Search

The mesoscale interaction between sea surface temperature (SST) and wind is a crucial factor influencing oceanic and atmospheric conditions. To investigate the mesoscale coupling characteristics of the Yellow Sea and East China Sea, we applied a locally weighted regression filtering method to extract mesoscale signals from Quik-SCAT wind field data and AMSR-E SST data and found that the mesoscale coupling intensity is stronger in the Yellow Sea during the spring and winter seasons. We calculated the mesoscale coupling coefficient to be approximately 0.009 N·m−2/°C. Subsequently, the Tikhonov regularization method was used to establish a mesoscale empirical coupling model, and the feedback effect of mesoscale coupling on the ocean was studied. The results show that the mesoscale SST–wind field coupling can lead to the enhancement of upwelling in the offshore area of the East China Sea, a decrease in the upper ocean temperature, and an increase in the eddy kinetic energy in the Yellow Sea. Diagnostic analyses suggested that mesoscale coupling-induced variations in horizontal advection and surface heat flux contribute most to the variation in SST. Moreover, the increase in the wind energy input to the eddy is the main factor explaining the increase in the eddy kinetic energy. [ABSTRACT FROM AUTHOR]

The article focuses on introducing a Dual Graph Ensemble Learning method for Knowledge Tracing (DGEKT) to improve knowledge tracing by incorporating heterogeneous exercise-concept associations and interaction transitions. Topics include hypergraph modeling, directed graph modeling, and online knowledge distillation to enhance prediction accuracy and ensemble modeling for better judgment quality in knowledge tracing tasks.

In this study, an ensemble forecasting model for in situ wind speed and wave height was developed using the Coupled Ocean–Atmosphere–Wave–Sediment Transport (COAWST) model. This model utilized four bias correction algorithms—Model Output Statistics (MOS), Back Propagation Neural Network (BPNN), Long Short-Term Memory (LSTM) neural network, and Convolutional Neural Network (CNN)—to construct ensemble forecasts. The training data were derived from the COAWST simulations of one year and observations from three buoy stations (Laohutan, Zhifudao, and Lianyungang) in the Yellow Sea and Bohai Sea. After the optimization of the bias correction model training, the subsequent evaluations on the ensemble forecasts showed that the in situ forecasting accuracy of wind speed and wave height was significantly improved. Although there were some uncertainties on bias correction performance levels for individual algorithms, the uncertainties were greatly reduced by the ensemble forecasts. Depending on the dynamic weight assignment, the ensemble forecasts presented a stable performance even when the corrected forecasts by three algorithms had an obvious negative bias. Specifically, the ensemble forecasting bias was found with a mean reduction of about 96%~99% and 91%~95% for wind speed and wave height, and a reduction of about 91%~98% and 16%~54% during the period of Typhoon "Muifa". For the four correction algorithms, the performance of bias correction was not directly related to the algorithm complexity. However, the strategies with more complex algorithms (i.e., CNN) were more conservative, and simple algorithms (i.e., MOS) might have induced unstable performance levels despite their lower bias in some cases. [ABSTRACT FROM AUTHOR]

The accuracy of stock index prediction is of great significance to national economic development. However, because of the nonlinearity and long-term dependence of stock index data, effective prediction of future stock index price becomes a challenge. In order to solve the above problems, this paper proposes a research method of stock index time series prediction based on ensemble learning model. This method first uses an Adaboost.R2 algorithm to iteratively train multiple LSTM models and then integrates these LSTM models based on the parameters obtained by iterative training. Finally, it uses the ensemble model to predict stock index time series data. This paper uses the Shanghai Composite index, CSI 300 index and Shenzhen Composite index as experimental data sets, and uses the BP model, CNN model and LSTM model as comparative models to conduct an experimental analysis. The experimental results show that the new ensemble learning model proposed in this paper has certain advantages in the research of stock index time series prediction. [ABSTRACT FROM AUTHOR]

