Search

With the transformation of health concepts and medical models, traditional medicine is receiving more and more attention from people all over the world. Traditional medicine has become an important part of health promotion and global health management improvement. This paper briefly describes the connotation and development trend of traditional medicine in the new era from the perspective of philosophical theory. Through a thorough analysis of the rational thinking and realistic basis for the integration and development of traditional medicine in the new era from the perspectives of time, space, content and technology, we come to the conclusion that, under the modern medical model supported by Big Health and Big Data, the deepening and improvement of the consolidation of multi-dimensional integration and multi-collaborative system of traditional medicine will play a more fundamental role in mankind’s health and hygiene. This paper also presents the paths of integrative development for the traditional medicine under the new circumstances, aiming to provide theoretical ideas for further promoting the systematic development of traditional medicine and helping the revival of traditional Chinese medical civilization.

Patient similarity analytics has emerged as an essential tool to identify cohorts of patients who have similar clinical characteristics to some specific patient of interest. In this study, we propose a patient similarity measure called D3K that incorporates domain knowledge and data-driven insights. Using the electronic health records (EHRs) of 169,434 patients with either diabetes, hypertension or dyslipidaemia (DHL), we construct patient feature vectors containing demographics, vital signs, laboratory test results, and prescribed medications. We discretize the variables of interest into various bins based on domain knowledge and make the patient similarity computation to be aligned with clinical guidelines. Key findings from this study are: (1) D3K outperforms baseline approaches in all seven sub-cohorts, (2) our domain knowledge-based binning strategy outperformed the traditional percentile-based binning in all seven sub-cohorts, (3) there is substantial agreement between D3K and physicians (κ = 0.746), indicating that D3K can be applied to facilitate shared decision making. This is the first study to use patient similarity analytics on a cardiometabolic syndrome-related dataset sourced from medical institutions in Singapore. We consider patient similarity among patient cohorts with the same medical conditions to develop localized models for personalized decision support to improve the outcomes of a target patient.

The grid power distribution network (PDN) is an essential component in electronic packages to provide the dc power to circuits. In this paper, an integral equation hybrid method is proposed for the efficient calculation of the impedance of the complex grid PDN. First, the original 3-D grid PDN is reduced to a simple 2-D plane PDN. Both of them have the same input and mutual parasitic impedances. After that, this equivalent plane PDN is solved by using the 2-D integral equation method to consider its arbitrary shape. By comparing between the obtained impedances from the proposed method, full-wave method, and measurement, the accuracy and efficiency of the proposed method is verified. The proposed method provides a simple and useful electronics design automation tool for the power integrity analysis of complex circuits.

Patients with Parkinson's disease suffer from gait disturbances, such as a shuffling and festinating gait, which reduces their quality of life. To circumvent this problem, external visual cues may be applied in gait training to maintain the integrity of motor function. However, conventional training methods, such as transverse lines stuck on the ground, are difficult to adjust and adapt to personalized gait ability. This study proposes a convenient instrumented wheel walker that provides gap adjustable visual cues and selectable projection modes onto the ground with and without motion relative to the user. Ten subjects with Parkinson's disease were recruited, and the efficacy of the proposed device for their gait training was assessed. We demonstrated the applicability of our device to address personalized demands in gait guidance. With a personalized setting for patients with Parkinson's disease, a significantly lengthened stride length may be achieved.

X-ray testing is one of the main methods of industrial non-destructive testing, and its testing results have been used as an important basis for defect analysis and quality evaluation. Manual recognition of defects is the traditional method of X-ray testing in the industry. Observers need to keep an eye on the monitor and artificially screen for defects based on experience and industry standards. In the case of working for a long time, the observer’s efficiency will be reduced, and there will be misjudgment or omission, so that the components that do not meet the quality grade will enter the market, resulting in unpredictable losses. Because different observers have different judgments on the test results, the evaluation results are difficult to be unified, which is disadvantageous to the defect detection and identification work. In order to reduce manual intervention flaw detection, this work uses the method based on computer deep learning to intelligently detect flaws in X-ray images. The X-ray data set comes from the American Society for Testing Material (ASTM) standard radiogram, Panyu Chu Kong Steel Pipe (Zhuhai) Co., Ltd. and Gree Electric Appliances, Inc. of Zhuhai, which are divided into the training set and verification set according to the proportion of 70% and 30%. In this work, the models such as FCN, YOLO and Mask R-CNN are used to detect inclusions or bubbles in X-ray images, and it is found that Mask R-CNN model has the highest detection precision and recognition rate, and this model has certain generalization ability.

