Explore the vast range of titles available.

Large volumes of automatic identification system (AIS) data provide new ideas and methods for ship data mining and navigation behavior pattern analysis. However, large volumes of big data have low unit values, resulting in the need for large-scale computing, storage, and display. Learning efficiency is low and learning direction is blind and untargeted. Therefore, key feature point (KFP) extraction from the ship trajectory plays an important role in fields such as ship navigation behavior analysis and big data mining. In this paper, we propose a ship spatiotemporal KFP online extraction algorithm that is applied to AIS trajectory data. The sliding window algorithm is modified for application to ship navigation angle deviation, position deviation, and the spatiotemporal characteristics of AIS data. Next, in order to facilitate the subsequent use of the algorithm, a recommended threshold range for the corresponding two parameters is discussed. Finally, the performance of the proposed method is compared with that of the Douglas–Peucker (DP) algorithm to assess its feature extraction accuracy and operational efficiency. The results show that the proposed improved sliding window algorithm can be applied to rapidly and easily extract the KFPs from AIS trajectory data. This ability provides significant benefits for ship traffic flow and navigational behavior learning.

In this paper, a dual online subspace-based learning method called dual-generalized discriminative common vectors (Dual-GDCV) is presented. The method extends incremental GDCV by exploiting simultaneously both the concepts of incremental and decremental learning for supervised feature extraction and classification. Our methodology is able to update the feature representation space without recalculating the full projection or accessing the previously processed training data. It allows both adding information and removing unnecessary data from a knowledge base in an efficient way, while retaining the previously acquired knowledge. The proposed method has been theoretically proved and empirically validated in six standard face recognition and classification datasets, under two scenarios: (1) removing and adding samples of existent classes, and (2) removing and adding new classes to a classification problem. Results show a considerable computational gain without compromising the accuracy of the model in comparison with both batch methodologies and other state-of-art adaptive methods.

Abstract Due to the current video key frame extraction algorithm is affected by the lens conversion, the extraction accuracy is poor. For this reason, a precise extraction algorithm of video action key frames for online aerobics teaching is studied. According to the color components of the video color RGB space, in order to ensure that the color distance is suitable for human vision. A non-uniform quantized HSV space method is adopted, and a one-dimensional feature vector is introduced to convert the online teaching video of aerobics into a one-dimensional histogram of 72 bins, which realizes the segmentation of video shots and reduces the impact of shot conversion. Sort the gray values of the histogram pixels of the video after the segmentation is completed, and construct the dynamic frames of the aerobics online teaching video. Sequence search constructs the processing dynamic frame, extracts the feature vector of the video sequence, and uses the multi-layer core aggregation algorithm to extract the key frame of the video action according to the extracted feature vector. Experimental results show that the algorithm can effectively extract the key frames of aerobics video action, the fidelity of the extracted key frames is higher than 0.9, and the precision and recall are both higher than 99%.

In this paper, we propose a real-time online shopper behavior analysis system consisting of two modules which simultaneously predicts the visitor’s shopping intent and Web site abandonment likelihood. In the first module, we predict the purchasing intention of the visitor using aggregated pageview data kept track during the visit along with some session and user information. The extracted features are fed to random forest (RF), support vector machines (SVMs), and multilayer perceptron (MLP) classifiers as input. We use oversampling and feature selection preprocessing steps to improve the performance and scalability of the classifiers. The results show that MLP that is calculated using resilient backpropagation algorithm with weight backtracking produces significantly higher accuracy and F1 Score than RF and SVM. Another finding is that although clickstream data obtained from the navigation path followed during the online visit convey important information about the purchasing intention of the visitor, combining them with session information-based features that possess unique information about the purchasing interest improves the success rate of the system. In the second module, using only sequential clickstream data, we train a long short-term memory-based recurrent neural network that generates a sigmoid output showing the probability estimate of visitor’s intention to leave the site without finalizing the transaction in a prediction horizon. The modules are used together to determine the visitors which have purchasing intention but are likely to leave the site in the prediction horizon and take actions accordingly to improve the Web site abandonment and purchase conversion rates. Our findings support the feasibility of accurate and scalable purchasing intention prediction for virtual shopping environment using clickstream and session information data.