Multi-task learning in Convolutional Neural Networks (CNNs) has led to remarkable success in a variety of applications of computer vision. Towards effective multi-task CNN architectures, recent studies automatically learn the optimal combinations of task-specific features at single network layers. However, they generally learn an unchanged operation of feature combination after training, regardless of the characteristic changes of task-specific features across different inputs. In this paper, we propose a novel Adaptive Feature Aggregation (AFA) layer for multi-task CNNs, in which a dynamic aggregation mechanism is designed to allow each task adaptively determines the degree to which the knowledge sharing or preserving between tasks is needed based on the characteristics of inputs. We introduce two types of aggregation modules to the AFA layer, which realize the adaptive feature aggregation by capturing the feature dependencies of different tasks along the channel and spatial axes, respectively. The AFA layer is a plug-and-play component with low parameter and computation overheads, and can be trained end-to-end along with backbone networks. For both pixel-level and image-level tasks, we empirically show that our approach substantially outperforms the previous state-of-the-art methods of multi-task CNNs. The code and models are available at https://github.com/zhenshen-mla/AFANet. [ABSTRACT FROM AUTHOR]

Time series prediction has been studied for decades due to its potential in a wide range of applications. As one of the most popular technical indicators, moving average summarizes the overall changing patterns over a past period and is frequently used to predict the future trend of time series. However, traditional moving average indicators are calculated by averaging the time series data with equal or predefined weights, and ignore the subtle difference in the importance of different time steps. Moreover, unchanged data weights will be applied across different time series, regardless of the differences in their inherent characteristics. In addition, the interaction between different dimensions of different indicators is ignored when using the moving averages of different scales to predict future trends. In this paper, we propose a learning-based moving average indicator, called the self-attentive moving average (SAMA). After encoding the input signals of time series based on recurrent neural networks, we introduce the self-attention mechanism to adaptively determine the data weights at different time steps for calculating the moving average. Furthermore, we use multiple self-attention heads to model the SAMA indicators of different scales, and finally combine them through a bilinear fusion network for time series prediction. Extensive experiments on two real-world datasets demonstrate the effectiveness of our approach. The data and codes of our work have been released. [ABSTRACT FROM AUTHOR]

The BP neural network model is a hot issue in recent academic research, and it has been successfully applied to many other fields, but few researchers apply the BP neural network model to the field of automobile insurance. The main method that has been used in the prediction of the total claim amount in automobile insurance is the generalized linear model, where the BP neural network model could provide a different approach to estimate the total claim loss. This paper uses a genetic algorithm to optimize the structure of the BP neural network at first, and the calculation speed is significantly improved. At the same time, by considering the overfitting problem, an early stop method is introduced to avoid the overfitting problem. In the model, a three-layer BP neural network model, which includes the input layer, hidden layer, and output layer, is trained. With consideration of various factors, a total claim amount prediction model is established, and the trained BP neural network model is used to predict the total claim amount of automobile insurance based on the data of the training set. The results show that the accuracy of the prediction by using the BP neural network model to both the data of Shandong Province and to the data of six cities is over 95%. Then, the predicted total claim amount is used to calculate premiums for five cities in Shandong Province according to credibility theory. The results show that the average premium of the five cities is slightly higher than the actual claim amount of the city. The combination of BP neural network and credibility theory can perform accurate claim amount estimation and pricing for automobile insurance, which can effectively improve the current situation of the automobile insurance business and promote the development of insurance industry. [ABSTRACT FROM AUTHOR]

The majority of existing methods for saliency detection based on low-level features failed to uniformly highlight the salient-object regions. In order to improve the accuracy and consistency of generated saliency maps, we propose a novel and efficient framework by combining low-level saliency priors and local similarity cues for image saliency detection. Firstly, we construct a multiple low-level prior map using location prior, color prior and background prior. Then, the prior maps employ a propagation mechanism based on Cellular Automata to enforce relevance of similar regions as a local similarity cue. Finally, a principle refinement framework by integrating multi-level prior maps and local similarity cue map are used to obtain an ultimate high-quality saliency map. Extensive experiments on publicly available datasets show that our designed approach is capable of producing accurate saliency maps compared with those generated results by the state-of-the-art saliency-detection methods. [ABSTRACT FROM AUTHOR]