A number of theories have been proposed for reducing sampling rate of light field. But these theories still need a great many of samples (images) to obtain sufficient geometric information. In this paper, we utilize the sparse representation of light field in Discrete Cosine Transform domain to present a Discrete Cosine Sparse Basis (DCSB). Thus, we can find out the zeros of DCSB to reduce sampling requirement of light field for alias-free rendering. Finally, experimental results demonstrate the effectiveness of our approach without lose information.

Credit risk assessment is an important process for every financial institution. Most current studies in this area focus on the development of the classifiers. However, besides the classifiers, data analysis and mining also play an important role in improving the accuracy. In this paper, an integrated feature selection method is proposed combing the analysis of the data in one feature but belongs to different labels and the analysis of the relationship between a single feature and labels. Back propagation neural network is used as the classification model. Experiments are done on Australian credit approval dataset. The results show that the proposed integrated feature selection can improve the accuracy of assessment compared with other previous work.

An intelligent power automatic switching device (ATS) is designed for communication vehicle, which can automatically detect the quality of the power supply, automatically switch the power supply according to the priority set by the system, automatically start and stop the diesel generators installed on the communication vehicle, and communicate with the PC of the communication vehicle. The device includes the corresponding power circuit, interface circuit, communication circuit, based on s7-1200 controller, and it finally realizes the system functions.

With the rise of mobile video, user-generated content, and social networks, there is a massive opportunity for disruptive innovations in the media and content industry. It is now a fast-changing landscape with rapid advances in AI-powered content creation, dissemination and interaction technologies. I believe the current trends are leading us towards a world where everyone is equally empowered to produce high-quality content in video, music, augmented reality or more – and to share their information, knowledge, and stories with a large global audience. This new AI- powered content platform can further lead to innovations in advertising, e-commerce, online education, and productivity. I will share the current research efforts at ByteDance connected to this emerging new platform through products such as Douyin and TikTok, and discuss the challenges and the direction of our future research.

This paper introduces the Attribute-Decomposed GAN, a novel generative model for controllable person image synthesis, which can produce realistic person images with desired human attributes (e.g., pose, head, upper clothes and pants) provided in various source inputs. The core idea of the proposed model is to embed human attributes into the latent space as independent codes and thus achieve flexible and continuous control of attributes via mixing and interpolation operations in explicit style representations. Specifically, a new architecture consisting of two encoding pathways with style block connections is proposed to decompose the original hard mapping into multiple more accessible subtasks. In source pathway, we further extract component layouts with an off-the-shelf human parser and feed them into a shared global texture encoder for decomposed latent codes. This strategy allows for the synthesis of more realistic output images and automatic separation of un-annotated attributes. Experimental results demonstrate the proposed method's superiority over the state of the art in pose transfer and its effectiveness in the brand-new task of component attribute transfer., Accepted by CVPR 2020 (Oral). Project Page: https://menyifang.github.io/projects/ADGAN/ADGAN.html

This paper investigates the foot types and structural features of Chinese four-syllable abbreviations. Results show that: first, whether for the need of complexity and richness of ideogram or for avoiding ambiguity, these abbreviated forms are an essential and meaningful part of modern Chinese abbreviations; second, most of the existing forms tend to present 2 + 2 balanced prosodic structure, nominal attribute in part of speech. In addition, there also exists the possibility of abbreviation further to less than four syllables under some specific conditions.

As a kind of geophysical method, the principle of transient electromagnetic method (TEM) observes the secondary vortex field in the gap of the primary field, which is applied to the investigation of resources and environmental engineering survey. In order to realize the rapid and precise detection of the deep targets (due to its relative great detection depth and high accuracy), the short offset time-domain electromagnetic method (SOTEM) is proposed in this paper. In addition, its feasibility, observation and interpretation techniques are researched. The results show that SOTEM has the same near source detection capability as the central loop configuration when choosing the step waveform, and as a result, the primary and secondary fields can be separated theoretically. The application has shown the effectiveness and superiority of SOTEM in the deep target exploration.