Previous literature has shown that music preferences (and thus preferred musical features) differ depending on the listening context and reasons for listening (RL). Yet, to our knowledge no research has investigated how features of music that people dance or move to relate to particular RL. Consequently, in two online surveys, participants (N = 173) were asked to name songs they move to (“dance music”). Additionally, participants (N = 105) from Survey 1 provided RL for their selected songs. To investigate relationships between the two, we first extracted audio features from dance music using the Spotify API and compared those features with a baseline dataset that is considered to represent music in general. Analyses revealed that, compared to the baseline, the dance music dataset had significantly higher levels of energy, danceability, valence, and loudness, and lower speechiness, instrumentalness and acousticness. Second, to identify potential subgroups of dance music, a cluster analysis was performed on its Spotify audio features. Results of this cluster analysis suggested five subgroups of dance music with varying combinations of Spotify audio features: “fast-lyrical”, “sad-instrumental”, “soft-acoustic”, “sad-energy”, and “happy-energy”. Third, a factor analysis revealed three main RL categories: “achieving self-awareness”, “regulation of arousal and mood”, and “expression of social relatedness”. Finally, we identified variations in people’s RL ratings for each subgroup of dance music. This suggests that certain characteristics of dance music are more suitable for listeners’ particular RL, which shape their music preferences. Importantly, the highest-rated RL items for dance music belonged to the “regulation of mood and arousal” category. This might be interpreted as the main function of dance music. We hope that future research will elaborate on connections between musical qualities of dance music and particular music listening functions.

Handwriting recognition is one of the challenging tasks in the area of pattern recognition and machine learning. Handwriting recognition has two flavors, namely, Offline Handwriting Recognition and Online Handwriting Recognition. Though, saturation level has been achieved in machine printed (Offline) character recognition. Presently, due to dramatical development in IT sector, touch-based devices are available in the market with efficient processing capabilities. With this revolution, research in the area of handwriting recognition has become more popular in real-time (Online) mode. In this paper, a comprehensive review has been reported for online handwriting recognition of non-Indic and Indic scripts. The six non-Indic-scripts and eight Indic script namely, Arabic, Chinese, Japanese, Persian, Roman, Thai, and, Assamese, Bangla, Devanagari, Gurmukhi, Kannada, Malayalam, Tamil, Telugu, respectively have been considered in this article. This study comprises introduction of online handwriting recognition process, various challenges, motivations, feature extraction, and classification methodologies, used for recognizing the various scripting languages. Moreover, an effort has been made to provide the list of publicly available online handwritten dataset for various scripting languages. This study also provides the recognition and beneficial assistance to the novice researchers in field of handwriting recognition by providing a nut shell studies of various feature extraction strategies and classification techniques, used for the recognition of both Indic and non-Indic scripts.

PurposeThis paper aims to reveal the factors patients consider when choosing a doctor for consultation on an online medical consultation (OMC) platform and how these factors influence doctors' consultation volumes.Design/methodology/approachIn Study 1, influencing factors reflected as service features were identified by applying a feature extraction method to physician reviews, and the importance of each feature was determined based on word frequencies and the PageRank algorithm. Sentiment analysis was used to analyze patient satisfaction with each service feature. In Study 2, regression models were used to analyze the relationships between the service features obtained from Study 1 and the doctor's consultation volume.FindingsThe study identified 14 service features of patients' concerns and found that patients mostly care about features such as trust, phraseology, overall service experience, word of mouth and personality traits, all of which describe a doctor's soft skills. These service features affect patients' trust in doctors, which, in turn, affects doctors' consultation volumes.Originality/valueThis research is important as it informs doctors about the features they should improve, to increase their consultation volume on OMC platforms. Furthermore, it not only enriches current trust-related research in the field of OMC, which has a certain reference significance for subsequent research on establishing trust in online doctor–patient relationships, but it also provides a reference for research concerning the antecedents of trust in general.

