hiding English Vocabulary Learning System, which can be embedded into a video-browser and automatically hide English words for students involved in the video caption when the students watches English Video through a Browser on the Internet.

Peer assessment refers to the students with the similar academic level of learning achievement or knowledge background. During a learning activity of typical peer assessment, in order to have more meaningful peer feedback, students must have sufficient evaluation professional degree. Therefore, it is necessary to train students how to write in a meaningful text and constructive comments. Besides, this student will also receive review comments for his/her work from other peers in the same group, then the student can correct his/her work according to the comments, thereby peer assessment can also promote cognitive growth and learn capabilities of the student. Traditionally, students use pens and papers to join the peer assessment activities. Today, due to the booming of Internet and technological advances in technology, peer assessment activities had be moved on the Internet platform called Networked Peer Assessment, which has many advantages, such as, ensuring anonymity, enabling teachers to monitor learning and review process, performing the process not limited by time and space. Moreover, in the networked peer assessment activity, the most critical factor in the success of improving learning achievement is the quality of peers' comments. But, the students maybe lack of expertise knowledge of the learning object, such that the students do not have enough abilities to give proper, useful and constructive comments for peers' works. Therefore, this paper proposed an Individual Evaluation Professionalism supplemented by adaptive level recommendation mechanism for comment quality achievements, learning quality achievements and peer assessment validity on Networked Peer Assessment. Research subjects were two classes selected from a junior high school in Taiwan; one as the experimental group were provided by the proposed assisted mechanism with the networked peer assessment, the other one as control group were treated by the typical networked peer assessment. The results showed that: (1) There were significant differences between the experimental group and the control group in the learning effectiveness; (2) the learning attitudes of students participated in the network peer assessment activities with assisted assessment guide were a positive affirmation.

The popularity smart phones, mobile applications (Apps) have exploded in recent years creating unlimited business opportunities. Today's customers desire to acquire not only software and products, but also memorable experiences. Using O2O (Online to Offline, Offline to Online) marketing strategies, customers interact with software and products in the real world that doing so will help them increase their imagination and enjoyment, providing them with different and memorable experiences. An exploratory qualitative study was undertaken using case study "LINE&". Data were collected from secondary documents and interviews with customers about LINE's O2O services for creating customer experiences over the past three years. This study focuses on LINE's integration of online App-based services and offline real-world marketing, categorizes these services into five types of experiential strategy models based on their different characteristics. We conclude with several lessons related to the integration of the virtual and physical worlds to create customer O2O experiences.

In this study, we fabricated 150-nm In 0.6 Ga 0.4 As mHEMT devices using in-house fabrication process. The devices were packaged using FCOB technology on organic substrates. The packaged device exhibited favorable DC characteristics with I DS = 350 mA/mm and a transconductance of 600 mS/mm at V DS = 0.5V with a maximum available gain (MAG) 6.5 dB at 60 GHz. The impact of temperature setting during the bonding process on the DC and RF performance was experimentally investigated. RF performance degradation was observed at high bonding temperature due to the induced thermomechanical stress. The stress was mainly from the mismatch in the coefficient of thermal expansion (CTE) of the GaAs chip and organic substrate.

Due to the rapid growth of wireless communication systems, high frequency packages become very important and they require compactness, low cost and high performances even at frequency up to 60 GHz. Flip-chip assembly using organic substrate at very high frequency has become a cost competitive packaging method in semiconductor industries.

Current Si-CMOS technology has come to a limit that novel semiconductors as alternative channel materials (Ge, InSb, In x Ga 1−x As) are urgently needed for high-speed and low-power logic devices for post CMOS era [1]. Recent research shows III–V heterostructure field-effect transistors demonstrate aggressive merits due to its high electron mobility and rather mature process technology [2]. The outstanding low field electron transport characteristics of III–V materials make ultrahigh-speed switching at very low supply voltage possible [3]. Here, we present the latest advancement of 40 nm InAs/In x Ga 1−x As composite channel High Electron Mobility Transistor (HEMT) devices that have achieved excellent digital logic characteristics at very low power level.