Large-scale image datasets and deep convolutional neural networks (DCNNs) are the two primary driving forces for the rapid progress in generic object recognition tasks in recent years. While lots of network architectures have been continuously designed to pursue lower error rates, few efforts are devoted to enlarging existing datasets due to high labeling costs and unfair comparison issues. In this article, we aim to achieve lower error rates by augmenting existing datasets in an automatic manner. Our method leverages both the web and DCNN, where the web provides massive images with rich contextual information, and DCNN replaces humans to automatically label images under the guidance of web contextual information. Experiments show that our method can automatically scale up existing datasets significantly from billions of web pages with high accuracy. The performance on object recognition tasks and transfer learning tasks have been significantly improved by using the automatically augmented datasets, which demonstrates that more supervisory information has been automatically gathered from the web. Both the dataset and models trained on the dataset have been made publicly available.

Critical transitions whereby small changes in conditions can cause large and irreversible changes in ecosystem states are a cause of increasing concern in ecology. Here, we focus on the irreversibility of these transitions, formally known as hysteresis. We explore how simple correlations between parameters in Lotka-Volterra predator-prey equations result in a variety of complicated hysteretic patterns. These patterns include “unattainable” stable states that once lost may never be recovered. We suspect these patterns to be common in natural systems, where interactions between diverse assemblages are unavoidable. Thus, understanding underlying hysteretic structures may be necessary for rescuing lost ecosystem states and avoiding future losses.

A magnetic field sensing system with a single primary sensor and multiple reference sensors deployed locally and orthogonally, was proposed for downlink signal reception and interference cancelling for Through-the-Earth Communication (TEC). This paper mathematically analyzes a design optimization process for a search coil magnetometer (SCM), and applies that process to minimize the bandwidth of the primary SCM for TEC signal reception and the volume of reference SCMs for multiple distributions. The primary SCM achieves a 3-㏈ bandwidth of 7 ㎐, a sensitivity threshold of 120 fT/√㎐, and a volume of 2.32 × 10 −4 ㎥. The entire sensing system volume is as small as 10 −2 ㎥. Experiments with interference from industrial frequency harmonics demonstrated an average of 36 ㏈ and 18 ㏈ improvements in signal-to-interference ratio and signal-to-interference plus noise ratio, respectively, using multichannel recursive-least-squares algorithm. Thus, the proposed sensing system can reduce the interference effectively and allows reliable downlink signal reception.

The aim of this study was to analyze the error variation in the applicator placement during the first and second radiotherapy session for cervical cancer. We recruited 22 patients with cervical cancer treated with radiotherapy. According to the image output in the first and second CT-Sim inspection, we conducted comparative analysis of image fusion to accurately measure the errors in applicator position in the horizontal (X-), longitudinal (Y-) and vertical (Z)-axes. The calibration processing was implemented in accordance with the data error measured and the location parameters, such as the angle and depth of the applicator. Electronic portal imaging technology (EPID) was used to calibrate posture change amplitude for the extracorporeal irradiation of patients, and dynamic measurement with applicator position was used to describe the error of the parameters. Finally, the data from two measurements in CT-Sim, digital reconstruction radiography (DRR) and EPID were compared. After calibration, the mean value of error of the applicator were significantly smaller. Image registration planning for error parameter calibration of applicator position can effectively reduce the applied horizontal spatial position error in radiotherapy treatment, and improve the accuracy and effectiveness during treatment.

We propose the Unified Visual-Semantic Embeddings (Unified VSE) for learning a joint space of visual representation and textual semantics. The model unifies the embeddings of concepts at different levels: objects, attributes, relations, and full scenes. We view the sentential semantics as a combination of different semantic components such as objects and relations; their embeddings are aligned with different image regions. A contrastive learning approach is proposed for the effective learning of this fine-grained alignment from only image-caption pairs. We also present a simple yet effective approach that enforces the coverage of caption embeddings on the semantic components that appear in the sentence. We demonstrate that the Unified VSE outperforms baselines on cross-modal retrieval tasks; the enforcement of the semantic coverage improves the model's robustness in defending text-domain adversarial attacks. Moreover, our model empowers the use of visual cues to accurately resolve word dependencies in novel sentences.