Multi-view hashing can well support large-scale near-duplicate video retrieval, due to its desirable advantages of mutual reinforcement of multiple features, low storage cost, and fast retrieval speed. However, there are still two limitations that impede its performance. First, existing methods only consider local structures in multiple features. They ignore the global structure that is important for near-duplicate video retrieval, and cannot fully exploit the dependence and complementarity of multiple features. Second, existing works always learn hashing functions bit by bit, which unfortunately increases the time complexity of hash function learning. In this paper, we propose a supervised hashing scheme, termed as joint multi-view hashing (JMVH), to address the aforementioned problems. It jointly preserves the global and local structures of multiple features while learning hashing functions efficiently. Specially, JMVH considers features of video as items, based on which an underlying Hamming space is learned by simultaneously preserving their local and global structures. In addition, a simple but efficient multi-bit hash function learning based on generalized eigenvalue decomposition is devised to learn multiple hash functions within a single step. It can significantly reduce the time complexity of conventional hash function learning processes that sequentially learn multiple hash functions bit by bit. The proposed JMVH is evaluated on two public databases: CC_WEB_VIDEO and UQ_VIDEO. Experimental results demonstrate that the proposed JMVH achieves more than a 5 percent improvement compared to several state-of-the-art methods which indicates the superior performance of JMVH. [ABSTRACT FROM AUTHOR]

In higher education, the initial studying period of each course plays a crucial role for students, and seriously influences the subsequent learning activities. However, given the large size of a course’s students at universities, it has become impossible for teachers to keep track of the performance of individual students. In this circumstance, an academic early warning system is desirable, which automatically detects students with difficulties in learning (i.e., at-risk students) prior to a course starting. However, previous studies are not well suited to this purpose for two reasons: 1) they have mainly concentrated on e-learning platforms, e.g., massive open online courses (MOOCs), and relied on the data about students’ online activities, which is hardly accessed in traditional teaching scenarios; and 2) they have only made performance prediction when a course is in progress or even close to the end. In this paper, for traditional classroom-teaching scenarios, we investigate the task of pre-course student performance prediction, which refers to detecting at-risk students for each course before its commencement. To better represent a student sample and utilize the correlations among courses, we cast the problem as a multi-instance multi-label (MIML) problem. Besides, given the problem of data scarcity, we propose a novel multi-task learning method, i.e., MIML-Circle, to predict the performance of students from different specialties in a unified framework. Extensive experiments are conducted on five real-world datasets, and the results demonstrate the superiority of our approach over the state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Image aesthetics assessment has emerged as a hot topic in recent years due to its potential in numerous high-level vision applications. In this paper, distinguished from existing studies relying on a single label, we propose quantifying image aesthetics by a distribution over multiple quality levels. The distribution-based representation characterizes the disagreement among users’ aesthetic preferences regarding the same image, and is also compatible with the traditional task of aesthetic label prediction. Our framework is developed based on fully convolutional networks and enables inputs of varying sizes. In this way, we circumvent the fixed-size constraint of prevalent convolutional neural networks, and avoid the risk of impairing the intrinsic aesthetic appeal of images. Moreover, given the fact that aesthetic perceiving is coupled with semantic understanding, we present a novel semantic-aware hybrid NEtwork (SANE), which harvests the information from object categorization and scene recognition to enhance image aesthetics assessment. Experiments on two benchmark datasets have well verified the effectiveness of our approach in both scenarios of aesthetic distribution prediction and aesthetic label prediction, and highlighted the benefits of input preserving as well as semantic understanding for images. [ABSTRACT FROM AUTHOR]

This paper studies the statistical estimation of the Gerber-Shiu discounted penalty functions in a general spectrally negative Lévy risk model. Suppose that the claims process and the surplus process can be observed at a sequence of discrete time points. Using the observed data, the Gerber-Shiu functions are estimated by the Laguerre series expansion method. Consistent properties are studied under the large sample setting, and simulation results are also presented when the sample size is finite. [ABSTRACT FROM AUTHOR]

