Search

Precision nitrogen fertilizer application depends on accurate estimation of plant nitrogen content. However, the assessment of plant nitrogen content at early growth stages of paddy rice through remote sensed images is complicated by the compound effects of backgrounds (e.g. flood water, bare soil, algae, etc.) on the band reflectance. The rapid changing of plant nitrogen content during the vegetative phase makes the development of an operational prediction model very difficult. In this study, aerial images acquired by a quadcopter unmanned aerial vehicle (UAV) equipped with a multispectral sensor were used to estimate plant nitrogen content at vegetative phase of rice crops. The experiments were conducted at the experimental farm of Taiwan Agricultural Research Institute (TARI) from 2018 to 2020. A variable, N-index (ratio between N content of plants to be evaluated and plants not receiving N fertilizers), was introduced to resolve the issues related to rapid changing of plant N content during the vegetative phase. After removing the interference on band reflectance by background from the aerial images, the most appropriate vegetation indices and period that can capture the variations of N-index of rice plants were identified. It was found that a normalized difference red edge index (NDRI) and red edge chlorophyll index (RECI) based model correlated well with the N-index values from c.a. 30 days after transplanting (DAT) to 55 DAT (i.e., the most crucial period for rice yield and grain quality). The developed model was then used to display the spatial and temporal heterogeneity in plant nitrogen status within an experimental field as an example to illustrate how to use the model. In the example, soil plant analysis development (SPAD) meter values at locations of various levels of estimated N-index were collected as surrogates of plant nitrogen content at various DATs to build relationships for converting N-index maps to SPAD maps for potential variable rate fertilizer application management.

Recently Helm (2020) provides an excellent review and preliminary assessment of the environmental impacts of the Coronavirus. He concludes that “The evidence from the pandemic is that it is not the case that decoupling has occurred at the global level, or even at the European level. Emissions and GDP have both fallen sharply” (p. 3). Using the case of Taiwan, this study provides a local evidence to support Helm (2020)’s proposition. GDP and emissions in Taiwan have been highly correlated in the coronavirus pandemic. However, we found that there exists a significant asymmetry of the correlation (or coupling) between GDP and CO2 emission under each period of prosperity and depression from 2005 to 2020. In this respect, the analysis of the causal link between CO2 emission and GDP under the depression period like that in the coronavirus pandemic cannot be considered as a trivial duplicate of the analysis under the prosperous period.

Parkinson’s disease (PD), a progressive disease that affects movement, is related to dopaminergic neuron degeneration. Tc-99m Trodat-1 brain (TRODAT) single-photon emission computed tomography (SPECT) aids the functional imaging of dopamine transporters and is used for dopaminergic neuron enumeration. Herein, we employed a convolutional neural network to facilitate PD diagnosis through TRODAT SPECT, which is simpler than models such as VGG16 and ResNet50. We retrospectively collected the data of 3188 patients (age range 20–107 years) who underwent TRODAT SPECT between June 2011 and December 2019. We developed a set of functional imaging multiclassification deep learning algorithms suitable for TRODAT SPECT on the basis of the annotations of medical experts. We then applied our self-proposed model and compared its results with those of four other models, including deep and machine learning models. TRODAT SPECT included three images collected from each patient: one presenting the maximum absorption of the metabolic function of the striatum and two adjacent images. An expert physician determined that our model’s accuracy, precision, recall, and F1-score were 0.98, 0.98, 0.98, and 0.98, respectively. Our TRODAT SPECT model provides an objective, more standardized classification correlating to the severity of PD-related diseases, thereby facilitating clinical diagnosis and preventing observer bias.

Collaborative filtering (CF) has been applied in various domains to resolve problems related to information overload. In a knowledge-intensive environment, most works are processed through teamwork. A user on a team can reference task-related documents from other trusted members to support work on the task. However, the traditional personalised recommender systems no longer meet the demand of teams or groups. Therefore, this work proposes a novel document recommendation method based on a group-based trust model. Our method will analyse the degrees of trust among users in a group and then identify the trustworthy users. The proposed group trust consists of a hybrid personal trust (HPT) model and users’ importance (i.e. users’ activity, similarity and reputation) in a group. Group-based trust is then integrated with the user-based CF to recommend documents to users. The experiments demonstrate that the proposed method can provide better performance than other trust-based recommendation methods; it not only obtains reliable trust values to increase the accuracy of predictions but also enhances the recommendation quality.

