Search

Pattern images are artificially designed images which are discriminative in aspects of elements, styles, arrangements and so on. Pattern images are widely used in fields like textile, clothing, art, fashion and graphic design. With the growth of image numbers, pattern image retrieval has great potential in commercial applications and industrial production. However, most of existing content-based image retrieval works mainly focus on describing simple attributes with clear conceptual boundaries, which are not suitable for pattern image retrieval. It is difficult to accurately represent and retrieve pattern images which include complex details and multiple elements. Therefore, in this paper, we collect a new pattern image dataset with multiple labels per image for the pattern image retrieval task. To extract discriminative semantic features of multi-label pattern images and construct high-level topology relationships between features, we further propose an Attention Mechanism Driven Graph Convolutional Network (AMD-GCN). Different layers of the multi-semantic attention module activate regions of interest corresponding to multiple labels, respectively. By embedding the learned labels from attention module into the graph convolutional network, which can capture the dependency of labels on the graph manifold, the AMD-GCN builds an end-to-end framework to extract high-level semantic features with label semantics and inner relationships for retrieval. Experiments on the pattern image dataset show that the proposed method highlights the relevant semantic regions of multiple labels, and achieves higher accuracy than state-of-the-art image retrieval methods.

Point Of Interest Auto-Completion (abbr. as POI-AC) is one of the featured functions for the search engine at Baidu Maps. It can dynamically suggest a list of POI candidates within milliseconds as a user enters each character (e.g., English, Chinese, or Pinyin character) into the search box. Ideally, a user may need to provide only one character and immediately obtain the desired POI at the top of the POI list suggested by POI-AC. In this way, the user's keystrokes can be dramatically saved, which significantly reduces the time and effort of typing, especially on mobile devices that have limited space for display and user interfaces. Despite using a user's profile and input prefixes for personalized POI suggestions, however, the state-of-the-art approach, i.e., P^3AC, still has a long way to go so as to generate not only personalized but, more importantly, time- and geography-aware suggestions. In this paper, we find that 17.9% of users tend to look for diverse POIs at different times or locations using the same prefix. This insight drives us to establish an end-to-end spatial-temporal POI-AC (abbr. as ST-PAC) module to replace P3AC at Baidu Maps. To alleviate the problem of the long-tail distribution of time- and location-specific data on POI-AC, we further propose a meta-learned ST-PAC (abbr. as MST-PAC) updated by an efficient MapReduce algorithm. MST-PAC can significantly overcome the "long-tail" issue and rapidly adapt to the cold-start POI-AC tasks with fewer examples. We sample several benchmark datasets from the large-scale search logs at Baidu Maps to assess the offline performance of MST-PAC in line with multiple metrics, including Mean Reciprocal Rank (MRR), Success Rate (SR) and normalized Discounted Cumulative Gain (nDCG). The consistent improvements on these metrics give us more confidence to launch this meta-learned POI-AC module online. As a result, the critical indicator on user satisfaction online, i.e., the average number of keystrokes in a POI-AC session, significantly decreases as well. For now, MST-PAC has already been deployed in production at Baidu Maps, handling billions of POI-AC requests every day. It confirms that MST-PAC is a practical and robust industrial solution for large-scale POI Search.

As a social media popular in Chinese young people, Bilibili's average daily active users in China has reached 51 million. However, Bilibili users often express their negative attitude towards product placement in videos of content creators, which may cause number of followers to drop for content creators and lead expected marketing effect to be unattainable for brand marketers. Therefore, the main purpose of the study is to investigate the reasons why users of Bilibili have negative attitudes towards the product placement in videos by content creators and propose three potential factors—content authenticity, content congruity and product involvement, which can influence Bilibili users’ negative attitude towards product placement. After sending questionnaires to Bilibili users and doing data analysis, the study gets the results that content authenticity, content congruity and product placement are all negatively correlated with Bilibili users’ negative attitude towards product placement, and product involvement is positively correlated with content authenticity and content congruity. This gives practical implications for both Bilibili content creators and brand marketers who want to seek Bilibili content creators for commercial cooperation.