As mobile device penetration increases, it has become pervasive for images to be associated with locations in the form of geotags. Geotags bridge the gap between the physical world and the cyberspace, giving rise to new opportunities to extract further insights into user preferences and behaviors. In this article, we aim to exploit geotagged photos from online photo-sharing sites for the purpose of personalized Point-of-Interest (POI) recommendation. Owing to the fact that most users have only very limited travel experiences, data sparseness poses a formidable challenge to personalized POI recommendation. To alleviate data sparseness, we propose to augment current collaborative filtering algorithms along from multiple perspectives. Specifically, hybrid preference cues comprising user-uploaded and user-favored photos are harvested to study users’ tastes. Moreover, heterogeneous high-order relationship information is jointly captured from user social networks and POI multimodal contents with hypergraph models. We also build upon the matrix factorization algorithm to integrate the disparate sources of preference and relationship information, and apply our approach to directly optimize user preference rankings. Extensive experiments on a large and publicly accessible dataset well verified the potential of our approach for addressing data sparseness and offering quality recommendations to users, especially for those who have only limited travel experiences. [ABSTRACT FROM AUTHOR]

Content-based near-duplicate video detection (NDVD) is essential for effective search and retrieval, and robust video fingerprinting is a good solution for NDVD. Most existing video fingerprinting methods use a single feature or concatenate different features to generate video fingerprints, and show good performance under single-mode modifications such as noise addition and blurring. However, when they suffer combined modifications, the performance is degraded to a certain extent because such features cannot characterize the video content completely. By contrast, the assistance and consensus among different features can improve the performance of video fingerprinting. Therefore, in the present study, we mine the assistance and consensus among different features based on a tensor model, and we present a new comprehensive feature to fully use them in the proposed video fingerprinting framework. We also analyze what the comprehensive feature really is for representing the original video. In this framework, the video is initially set as a high-order tensor that consists of different features, and the video tensor is decomposed via the Tucker model with a solution that determines the number of components. Subsequently, the comprehensive feature is generated by the low-order tensor obtained from tensor decomposition. Finally, the video fingerprint is computed using this feature. A matching strategy used for narrowing the search is also proposed based on the core tensor. The robust video fingerprinting framework is resistant not only to single-mode modifications but also to their combination. [ABSTRACT FROM PUBLISHER]

High quality tags play a critical role in applications involving online multimedia search, such as social image annotation, sharing and browsing. However, user-generated tags in real world are often imprecise and incomplete to describe the image contents, which severely degrades the performance of current search systems. To improve the descriptive powers of social tags, a fundamental issue is tag relevance learning, which concerns how to interpret the relevance of a tag with respect to the contents of an image effectively. In this paper, we investigate the problem from a new perspective of learning to rank, and develop a novel approach to facilitate tag relevance learning to directly optimize the ranking performance of tag-based image search. Specifically, a supervision step is introduced into the neighbor voting scheme, in which the tag relevance is estimated by accumulating votes from visual neighbors. Through explicitly modeling the neighbor weights and tag correlations, the risk of making heuristic assumptions is effectively avoided. Besides, our approach does not suffer from the scalability problem since a generic model is learned that can be applied to all tags. Extensive experiments on two benchmark datasets in comparison with the state-of-the-art methods demonstrate the promise of our approach. [ABSTRACT FROM AUTHOR]

In this paper, we present a hybrid relevant-diverse image re-ranking approach that combines the strengths of two previous methods: the reciprocal election algorithm proposed by R. van Leuken et al. and the greedy search algorithm proposed by T. Deselaers et al. Our approach is a cluster-based re-ranking method. We select several candidate representatives based on the reciprocal election algorithm and employ a bounded greedy search algorithm to find the most relevant-diverse one as the cluster representative. We fuse multiple features to calculate image similarity, including colour and shape and especially topic content features, and discuss the benefits of integrating different features. In addition, we quantitatively evaluate our approach on a real-world Web image data set and obtain results which outperform the state of the art. [ABSTRACT FROM AUTHOR]

Taking mortality distribution, surrender value, and tax relief factors into consideration, the authors construct an actuarial model for the influence of personal income tax deferred commercial pension insurance on changes in personal pension wealth and adopts a numerical simulation to deliver the corresponding changes in personal pension wealth to different initial insured age and different initial insured annual salary. In order to better measure the security level of the commercial pension insurance, the model for the net replacement rate of pension of the commercial pension insurance was further constructed. The results show that the effect of participating in the personal income tax deferred commercial pension insurance on the present value of personal pension wealth depends on the combined action of the initial insured age and the initial annual salary. Under the same insured age, because men retire later and work longer than women, men can obtain a higher accumulation of personal pension wealth than women. For insured persons with different income levels, high-income groups can obtain higher personal pension wealth growth, and although low-income groups cannot obtain higher personal pension wealth growth, they can obtain a significant increase in the pension replacement rate by participating in the insurance, thereby better guaranteeing their living standards after retirement. Regardless of the income level, tax relief can be obtained once participating in the insurance, but the value may vary. The optimal tax-saving age for men is 23 years old, and for women 25 years old. [ABSTRACT FROM AUTHOR]