New starburst small molecules bearing carbazole end-groups are step-economically synthesized in good isolated yields (70–84%) by sequential C–H activation/arylation reactions. Perovskite solar cells employing these molecules (YC05–08) as hole-transporting materials (HTMs) display a promising power conversion efficiency of up to 17.4%.

The widely deployed Internet-of-Things (IoT) devices provide intelligent services with its cognition capability. Since the IoT data are usually transmitted to the server for recognition (e.g., image classification) due to low computational capability and limited power supply, achieving recognition accuracy under limited bandwidth and noisy channel of wireless networks is a crucial but challenging task. In this paper, we propose a deep learning-constructed joint transmission-recognition scheme for the IoT devices to effectively transmit data wirelessly to the server for recognition, jointly considering transmission bandwidth, transmission reliability, complexity, and recognition accuracy. Compared with other schemes that may be deployed on the IoT devices, i.e., a scheme based on JPEG compression and two compressed sensing-based schemes, the proposed deep neural network-based scheme has much higher recognition accuracy under various transmission scenarios at all signal-to-noise ratios (SNRs). In particular, the proposed scheme maintains good performance at the very low SNR. Moreover, the complexity of the proposed scheme is low, making it suitable for IoT applications. Finally, a transfer learning-based training method is proposed to effectively mitigate the computing burden and reduce the overhead of online training.

Visual recognition is a fast and accurate process. This study adopted a progressive revelation task which demonstrated the dynamic visual process to investigate narrative scene perception. Experiment 1 examined the recognition of character and surroundings that are presented in isolation. The results demonstrated an accumulation of perceptual evidence for recognition as a psychometric function, and character identification was found to be superior to surroundings identification (character advantage effect). Experiment 2 examined the character and surroundings identification in isolated and concurrent presentations. The results showed that less accumulated evidence was required for the concurrent condition than for the isolated condition in the surroundings identification but not in the character identification. The character advantage effect only appeared in the isolated condition. In Experiment 3, both the character and surroundings in semantically consistent or inconsistent scenes were reported....

Financial technology (FinTech) draws much attention in these years, with the advances of machine learning and deep learning. In this work, given historical time series of stock prices of companies, we aim at forecasting upcoming anomalous financial items, i.e., abrupt soaring or diving stocks, in financial time series, and recommending the corresponding stocks to support financial operations. We propose a novel joint convolutional and recurrent neural network model, Financial Event Neural Network (FineNet), to forecast and recommend anomalous stocks. Experiments conducted on the time series of stock prices of 300 well-known companies exhibit the promising performance of FineNet in terms of precision and recall. We build FineNet as a Web platform for live demonstration.

This study applied the modified Cournot duopoly model proposed by Buehler and Schmutzler (2008) and Milliou and Pavlou (2013) to analyze and compare the economic efficiency of the separation model and vertical integration model. For the industry of downstream Research and Development (R&D), this model examines the effect of market size and product substitutability on the economic output. We establish the following results: (i) the benefits of a downstream firm on vertical integration will increase when the product substitutability is lower. (ii) While the market size becomes bigger, the benefit is further enhanced this conclusion. The integration firm will promote investment in R&D to cause intimidation effect. (iii) When the product substitutability is higher to a certain degree, the benefits will also increase. Thus, high degree of product homogeneity and high degree of product heterogeneity are more suitable for vertical integration. In the long run, the industry which will extend or increase in demand suggests a merger as early as possible.

In contrast to the traditional multistep synthesis, we demonstrate herein a two-step synthesis shortcut to triphenylamine-based hole-transporting materials (HTMs) through sequential direct C-H arylations. These hole-transporting molecules are fabricated in perovskite-based solar cells (PSCs) that exhibit promising efficiencies up to 17.69 %, which is comparable to PSCs utilizing commercially available 2,2',7,7'-tetrakis[N,N-di(4-methoxyphenyl)amino]-9,9'-spirobifluorene (spiro-OMeTAD) as the HTM. This is the first report describing the use of step-saving C-H activations/arylations in the facile synthesis of small-molecule HTMs for perovskite solar cells.

This study conducted a questionnaire survey to understand citizen cognition of the stray dog issue in Taipei. This study collected and analyzed 480 effective responses regarding demographic factors, and the cognition and attitudes toward the stray dog issue in Taipei from respondents with or without watching the film Twelve Nights. This work revealed two key findings. First, the respondents' gender, age, education level, and occupation were strongly related to their cognition and attitudes toward stray dogs. Second, the movie Twelve Nights significantly raised the social concern and awareness toward stray dogs issue.