with the diversification of power services and the access of mass terminals, different services have different QoS requirements, large bandwidth, low delay and low cost. In this regard, the power heterogeneous network parallel transmission technology is used to improve the power business transmission capacity, and data diversion is an important problem of parallel transmission. The minimum transmission delay model and cost model of wireless heterogeneous network are established. Combined with the improved nsga-2 algorithm, the multi-objective function model is solved. The results show that the improved nsga-2 algorithm can quickly solve the Pareto optimal solution and the optimal data transmission parameters of each link.

Today, businesses rely on efficiently running analytics on large amounts of operational and historical data to gain business insights and competitive advantage. Increasingly, such analytics are run using cloud-based data analytics services, such as Google BigQuery, Microsoft Azure Synapse, Amazon Redshift, and Snowflake. These services persist and process data in compressed, columnar formats, stored in large blocks, each of which contains thousands or millions of records. For these services, disk I/O from (remote) cloud storage is often one of the dominant costs for query processing. To reduce the amount of I/O, services often maintain per-block metadata, such as zone maps, which are used to skip blocks that are irrelevant to the query, leading to lower query execution times. However, the effectiveness of block skipping via zone maps is dependent on how the records are assigned to blocks. Recent work on instance-optimized data layouts aims to maximize block skipping by specializing the block assignment strategy to a specific dataset and workload. However, these existing approaches only optimize the layout for a single table. In this paper, we propose MTO, an instance-optimized data layout framework that determines the blocking strategy for all tables in a multi-table database in the presence of joins, such as in a star or snowflake schema common in real-world workloads. MTO takes advantage of sideways information passing through joins to jointly optimize the layout for all tables, which results in better block skipping and hence reduced query execution times. Experiments on a commercial cloud-based analytics service show that MTO achieves up to 93% reduction in blocks accessed and 75% reduction in end-to-end query times compared to alternative blocking strategies.

In recent years, the technology of identity authentication based on biometrics is constantly improving and maturing. Due to the unique advantages of long-distance and multi equipment data acquisition, voiceprint recognition has been gradually commercialized in the past fifty years. However, the traditional voiceprint recognition method will reduce the model performance under the condition of large-scale data. In view of the above problem, this paper mainly studies text independent speaker verification system, using Mel Frequency Cepstrum Coefficient (MFCC) as speech feature parameter based on deep neural network (DNN). Compared with the traditional neural network, this method has the fast-learning ability of the network weight and high recognition rate. In this experiment, different speed speech is added to the training, which improves the recognition accuracy, and promotes the robustness of the model.

With the continuous improvement of virtual technology, the current consumer experience has become more and more 'immersive', and a number of new concepts have been derived, such as the immersion media communication environment created by VR and AR technology. By fully mobilizing people's facial feelings, people can focus on the target situation created by the designer. This paper is based on the implementation of human-computer interaction application of "cultural tourism integration", focusing on exploring how to innovate the display mode of folk life along the Grand Canal. Through the concept of immersive experience, the application of human-computer interaction technology to the construction of tourist areas along the grand canal can greatly enrich the tourist experience of the audience, and provide reference for the implementation of the national strategy And make contributions to the important spirit of protecting the cultural heritage of the Grand Canal.

Canal clay pottery is one of the important intangible cultural heritages of the Zaozhuang section of the canal. It germinates due to the canal gene nurturing and develops with the canal culture. The craftsmanship is exquisite, and it has the aesthetic thinking of "creating an image to fulfill its meaning. In the era of the Internet and the Internet of Things, information interaction design is an innovative channel where technology and art complement each other, providing new ideas for the inheritance of canal pottery. This article will specifically discuss the implementation path and method based on this idea

