Weighted nuclear norm minimization (WNNM) has produced remarkable denoising results; however, it still has some limitations, including only being able to measure the similarity of noisy patches by Euclidean distance, fixing the feedback proportion of method noise for all noise levels, and establishing an inflexible number of iterations for each image. In this paper, three strategies are used to improve the abovementioned shortcomings. The first strategy is to calculate the correlation coefficient using Grey theory, and it is combined with Euclidean distance as the final similarity measure value. The second strategy is to adaptively add the feedback coefficient of the method noise according to various noise levels. The last strategy is to apply a stopping criterion based on residual noise to the iteration process. Experimental results show our method provides better results compared to various state-of-the-art algorithms.
Many aerial images with similar appearances have different but correlated scene labels, which causes the label ambiguity. Label distribution learning (LDL) can express label ambiguity by giving each sample a label distribution. Thus, a sample contributes to the learning of its ground-truth label as well as correlated labels, which improve data utilization. LDL has gained success in many fields, such as age estimation, in which label ambiguity can be easily modeled on the basis of the prior knowledge about local sample similarity and global label correlations. However, LDL has never been applied to scene classification, because there is no knowledge about the local similarity and label correlations and thus it is hard to model label ambiguity. In this paper, we uncover the sample neighbors that cause label ambiguity by jointly capturing the local similarity and label correlations and propose neighbor-based LDL (N-LDL) for aerial scene classification. We define a subspace learning problem, which formulates the neighboring relations as a coefficient matrix that is regularized by a sparse constraint and label correlations. The sparse constraint provides a few nearest neighbors, which captures local similarity. The label correlations are predefined according to the confusion matrices on validation sets. During subspace learning, the neighboring relations are encouraged to agree with the label correlations, which ensures that the uncovered neighbors have correlated labels. Finally, the label propagation among the neighbors forms the label distributions, which leads to label smoothing in terms of label ambiguity. The label distributions are used to train convolutional neural networks (CNNs). Experiments on the aerial image dataset (AID) and NWPU_RESISC45 (NR) datasets demonstrate that using the label distributions clearly improves the classification performance by assisting feature learning and mitigating over-fitting problems, and our method achieves state-of-the-art performance.
Offline stores are seriously challenged by online shops. To attract more customers to compete with online shops, the patterns of customer flows and their influence factors are important knowledge. To address this issue, we collected indoor positioning data of 534,641 and 59,160 customers in two shopping malls (i.e., Dayuecheng (DYC) in Beijing and Longhu (LH) in Chongqing, China) for one week, respectively. The temporal patterns of the customer flows show that (1) total customer flows are high on weekends and low midweek and (2) peak hourly flow is related to mealtimes for LH and only on weekdays for DYC. The difference in temporal patterns between the two malls may be attributed to the difference in their locations. The customer flows to stores reveal that the customer flows to clothing, food and general stores are the highest; specifically, in DYC, the order is clothing, food and general, while in LH, it is food, clothing and general. To identify the factors influencing customer flow, we applied linear regression to the inflow density of stores (customers per square meter) of two major classes (clothing and food stores), with 10 locational and social factors as independent variables. The results indicate that flow density is significantly influenced by store location, visibility (except for food stores in DYC) and reputation. Besides, the difference between the two store classes is that clothing stores are influenced by more convenience factors, including distance to an elevator and distance to the floor center (only for LH). Overall, the two shopping malls demonstrate similar customer flow patterns and influencing factors with some obvious differences also attributed to their layout, functions and locations.
Customer profiles that include gender and age information are important to businesses and can be used to promote sales and provide personalized services. This information is gathered in e-commerce by analyzing customer visit records in virtual web space. However, such practice is difficult in brick-and-mortar businesses because the data that can be utilized to infer customer profiles are limited in physical spaces. In this paper, we attempt to infer the gender and age of customers using indoor positioning data generated by the Wi-Fi engine. To achieve this, we first construct a synthesized features vector to distinguish different profiles. This vector contains both customer spatial–temporal mobility characteristics and interest preferences. A hidden Markov model group detection method is then applied to detect customers who shop together because they usually show the same shopping behavior and it is difficult to distinguish their profiles. Finally, a random forest inference model is proposed to infer profiles of customers who shop alone. The indoor positioning data collected in the Longhu Tianjie Plaza in Chongqing were used as a case study. The result shows that customer profiles are indeed inferable from indoor positioning data. The accuracy of the gender inference model reaches 73.9%, while that of the age inference model is 67.9%. This demonstrates the potential value of new “big data” for promoting precision marketing and customer management in brick-and-mortar businesses.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)