This study examined whether various doses of ethanol induced reward or aversion and then evaluated Grigson’s reward comparison hypothesis (1997). Rats were given a 0.1% saccharin solution (conditioned stimulus 1 [CS1]) 15 min prior to administration of a 0, 0.05, 0.125, 0.20, 0.35, or 0.50 g/kg dose of ethanol (unconditioned stimulus [US]). The rats were then exposed to a paired compartment (CS2) for 30 min. The low dose of 0.05 g/kg ethanol did not induce conditioned suppression (i.e., conditioned taste aversion [CTA]) or conditioned place preference (CPP). The dose of 0.125 g/kg ethanol induced CPP but not CTA. High doses of ethanol, including 0.35 g/kg and 0.50 g/kg, produced CTA but not CPP. The middle dose of 0.20 g/kg ethanol simultaneously induced CTA and CPP. As a result, the reward comparison hypothesis cannot explain the present finding that the middle dose of ethanol induced CTA and CPP. Meanwhile, the high doses of ethanol induced motivationally aversive CTA but not rewarding CPP. The reward comparison hypothesis should be updated further.

A combination of a solution process and the control of the electric potential for magnetism represents a new approach to operating spintronic devices with a highly controlled efficiency and lower power consumption with reduced production cost. As a paradigmatic example, we investigated Co/Pt(111) in the Bloch-wall regime. The depression in coercive force was detected by applying a negative electric potential in an electrolytic solution. The reversible control of coercive force by varying the electric potential within few hundred millivolts is demonstrated. By changing the electric potential in ferromagnetic layers with smaller thicknesses, the efficiency for controlling the tunable coercive force becomes higher. Assuming that the pinning domains are independent of the applied electric potential, an electric potential tuning-magnetic anisotropy energy model was derived and provided insights into our knowledge of the relation between the electric potential tuning coercive force and the thickness of the ferromagnetic layer. Based on the fact that the coercive force can be tuned by changing the electric potential using a solution process, we developed a novel concept of electric-potential-tuned magnetic recording, resulting in a stable recording media with a high degree of writing ability.

How to improve performance of software-as-a-service remains a problem. Drawing from the service-dominant logic, the relational view, and the social exchange theory, we develop a model to delineate the relationships between performance, relationship-specific services, and internal services. The premise of this model is that the service provider's internal services play a key role in motivating the client to collaborate with the vendor. This helps the provider offer relationship-specific services, characterized as knowledge-based and process-based services and the provider's performance relies on these services. Results of empirical testing using responses from 102 firms support the proposed hypotheses. Our findings provide a number of useful implications for research and managers.

This paper presents an innovative Method to Measure Errors Of Mobile Robots Location. The innovative method is composed of a Laser pointer with Optical cross mark and a Grid paper (LOG method). The errors of a mobile robots location are included the precision and accuracy of the translation and rotation. Using the measured errors, the performance of a mobile robot can be evaluated. The Mecanum wheel is a conventional wheel with a series of rollers attached to its circumference and can move in any direction. In this research, the three encoders are used on a mobile robot with three Mecanum wheels. The three Mecanum wheels are mounted on the robot round frame as a base of the moving platform. Mecanum wheels are independently powered using three units of precisian gear DC motors and the wheel/motor assemblies were mounted directly to the robot round frame. The slip occurs always by the rollers attached to conventional wheel's circumference. The force modeling and the innovative evaluating system for the precision and accuracy of the translation and rotation is focused on this research. By using a laser pointer with cross optical mark and a sheet of mm grid paper on the mobile robot, a fast and stable testing set up for the precision and accuracy of the translation and rotation is build. Through LOG method, and according certain testing process the precision and accuracy of the translation and rotation of a mobile robot can be fast evaluated and analyzed, respectively. The experiment result shows that the mobile robot moves sideways the distance from 0.25m to1.5m at the speed 1m/s fast, stable and easily to finish. The error of translation and rotation of the mobile robot can be fast determined respectively. And the performance of a mobile robot can be fast evaluated by every run at the stop localization immediately.