Foodborne Disease (FBD) is the most common disease around the world. Based on the surveillance data of FBD in Zhejiang Province from Jan 1, 2013 to Dec 31, 2015, including 26721 illness cases, 1591 hospitalizations and 6 deaths. We conducted exploratory data analysis to evaluate eidemiological characteristics. The spatial autocorrelation analysis was performed to detect the clusters at county level. The results show that the high peaks usually occurred in summer and autumn, and high-high pattern areas were mainly concentrated in the southwest counties. Most likely clusters were mainly located in the northern counties over the period from May to October, which deserved public's attention. Incidence rates of illness and hospitalization for males were higher than females. People in 18-35 group were suspected to FBD and the elderly over 60 had a highest hospitalized percentage. Peasants and children accounted for about 40% of illness cases, and over 50% of hospitalized cases were peasants (especially in spring and winter). Norovirus was the major cause in spring and winter, while Vibrio parahaemolyticus and Salmonella were the top two pathogens in summer and autumn.

For the recognition of feature positions and feature points in face images, it is often necessary to test and iteratively calculate the extracted face feature images with the help of various data extraction and information optimization algorithms, to obtain the global optimal solution inside the feature space. This article relies on AdaBoost the training algorithm uses several weak classifiers to expand the training of data samples, sets the same initialization weight for each sample, and expands the retrieval and exclusion operations of multi-layer target images in the Cascade Classifier, repeat the weighted summation of the calculated Image weight and the threshold, gradually eliminate the redundant data information in the face feature space, and ensure the optimal solution of the image algorithm test in the face recognition system.

In this paper, consumer overconfidence and the decision behavior of manufacturer are incorporated into the retailer's price decision in the supply chain. Through the analysis of overconfident consumers' purchasing decision with linear utility function, this paper study the retail and wholesale price decision of perishable products and get the profit of manufacturer and retailer in centralized and decentralized supply chain. In addition, the revenue sharing ratio among the manufacturer and retailer, the effect of consumer overconfidence level on the price decisions and profits are analyzed.

As the Internet becomes more popular among Chinese citizens, Ecommerce is developing. In the meanwhile, the traditional retail industry has been greatly affected, so the transformation is urgent. In 2016, Jack Ma proposed a new concept of "New Retail" and then the government encouraged it. After that, retail companies began to transform. With the development of AI, cloud computing and big data, several companies explored new profit model, seeking online and offline development. GOME Retail Holdings Limited is one of the leading transformed retail companies, which is representative. Based on the profit model proposed by Guohong Zhen, this paper takes GOME as the research object, analyzing its profit model under the new retail. By comparing its financial evaluation before and after transformation, the paper analyzes its advantages and disadvantages and makes some suggestions for the company. Other companies in retail industry may gain some implications and find out how to survive and develop well in the trend of new retail.

Web and mobile technologies enable ubiquitous access to information. Yet, it is getting harder, even for subject matter experts, to quickly identify quality, trustworthy, and reliable content available online through search engines powered by advanced knowledge graphs. This paper explores the practical applications of Domain Specific Knowledge Graphs that allow for the extraction of information from trusted published and unpublished sources, to map the extracted information to an ontology defined in collaboration with sector experts, and to enable the public to go from single queries into ongoing conversations meeting their knowledge needs reliably. We focused on Social-Impact Funding, an area of need for over one million nonprofit organizations, foundations, government entities, social entrepreneurs, impact investors, and academic institutions in the US.

Our voices can convey many different types of thoughts and intent; how our voices carry them is often not consciously controlled and as a consequence, unintended effects may arise that negatively impact our relationships. How we say things is as important as what we say. This paper presents methodologies for computing a set of physical properties from sound waves of a speaker's voice directly, referred to as acoustic measures. Experiments are designed and conducted to establish the correlations between physical properties and auditory measures for human perception of sound waves. Based on these correlations, a voice coaching app can guide users, in real-time or deferred retrospective, to modify their speech's auditory measures, such as rate of speech, energy level, and intonation, to achieve their intended communication goals.