The usage of anomaly detection is of critical importance to numerous domains, including structural health monitoring (SHM). In this study, we examine an online setting for damage detection in the Z24 bridge. We evaluate and compare the performance of the elliptic envelope, incremental one-class support vector classification, local outlier factor, half-space trees, and entropy-guided envelopes. Our findings demonstrate that XGBoost exhibits enhanced performance in identifying a limited set of significant features. Additionally, we present a novel approach to manage drift through the application of entropy measures to structural state instances. The study is the first to assess the applicability of one-class classification for anomaly detection on the short-term structural health data of the Z24 bridge.

The control of complex systems is of critical importance in many branches of science, engineering, and industry, many of which are governed by nonlinear partial differential equations. Controlling an unsteady fluid flow is particularly important, as flow control is a key enabler for technologies in energy (e.g., wind, tidal, and combustion), transportation (e.g., planes, trains, and automobiles), security (e.g., tracking airborne contamination), and health (e.g., artificial hearts and artificial respiration). However, the high-dimensional, nonlinear, and multi-scale dynamics make real-time feedback control infeasible. Fortunately, these high-dimensional systems exhibit dominant, low-dimensional patterns of activity that can be exploited for effective control in the sense that knowledge of the entire state of a system is not required. Advances in machine learning have the potential to revolutionize flow control given its ability to extract principled, low-rank feature spaces characterizing such complex systems. We present a novel deep learning model predictive control framework that exploits low-rank features of the flow in order to achieve considerable improvements to control performance. Instead of predicting the entire fluid state, we use a recurrent neural network (RNN) to accurately predict the control relevant quantities of the system, which are then embedded into an MPC framework to construct a feedback loop. In order to lower the data requirements and to improve the prediction accuracy and thus the control performance, incoming sensor data are used to update the RNN online. The results are validated using varying fluid flow examples of increasing complexity.

Since the outbreak of COVID-19, the development of dashboards as dynamic, visual tools for communicating COVID-19 data has surged worldwide. Dashboards can inform decision-making and support behavior change. To do so, they must be actionable. The features that constitute an actionable dashboard in the context of the COVID-19 pandemic have not been rigorously assessed. The aim of this study is to explore the characteristics of public web-based COVID-19 dashboards by assessing their purpose and users (\"why\"), content and data (\"what\"), and analyses and displays (\"how\" they communicate COVID-19 data), and ultimately to appraise the common features of highly actionable dashboards. We conducted a descriptive assessment and scoring using nominal group technique with an international panel of experts (n=17) on a global sample of COVID-19 dashboards in July 2020. The sequence of steps included multimethod sampling of dashboards; development and piloting of an assessment tool; data extraction and an initial round of actionability scoring; a workshop based on a preliminary analysis of the results; and reconsideration of actionability scores followed by joint determination of common features of highly actionable dashboards. We used descriptive statistics and thematic analysis to explore the findings by research question. A total of 158 dashboards from 53 countries were assessed. Dashboards were predominately developed by government authorities (100/158, 63.0%) and were national (93/158, 58.9%) in scope. We found that only 20 of the 158 dashboards (12.7%) stated both their primary purpose and intended audience. Nearly all dashboards reported epidemiological indicators (155/158, 98.1%), followed by health system management indicators (85/158, 53.8%), whereas indicators on social and economic impact and behavioral insights were the least reported (7/158, 4.4% and 2/158, 1.3%, respectively). Approximately a quarter of the dashboards (39/158, 24.7%) did not report their data sources. The dashboards predominately reported time trends and disaggregated data by two geographic levels and by age and sex. The dashboards used an average of 2.2 types of displays (SD 0.86); these were mostly graphs and maps, followed by tables. To support data interpretation, color-coding was common (93/158, 89.4%), although only one-fifth of the dashboards (31/158, 19.6%) included text explaining the quality and meaning of the data. In total, 20/158 dashboards (12.7%) were appraised as highly actionable, and seven common features were identified between them. Actionable COVID-19 dashboards (1) know their audience and information needs; (2) manage the type, volume, and flow of displayed information; (3) report data sources and methods clearly; (4) link time trends to policy decisions; (5) provide data that are \"close to home\"; (6) break down the population into relevant subgroups; and (7) use storytelling and visual cues. COVID-19 dashboards are diverse in the why, what, and how by which they communicate insights on the pandemic and support data-driven decision-making. To leverage their full potential, dashboard developers should consider adopting the seven actionability features identified.