Load aggregators can help residents aggregate their loads to better participate in demand-side response in electricity market. For residents, the characteristics of flexible loads can be classified as adjustable and non-adjustable characteristics, transferable and non-transferable characteristics, interruptible and non-interruptible characteristics. Based on these characteristics, an optimal dispatching strategy of aggregated loads is proposed in this paper. For load aggregators, with the object of peak load shifting, an optimal dispatching strategy is developed. Then, to estimate the revenue of load aggregators, a settlement model is established. Case study results verified the validity of the proposed dispatching strategy, which can not only help distribution network reduce the peak and valley difference, but bring these load aggregators and residents extra benefits from the distribution network.

Flow of cut tobacco is an important processing parameter of high speed cigarette machine. In order to meet the demand of cut tobacco flow statistics in the whole sealing pipes, this paper shares a machine vision based on-line statistical system of cut tobacco flow. Besides the introduction of the system’s hardware, this paper focuses on the image processing algorithms such as cut tobacco image preprocessing, thresholding and morphological processing. Firstly, after the preprocessing of image denoising, the cut tobacco color image is transformed into gray-scale image by weighted average method. Secondly, the gray-scale image is converted into to binary image by the maximum entropy threshold method. Finally, the binary cut tobacco image is processed by dilation and erosion of the morphological image processing, which lay the foundation for the on-line statistic process of cut tobacco flow. The practical engineering application of the system shows that the on-line statistic of cut tobacco flow is real-time with high reliability, which can meet the requirements of the production process of tobacco companies and greatly improve the automation and intelligentization of high-speed cigarette machine, resulting in certain generalization value in the industry.

In the cigarette production industry, the automatic cigarette sampling device varies. This paper introduces a new batch sampling device based on PLC technology, which involves mainly the system's mechanical structure, its operating principle, the PLC program control flow and so on. Sampling by the device is completely mechanical, which means it's easy to operate and is with good stability, safety and reliability. Meanwhile, it does not interfere with or damage the sample flow. So it can be used for non-destructive batch sampling of rod-like products such as large flow of cigarettes in high-speed production environment, which greatly reduces the labour intensity of quality checkers by improving the automation of cigarette testing, resulting in great industrial application value.

Taking Hangzhou cigarette factory as an example, this paper establishes a key performance index system through integrating the balanced scorecard (BSC) and key performance index (KPI). The index system includes the first-level and 13 second-level indicators in four dimensions of finance, customers, internal processes, as well as learning and growth. Through Analytic Hierarchy Process (AHP) and fuzzy comprehensive evaluation method, it constructs an evaluation model of key corporate performance, finally achieves the actual performance levels of the enterprise, as well as puts forward the improvement suggestions. This evaluation method mainly includes the steps of determining the domain of evaluation factors at all levels, determining the evaluation level, determining fuzzy weight vectors of various indexes, establishing the fuzzy relation matrix and calculating comprehensive evaluation results. It has high accuracy in the process of performance evaluation, which has certain reference and guiding significance for improving the comprehensive competitiveness.

The air quality in urban areas is a major concern in modern cities due to significant impacts of air pollution on public health, global environment, and worldwide economy. Recent studies reveal the importance of micro-level pollution information, including human personal exposure and acute exposure to air pollutants. A real-time system with high spatio-temporal resolution is essential because of the limited data availability and non-scalability of conventional air pollution monitoring systems. Currently, researchers focus on the concept of The Next Generation Air Pollution Monitoring System (TNGAPMS) and have achieved significant breakthroughs by utilizing the advance sensing technologies, MicroElectroMechanical Systems (MEMS) and Wireless Sensor Network (WSN). However, there exist potential problems of these newly proposed systems, namely the lack of 3D data acquisition ability and the flexibility of the sensor network. In this paper, we classify the existing works into three categories as Static Sensor Network (SSN), Community Sensor Network (CSN) and Vehicle Sensor Network (VSN) based on the carriers of the sensors. Comprehensive reviews and comparisons among these three types of sensor networks were also performed. Last but not least, we discuss the limitations of the existing works and conclude the objectives that we want to achieve in future systems.