Point of interest auto-completion (POI-AC) is a featured function in the search engine of many Web mapping services. This function keeps suggesting a dynamic list of POIs as a user types each character, and it can dramatically save the effort of typing, which is quite useful on mobile devices. Existing approaches on POI-AC for industrial use mainly adopt various learning to rank (LTR) models with handcrafted features and even historically clicked POIs are taken into account for personalization. However, these prior arts tend to reach performance bottlenecks as both heuristic features and search history of users cannot directly model personal input habits. In this paper, we present an end-to-end neural-based framework for POI-AC, which has been recently deployed in the search engine of Baidu Maps, one of the largest Web mapping applications with hundreds of millions monthly active users worldwide. In order to establish connections among users, their personal input habits, and correspondingly interested POIs, the proposed framework (abbr. P3AC) is composed of three components, i.e., a multi-layer Bi-LSTM network to adapt to personalized prefixes, a CNN-based network to model multi-sourced information on POIs, and a triplet ranking loss function to optimize both personalized prefix embeddings and distributed representations of POIs. We first use large-scale real-world search logs of Baidu Maps to assess the performance of P3AC offline measured by multiple metrics, including Mean Reciprocal Rank (MRR), Success Rate (SR), and normalized Discounted Cumulative Gain (nDCG). Extensive experimental results demonstrate that it can achieve substantial improvements. Then we decide to launch it online and observe that some other critical indicators on user satisfaction, such as the average number of keystrokes and the average typing speed at keystrokes in a POI-AC session, which significantly decrease as well. In addition, we have released both the source codes of P3AC and the experimental data to the public for reproducibility tests.

In June 2019, Ministry of Industry and Information Technology awarded 5G licenses to three operators and National Radio and Television Administration. This marks that China has officially entered the 5G era. 5G is not only a technological upgrading of the communication industry, but also can speed up the upgrading of other vertical industries. In this paper, we introduce three key technologies of 5G: network slicing, network function virtualization (NFV) and software defined network (SDN), we also elaborate the application status of 5G network slicing technology in power communication network. Finally, we prospect the development of network slicing technology in power communication network.

With the growth of images on the Internet, plenty of hashing methods are developed to handle the large-scale image retrieval task. Hashing methods map data from high dimension to compact codes, so that they can effectively cope with complicated image features. However, the quantization process of hashing results in unescapable information loss. As a consequence, it is a challenge to measure the similarity between images with generated binary codes. The latest works usually focus on learning deep features and hashing functions simultaneously to preserve the similarity between images, while the similarity metric is fixed. In this paper, we propose a Rank-embedded Hashing (ReHash) algorithm where the ranking list is automatically optimized together with the feedback of the supervised hashing. Specifically, ReHash jointly trains the metric learning and the hashing codes in an end-to-end model. In this way, the similarity between images are enhanced by the ranking process. Meanwhile, the ranking results are an additional supervision for the hashing function learning as well. Extensive experiments show that our ReHash outperforms the state-of-the-art hashing methods for large-scale image retrieval.

In the field of natural language processing, recurrent neural networks are good at capturing long-range dependent information and can effectively complete text classification tasks. However, Recurrent neural network is model the entire sentence in the process of text feature extraction, which easily ignores the deep semantic information of the local phrase of the text. To further enhance the expressiveness of text features, we propose a text classification model base on region embedding and LSTM (RELSTM). RELSTM first divides regions for text and then generates region embedding. We introduce the learnable local context unit(LCU) to calculate the relative position information of the middle word and its influence on the context words in the region, and obtain a region matrix representation. In order to reduce the complexity of the model, the max pooling operation is applied to the region matrix and we obtain a dense region embedding. Then, we use LSTM's long-term memory of text information to extract the global characteristics. The model is verified on public data sets, and the results are compared using 5 benchmark models. Experimental results on three dataset show that RELSTM has better overall performance and is effective in improving the accuracy of text classification compared with traditional deep learning models.