The fabrication process of an 80 nm In 0.7 Ga 0.3 As MHEMT device with flip-chip packaging on Al 2 O 3 substrate is presented. The flip-chip packaged device exhibited good dc characteristics with high I DS = 425 mA/mm and high g m = 970 mS/mm at V DS = 1.5 V. Besides, the RF performances revealed high gain of 10 dB at 50 GHz and low minimum noise figure (NF min )below 2 dB at 60 GHz, showing the feasibility of flip-chip packaged In 0.7 Ga 0.3 As MHEMT device for low noise applications at W-band.

80nm InAs/In0.7Ga0.3As HEMTs using Pt gate sinking were characterized for ultra-low power low noise applications. While the epitaxial structure of the device was optimized, the reduction of gate-to-channel distance was achieved from gate sinking process. The device exhibited very high drain current density of 1066 mA/mm and maximum gm of 1900 mS/mm at Vds = 0.5 V. Excellent fT (fmax) up to 113 GHz (110 GHz) at VDS = 0.1 V was achieved with very low NFmin of 1.05dB and associated gain of 8.6dB at 17 GHz were achieved at 0.1V VDS. These results demonstrated the outstanding potential for state-of-art ultra-low power and low noise applications.

The DC and RF performances of 40 nm high electron mobility transistors (HEMTs) with composite channel of InAs channel and In0.53Ga0.47As sub-channel were demonstrated. The drain current was 870 mA/mm (Vds=0.4V, Vgs=0V) and maximum gm was 1750 mS/mm (Vds=0.5V, Vgs=-0.65V). The devices showed high current gain cutoff frequency (fT) of 420 GHz and low gate delay time of 0.77 ps owing to the nanometer gate length and extremely high electron mobility of the InAs channel. Under low DC power consumption; these InAs HEMTs still exhibited excellent RF and logic performance which indicates these devices have great potential for future high-speed and low-voltage applications.

80-nm InAs channel HEMTs with different lattice matched sub-channels, In0.53Ga0.47As and In0.7Ga0.3As, have been fabricated. The device with InAs/In0.7Ga0.3As composite channel exhibits high drain current density (1101 mA/mm), and high transconductance (1605 mS/mm) at drain bias VDS = 0.8 V. The high current gain cutoff frequency (ft) of 360 GHz and maximum oscillation frequency (fmax) of 380 GHz of the device with InAs/In0.7Ga0.3As were obtained at VDS = 0.7 V in comparison to the InAs/In0.53Ga0.47 As channel HEMTs with ft = 310 and fmax = 330 GHz. This is due to the high electron mobility and electron confinement in the InAs/In0.7Ga0.3As channel. In addition, a low gate delay time 0.84 psec was obtained at VDS = 0.5 V. The excellent performance of the InAs channel HEMTs demonstrated in this study shows great potential for high speed and very low power logic applications with the optimal design of In0.7Ga0.3As/InAs/In0.7Ga0.3As composite channel.

An 80-nm InP HEMT with InAs channel and InGaAs sub-channels has been fabricated. The high current gain cutoff frequency (ft) of 350 GHz and maximum oscillation frequency (fmax) of 360 GHz were obtained at VDS = 0.7 V due to the high electron mobility in the InAs channel. DC and RF characterizations on the device have been performed and the on- sate breakdown voltage of the device was measured to be 1.75 V. A 2-stage MMIC gain block was designed using such device. A simulated gain of better than 12 dB from 54 GHz to 71 GHz with only 14 mW DC power consumption was achieved. Such high performance HEMTs demonstrated in this study shows great potential for very low power millimeter wave applications.

Experiments show that CAIM discretization algorithm is superior to all the other top-down discretization algorithms. However, CAIM algorithm does not take the data distribution into account. The discretization formula used in CAIM also gives a high factor to the numbers of generated intervals. The two disadvantages make CAIM may generate irrational discrete results in some cases and further leads to the decrease of predictive accuracy of a classifier. In this paper we propose the class-attribute contingency coefficient discretization algorithm. The experimental results showed that compared with CAIM, our method can generate a better discretization scheme to bring on the improvement of accuracy of classification. With regard to the number of generated rules and execution time of a classifier, CACC and CAIM achieve comparable results.