In this paper, we assume that the reserve level of an insurance company can only be observed at discrete time points, then a new risk model is proposed by introducing a periodic capital injection strategy and a barrier dividend strategy into the classical risk model. We derive the equations and the boundary conditions satisfied by the Gerber-Shiu function, the expected discounted capital injection function and the expected discounted dividend function by assuming that the observation interval and claim amount are exponentially distributed, respectively. Numerical examples are also given to further analyze the influence of relevant parameters on the actuarial function of the risk model. [ABSTRACT FROM AUTHOR]

Recently, the valuation of variable annuity products has become a hot topic in actuarial science. In this paper, we use the Fourier cosine series expansion (COS) method to value the guaranteed minimum death benefit (GMDB) products. We first express the value of GMDB by the discounted density function approach, then we use the COS method to approximate the valuation Equations. When the distribution of the time-until-death random variable is approximated by a combination of exponential distributions and the price of the fund is modeled by an exponential Lévy process, explicit equations for the cosine coefficients are given. Some numerical experiments are also made to illustrate the efficiency of our method. [ABSTRACT FROM AUTHOR]

The absolute ruin insurance risk model is modified by including some valuable market economic information factors, such as credit interest, debit interest and dividend payments. Such information is especially important for insurance companies to control risks. We further assume that the insurance company is able to finance and continue to operate when its reserve is negative. We investigate the integro-differential equations for some interest actuarial diagnostics. We also provide numerical examples to explain the effects of relevant parameters on actuarial diagnostics. [ABSTRACT FROM AUTHOR]

• A unified weakly-supervised hash learning framework for social image retrieval. • An efficient discrete hash optimization technique to iteratively solve hash codes. • Experiments demonstrate the superiority of the proposed approach. Hashing has been widely exploited for information retrieval recently, because of its high computation efficiency and low storage cost. However, many existing hashing methods cannot perform well on large-scale social image retrieval, due to the relaxed hash optimization and the lack of supervised semantic labels. In this paper, we propose an efficient Weakly-supervised Discrete Hashing (WDH) to solve the limitations. We formulate a unified weakly-supervised hash learning framework. It could effectively enrich the semantics of image hash codes with the freely obtained user-provided social tags and simultaneously remove their involved adverse noises. Furthermore, instead of relaxed hash optimization, we propose an efficient discrete hash optimization method based on Augmented Lagrangian Multiplier (ALM) to directly solve the hash codes without quantization information loss. Experiments on two standard social image datasets demonstrate the superior performance of the proposed method compared with several state-of-the-art hashing techniques. [ABSTRACT FROM AUTHOR]

• We introduce and compare six typically fusion strategies for saliency detection. • Overall, the non-linear fusion strategies outperform the linear-based strategies. • The devised non-linear fusion strategy based on least square method is optimum. Saliency detection is a hot topic in the field of computer vision and pattern recognition, thus plenty of saliency models have been exploited and extended to various visual correspondence applications. Currently, it's still confronted with a variety of obstacles and challenges, although it has been studied for decades. With the progress of saliency detection, different computational models and salient features have been proposed and some of them improve and compensate the deficiencies of the others. In this paper, we focus on investigating the salient feature fusion strategies in human visual attention mechanism for saliency detection (e.g., linear and non-linear), in order to efficiently incorporate various salient cues for achieving a better result. Based on the complementary principle, we firstly construct a saliency map based on the information of the image background. Then, we generate a supplemental saliency map from the compactness saliency features. Finally, we evaluate the performance of six individual fusion strategies including both linear and non-linear models in terms of three publicly available image datasets. Experimental results show that our designed non-linear fusion strategy based on least-square method outperforms the other fusion strategies in saliency detection. [ABSTRACT FROM AUTHOR]