The purpose of this paper is to examine whether market structure or technical efficiency is correlated with the performance of ecommerce in China. The paper estimates technical efficiencies by using input-oriented data envelopment analysis (DEA) model. The fixed effect panel regressions by the estimation of pool least square (PLS) are employed to test the relationship of market structure, technical efficiency and performance. The findings indicate that market structure of e-commerce in China undergoes transition from monopoly to oligopoly. Meanwhile, overall technical inefficiencies are not only caused by poor pure technical efficiency but also unsuccessful operation at decreasing returns to scale. The results also show that market structure and market size have an insignificant relationship with performance; but market share and pure technical efficiency have a significant positive effect on performance; and scale efficiency has a negative effect on the performance. Therefore, modified efficient structure (MES) hypothesis holds in China.

In this paper, the decision maker's overconfidence is taken into consideration for the different members in the supply chain. Through the analysis of overconfident consumer's decision making process, this paper construct the profit functions for the fully rational retailer and overconfident retailer, conduct the optimal analysis to get the optimal ordering decisions, and conduct comparative study to analyze the effect of retailer's overconfidence level and consumer's overconfidence level on the optimal decisions' deviation.

Digital coupon has become an uprising power in internet marketing. As an integration of online and offline resources, it is widely applied in takeaway market. The key to realize the marketing value of digital takeaway coupon is to spur consumers' intention to use it. Literature at home and abroad has investigated the factors influencing consumers' intention to use mobile coupons, but there is little research on digital takeaway coupons. This study draws on previous study results, integrates Technology Acceptance Model with consumer characteristics to investigate university students' attitude toward the coupons and the determinants influencing their intention to use them. The results of a questionnaire-based research show that perceived economic benefit, attitude, perceived enjoyment and coupon proneness have significant influence on intention to use digital takeaway coupons among university students, with perceived economic benefit having the greatest impact while coupon proneness the least.

The value assessment of private properties is an attractive but challenging task which is widely concerned by a majority of people around the world. A prolonged topic among us is "how much is my house worth?". To answer this question, most experienced agencies would like to price a property given the factors of its attributes as well as the demographics and the public facilities around it. However, no one knows the exact prices of these factors, especially the values of public facilities which may help assess private properties. In this paper, we introduce our newly launched project "Monopoly" (named after a classic board game) in which we propose a distributed approach for revaluing private properties by learning to price public facilities (such as hospitals, schools, and metros) with the large-scale urban data we have accumulated via Baidu Maps. To be specific, our method organizes many points of interest (POIs) into an undirected weighted graph and formulates multiple factors including the virtual prices of surrounding public facilities as adaptive variables to parallelly estimate the housing prices we know. Then the prices of both public facilities and private properties can be iteratively updated according to the loss of prediction until convergence. We have conducted extensive experiments with the large-scale urban data of several metropolises in China. Results show that our approach outperforms several mainstream methods with significant margins. Further insights from more in-depth discussions demonstrate that the "Monopoly" is an innovative application in the interdisciplinary field of business intelligence and urban computing, and it will be beneficial to tens of millions of our users for investments and to the governments for urban planning as well as taxation.

In this paper we approach to accelerate the data processing of storage and computing separated big data systems. We propose an improved Network Interface Card with Query Filter (NIC-QF), implemented by FPGA on storage nodes, to accelerate the data queries, which can also reduce the workload and communication overhead on computing nodes. NIC-QF is designed with query filtering accelerator and Network Interface Card (NIC) communicator, which can filter the original data on storage nodes as an implicit coprocessor and directly send the filtered data to computing nodes of big data systems. Filter units in NIC-QF can perform multiple SQL tasks in parallel, and each filter unit is internally pipelined, which can further speed up the data processing. Experiments with two benchmarks demonstrate the efficiency of our approach, which can achieve average up to 65.56% faster than the traditional approach.

360-degree cinematic virtual reality(CVR) could bring users a full perspective experience, but high freedom will cause the loss of important information. To bring the user back to the narrative thread, researchers have made a variety of guided attempts, such as using note reminders, voice guidance, and so on. However, these methods are less efficient and less controllable for CVR for the purpose of science education. To solve this issue, the authors proposed visual guidance technology by synchronizing camera animation and 360-degree video in VR. When the user does not operate the mouse, the CVR viewing angle is controlled by the camera animation synchronously. This technology effectively transmits important information while allowing users to interact with the 360 video freely, thus improving the cognitive effect in science education.