This study aimed to examine which brain oscillatory activities and peripheral physiological measures were influenced by attention levels. A new experimental procedure was designed. Participants were asked to count the number of target events while viewing eight moving white circles. An event occurred when two of the circles changed from white to red or blue. In the low-attention task, similar to a feature search, the target events were defined by color only. In the high-attention task, similar to a conjunction search, the target events were defined by both color and size. In the control task, participants were asked to passively watch the series of events while remembering a number. Based on Feature Integration Theory, our high-attention task would demand more attentional investment than the low-attention task. Given the identical visual stimuli and requirement of keeping a number in working memory for all three tasks, the changes in brain oscillatory activities can be attributed to attention level rather than to perceptual content or memory processes. Peripheral measures such as heart rate, heart rate variability (HRV), respiration rate, eye blinks, and skin conductance level were also evaluated. In comparing the high-attention task with the low-attention task, theta synchronization at the Fz, Cz, and Pz electrodes as a group, alpha2 desynchronization at the Fz, Cz, Pz, and Oz electrodes as a group, and a decrease in the low-frequency component and ratio measure of HRV were evident. These measures are considered to be promising indices for discriminating between attention levels.

Grounded on expectation-confirmation theory (ECT), this study proposes an integrated model aiming to understand how participants' continuance intention of knowledge creation and satisfaction are affected by their perceived identity verification (PIV) and performance expectancy in an online community. To assess the theoretical model, we surveyed two online communities-Baidu (China) and Yahoo knowledge+ (Taiwan), including 213 and 216 useable responses, respectively. Both PIV and performance expectancy are positively associated with continuance intention of knowledge creation. In Baidu community, satisfaction is affected either directly by performance expectancy or indirectly by PIV (through performance expectancy). Our findings have implications for both practice and theory.

Improving the performance of ERP systems remains an important issue. This study examines ERP performance at the post-implementation stage, particularly from the perspective of managerial intervention. Specifically, we proposed that both customization and organizational mechanisms affect intermediate benefits (including coordination improvement and task efficiency), which in turn influence overall benefits. A firm-level survey was used to collect data. Our findings support the proposed hypotheses. We also provide implications for both managers and researchers.

We investigated the attentional capture effect of emotional faces under sufficient or restricted attentional conditions. In a modified visual search paradigm, three kinds of schematic faces (angry, happy, and neutral) served as stimuli. Participants were instructed to search for a target face indicated by a dot and to respond to the dot's position. In this design, the emotional content of the face is task-irrelevant and does not need to be attended. The results of Experiment 1 demonstrate that having an angry face as the target face elicited a faster response than did the neutral target face, and when the angry face is used as a distractor, the response to the target was delayed compared to the response with no such distractor. Experiment 2 included inverted faces to decrease emotional content; results showed that inversion of the faces reduced the effect of angry faces on the search performance. When attention was cued to a specific area in Experiment 3, the effect of angry faces outside of the cued area became weaker. In conclusion, the results indicate that a task-irrelevant angry face can capture attention beyond top-down control, but this effect is modulated by the availability of attentional resources. (PsycINFO Database Record (c) 2011 APA, all rights reserved).

Grounded on both social cognitive theory and expectation-disconfirmation theory (EDT), this study aims to investigate how individuals’ continuance intention of knowledge creation and satisfaction are affected by their perceived identity verification (PIV) and performance expectancy. In addition, we also explore how the foregoing relationships are contingent on trust—affective trust and cognitive trust. To assess the theoretical model, we surveyed two online communities—Baidu (Cina) and Yahoo knowledge+ (Taiwan), including 213 and 216 useful responses respectively. Both PIV and performance expectancy are positively associated with continuance intention of knowledge creation. While performance expectancy has a significant impact on satisfaction, in Baidu community, PIV affects satisfaction indirectly (through performance expectancy) rather than directly. Finally, neither the relationships between PIV and continuance intention nor those between performance and continuance intention are contingent on trust based on our findings. This study provides a new insight into how knowledge creation can be facilitated by PIV and performance expectancy, and how the foregoing relationships are contingent on trust. Our findings have implications for both practice and theory.

How to achieve knowledge creation effectively remains a problem, this becomes even more difficult in a loosely knitted community such as networks of practice. This is because knowledge creation entails not only knowledge acquisition but also knowledge conversion. Both of them necessitate complicated social interactions. Since individuals are less likely to share precious insights with strangers, knowledge creation seems infeasible in networks of practice. To address this in a more comprehensive way, this study integrates social capital and TPB (theory of planned behavior) as well as individual motivations. The purpose of this paper is to delineate the relationships among salient antecedents of social capital and TPB, from which the factors that stimulate or impede individuals' knowledge creation can be specified. A survey instrument was used to collect data from a famous network of practice. 202 useful responses were employed to test our model. 11 of the 17 hypotheses are supported by the empirical findings. This paper also provides implications for both academics and practitioners.