This paper presents an efficient procedure to locate joint centers on 3D scanned human models. Using the hip joint centers (HJCs) as an example, we conducted an experiment to study the validity of the procedure. We carried out 3D body scanning on six subjects, which were attached with the necessary body landmarks. Another 12 participants were recruited to perform two tasks. The first task was to identify the locations of the body landmarks. We pre-programmed the estimation method to an in-house rigging system. The lower body bones were then placed by the system according to the estimated HJCs. The second task was to place the lower body bones based on visual judgments. The results show that the HJCs obtained in the first task are more accurate. The results are also visually demonstrated. The promising results of the experiment indicate the possibility of building skeletal structures with 3D body scanning.

Multiple Traveling Salesman Problem (MTSP) is able to model and solve various real-life applications such as multiple scheduling, multiple vehicle routing and multiple path planning problems, etc. While Traveling Salesman Problem (TSP) focuses on searching a path of minimum traveling distance to visit all cities exactly once by one salesman, the objective of the MTSP is to find m paths for m salesmen with a minimized total cost - the sum of traveling distances of all salesmen through all of the respective cities covered. They have to start from a designated depot which is the departing and returning location of all salesmen. Since the MTSP is a NP-hard problem, a new effective Genetic Algorithm with Local operators (GAL) is proposed in this paper to solve the MTSP and generate high quality solution within a reasonable amount of time for real-life applications. Two new local operators, Branch and Bound (BaB) and Cross Elimination (CE), are designed to speed up the convergence of the search process and improve the solution quality. Results demonstrate that GAL finds a better set of paths with a 9.62% saving on average in cost comparing to two existing MTSP algorithms.

Market research shows that one of the most intolerable issues in the pack of cigarettes is the cigarette missing. This issue makes substantial adverse effects on a company which needs to be avoided completely. Existing research uses a weight detection method to identity packages with issues. However, the accuracy of weight detection methods is low due to instrument error and complex workshop environment. Nowadays, deep learning is a hot topic in the area of detection and classification. The input data for this work is the X-ray image which provides an opportunity to detect cigarette missing based on image classification. This work aims to propose a new detection method based on convolutional neural network. The convolutional neural network, which is the most popular deep learning method in the field of images, performs well when hand-crafted features are difficult to obtain. Besides the cigarette missing detection, this paper also succeeds to detect the packet damage, packet missing and cigarette misplaced. Experiments show that the convolutional neural network outperforms all other classical classifiers and the accuracy of cigarette missing detection is more than 99%. The proposed method is a promising cigarette missing detection method and can be used for the detection of packet damage, packet missing and cigarette misplaced.

Since 1997, internet-based e-government has been promoted by the Research and Development Review Board of Executive Yuan, R.O.C., and the application of electronic forms is one of the e-government processes. Electronic form system includes integrated document management and files management systems and is integrated with an electronic attendance system as well. In this way, other than merely achieving creating a paperless system, it also reduces the manpower required for paper-form transmission and shortens document delivery time.

In order to help individuals effectively manage and record these health check physiological measurement data, this research developed an “Integrated Health Check Report Analysis and Tracking Platform” together with H&B Health Centers. Using this platform, the public can query health check data and analysis charts from health check centers through a website. The information will include suggestions from doctors, nutritionists and representatives of various health check categories. The meanings behind various data can be explained to the general public using illustrations.

An incisive understanding of user personality is not only essential to many scientific disciplines, but also has a profound business impact on practical applications such as digital marketing, personalized recommendation, mental diagnosis, and human resources management. Previous studies have demonstrated that language usage in social media is effective in personality prediction. However, except for single language features, a less researched direction is how to leverage the heterogeneous information on social media to have a better understanding of user personality. In this paper, we propose a Heterogeneous Information Ensemble framework, called HIE, to predict users' personality traits by integrating heterogeneous information including self-language usage, avatar, emoticon, and responsive patterns. In our framework, to improve the performance of personality prediction, we have designed different strategies extracting semantic representations to fully leverage heterogeneous information on social media. We evaluate our methods with extensive experiments based on a real-world data covering both personality survey results and social media usage from thousands of volunteers. The results reveal that our approaches significantly outperform several widely adopted state-of-the-art baseline methods. To figure out the utility of HIE in a real-world interactive setting, we also present DiPsy, a personalized chatbot to predict user personality through heterogeneous information in digital traces and conversation logs.