Tags of a Point of Interest (POI) can facilitate location-based services from many aspects like location search and place recommendation. However, many POI tags are often incomplete or imprecise, which may lead to performance degradation of tag-dependent applications. In this paper, we study the POI tag refinement problem which aims to automatically fill in the missing tags as well as correct noisy tags for POIs. We propose a tri-adaptive collaborative learning framework to search for an optimal POI-tag score matrix. The framework integrates three components to collaboratively (i) model the similarity matching between POI and tag, (ii) recover the POI-tag pattern via matrix factorization and (iii) learn to infer the most possible tags by maximum likelihood estimation. We devise an adaptively joint training process to optimize the model and regularize each component simultaneously. And the final refinement results are the consensus of multiple views from different components. We also discuss how to utilize various data sources to construct features for tag refinement, including user profile data, query data on Baidu Maps and basic properties of POIs. Finally, we conduct extensive experiments to demonstrate the effectiveness of our framework. And we further present a case study of the deployment of our framework on Baidu Maps.

Under the trend of fully advocating "consumer demand", this study intends to understand the current situation of consumers' perceptions and satisfactions of the multi-channels constructed by insurance companies, and then to understand the experience and cognition of consumers using channels. Seeking advice from the customer side, thereby reducing the gap between buyers and sellers for service quality and satisfaction. Therefore, this study first uses the independence test to test whether the demographic variables are related to the consumer's choice of pathway type, and then use the independence test to test the measure in each satisfaction facet. Whether there is a relationship between satisfactions. The results of the study show that the channel utilization rate of direct sales units is as high as 75%, which is still the mainstream of sales channels; the consumer groups are mainly concentrated in "married", "36-45 years old" and "46-55 years old". "The age of 55 or above", "low education level", "the number of 1-2 sheets of insurance policy" is the highest; in addition, the utilization rate of insurance generation channels is 17%, ranking second, and the consumer group is characterized by: The number of young people, high education level, and policy sheets is more than 4, while the bank access rate ranks third. In addition, consumers' perceptions and usage rates of emerging pathways are very low, and they do not meet the expected results of the industry. Consumers' consideration of the choice of channels is "completed after-sales service" and "complete planning for different needs". "Professional knowledge provision" is the mainstay, and up to 60% of the people are dissatisfied with the level of service provided by the direct sales unit clerk, but in the insurance generation, more than half of the people agree Satisfaction with the quality of the services provided by this pathway.

According to the characteristics of mechanics course in university, the author designs and practices the participatory teaching based on BOPPPS mode in teaching, This paper describes how the author form the teaching design based on BOPPPS mode with teaching examples, and focuses on the application of "4F guiding strategy" in guiding students to participate in teaching. The practice shows that the degree of student's participation in class increased, and the teaching results show that the use of the BOPPPS mode increases the students' interest in leaning and the ability of the problem-solving. Based on the questionnaire over 52% of students clearly believed that BOPPPS mode is effective for their learning, and 57% of the students like the form of group discussion based on 4F guiding strategy. This research may help university faculty promote their teaching skills

In the field of oil and gas exploration, reservoir parameter prediction is often affected by multi-solution of the seismic attribute combination, which leads to low prediction accuracy. In this paper, a feature selection method based on Gamma test is proposed to optimize the attribute combination and then combine it with deep neural network to accomplish reservoir parameter prediction. By computing the value of the statistics, it not only provides the best combination of the corresponding attributes to predict the target but also provides the proper training mean square error of neural network and the proper size of the training set. With this guide, over-fitting can be effectively avoided and the prediction accuracy is improved. The selected seismic attributes combination is used as the optimized network input, then use extreme learning machine to accomplish the regression problem. Through the analysis of the real seismic data experimental results, it is proved that the Gamma test is an effective nonparametric tool for feature selection.