80-nm-gate In0.7Ga0.3As/InAs/In0.7Ga0.3As composite channel high-electron mobility transistors (HEMTs) fabricated using platinum (Pt) buried gate as the Schottky contact metal were evaluated for RF and logic application. After gate sinking at the 250degC for 3 minutes, the device exhibited a high gm value of 1590 mS/mm at Vd = 0.5 V and the current gain cutoff frequency fT was measured to be 494 GHz. The intrinsic gate delay time was calculated to be 0.78 psec at supply voltage of 0.6 V. This is the highest fT achieved for 80 nm gate length HEMT devices. These superior performances are attributed to the reduction of distance between gate and channel, and the reduction of parasitic gate capacitances during gate-sinking process.

In a database, the concept of an example might change along with time, which is known as concept drift. When the concept drift occurs, the classification model built by using old dataset is not suitable for predicting new coming dataset. Although many algorithms had been proposed to solve this problem, they focus only on updating the classification model. However, in a real life users might be very interested in the rules of concept drift. For example, doctors would desire to know the main causes more for disease variation since such rules would enable them to diagnose patients more correctly and quickly. In this paper, we propose a concept drift rule mining tree to accurately discover the rule of concept drift. The main contributions of this paper are: a) we address the problem of mining concept-drifting rule which was ignored in the past; b) our method can accurately mine the rule of concept drift.

The magnetic anisotropy properties of (Ga, Mn)As epilayers are dependent on strain, thickness, substrates orientation, measurement temperature and shape. Systematic study magnetic anisotropy is important owing to they will strongly influence its application of spintronic devices. The effect of low-temperature annealing on the lattice constant and magnetic properties of ternary (Ga, Mn)As grown on GaAs substrates has been intensely studied. Here, we report magnetic anisotropy properties of Ga0.93Mn0.07As by low-temperature annealing.The Ga0.93Mn0.07As sample was grown on (001) semi-insulating GaAs substrates by a Varian Modular GEN-II MBE system. The surface reconstruction of the samples during growth was monitored in situ with the reflection high-energy electron diffraction (RHEED). After growth, sample was cleaved into pieces for various characterizations. The Mn concentrations were determined by a calibration curve based on the electron microprobe analysis (EMPA). The crystalline structures were examined by double-crystal x-ray diffraction (DXRD). Magnetic properties were measured by a Quantum Design MPMS2 superconducting quantum interference device (SQUID) magnetometer. To measure the magnetic properties of Ga0.93Mn0.07As epilayers the samples are cooled down to 10 K under an external field of 1000 Oe applied along the measurement axis. Then, field is removed, and projection of the remnant magnetization along the measurement axis is recorded as a function of increasing temperature. The temperature dependence of magnetization (M-T) curves of as-grown and annealed samples, were measured in small magnetic field (H = 1 Oe) applied in-plane directions. The M-T curves show in Fig. 1. Post-annealing with 250 oC for 1.5 hrs, the Curie temperatures (TC) value increases from 80 Kto 160 K.The magnetic field dependence of magnetization (M-H) curves was measured with applied fields along [110], [-110], [100] and [010] axis, respectively. M-H curves of as-grown and annealed samples are shown. The easy axis can be rotated from [100] direction to [010] direction by low-temperature annealing. And the coercivity field is from 50 G decrease to 18 G. Detailed studies of the effect of low-temperature annealing on magnetic anisotropy properties were discussed.

To further decrease the power gain loss and noise figure of an RF LNA circuit from the on-chip ESD protection circuit, novel LC-tank RF ESD protection circuits is proposed in this paper which has been successfully verified in a 0.25-/spl mu/m CMOS process with top thick metal. With the resonance of the LC-tank, the LC-tank RF ESD protection circuit can reduce the power gain loss and noise figure of the ESD at the operation frequency. From the experimental results, the 4.5 circles inductor of the LC-tank is the best choice for he requirement of 2 kV HBM ESD level. The proposed LC-tank ESD protection circuit will be one of the most effective ESD protection solutions for RF circuits in the higher frequency band (>10 GHz).