There is a clear lack of respect about traditional network adaptive learning system teaching on individualized assessment knowledge and building tacit knowledge, and this paper presents a knowledge model that supports personalized knowledge assessment, knowledge stored in the cloud computing environment, and construct tacit knowledge for learning body, to provide a personalized learning services for learners to achieve user to adapt the system to adapt to the user's system and two-way adaptation, this paper has guiding significance for further studies of adaptive learning systems.

Knowledge based question answering (KBQA) has attracted much attention from both academia and industry in the field of Artificial Intelligence. However, many existing knowledge bases (KBs) are built by static triples. It is hard to answer user questions with different conditions, which will lead to significant answer variances in questions with similar intent. In this work, we propose to extract conditional knowledge base (CKB) from user question-answer pairs for answering user questions with different conditions through dialogue. Given a subject, we first learn user question patterns and conditions. Then we propose an embedding based co-clustering algorithm to simultaneously group the patterns and conditions by leveraging the answers as supervisor information. After that, we extract the answers to questions conditioned on both question pattern clusters and condition clusters as a CKB. As a result, when users ask a question without clearly specifying the conditions, we use dialogues in natural language to chat with users for question specification and answer retrieval. Experiments on real question answering (QA) data show that the dialogue model using automatically extracted CKB can more accurately answer user questions and significantly improve user satisfaction for questions with missing conditions.

It combines the augmented reality technology and mobile learning devices, and integrates the real scene and virtual augmented information in the mobile learning device. This paper discusses the principle of Augmented Reality and the mobile learning based on Augmented Reality based on the cloud computing. It will have some reference value for the future development.

Mistakes often occurs in map matching when road condition is complex, especially there is island ring, or exit and entrance of highway, in vehicle navigation system. Wrong map matching will lead to wrong map display and wrong route guidance, thus system gives a bad impact to driver. In order to solve this problem, this paper includes an improved map matching algorithm based on NDS, which takes curvature and slope into account. This Algorithm coding has been realized by c and proved its availability at last.

In vehicle navigation system, path matching often go wrong when road condition is complex, especially there are island ring, or exit and entrance of highway. This lead to wrong map display and wrong route guidance, thus system give a bad impact to driver. In order to solve this problem, this paper includes a map matching algorithm based on curvature and slope. This Algorithm coding has been realized by c and proved its availability at last.

There are various kinds of valuable semantic information about real-world entities embedded in webpages and databases. Extracting and integrating these entity information from the Web is of great significance. Comparing to traditional information extraction problems, web entity extraction needs to solve several new challenges to fully take advantage of the unique characteristic of the Web. In this paper, we introduce our recent work on statistical extraction of structured entities, named entities, entity facts and relations from Web. We also briefly introduce iKnoweb, an interactive knowledge mining framework for entity information integration. We will use two novel web applications, Microsoft Academic Search (aka Libra) and EntityCube, as working examples.

Easy photo-taking and photo-sharing today make image an increasingly important type of media in people's everyday life, which arouses a growing demand for a practical image understanding technique. Traditional computer vision or machine learning methods which learn models based on a set of training data are still in the stage of tackling hundreds of object categories. Such a scale is far from practical usage. In recent years, the technique of search-based image annotation on a large-scale data set has demonstrated great success. Rather than directly mapping visual features to texts which is inevitably hindered by the semantic gap, it understands the content of an image by propagating labels of its similar images in a large-scale data set. Since similarity search is performed among homogenous data, the difficulty is greatly reduced. This paper summarizes the extensive work on web image annotation using the large-scale metadata and social information available on the Web, and introduces the Arista system, which is a nonparametric image annotation platform built upon two billion web images. We propose a highly efficient and scalable duplicate-search technique so that the Arista system can be deployed on a few servers. A few interesting applications such as building large-scale celebrity face database and text-to-image translation are also presented in this paper.

Chinese medicinal materials (CMMs) are easily to be contaminated by all kinds of molds to produce various mycotoxins due to their internal factors and the external environmental conditions during the growth, harvesting, processing, and especially storage processes. This will not only affect the quality of CMMs, resulting in enormous financial loss, but also influence the safety and effectiveness of CMMs, posing potential threats to human health. With the increase in awareness of "traditional Chinese medicine health" idea, more and more attention has been paid on how to prevent and control these CMMs from being mouldy to guarantee their safety. Some physical and chemical techniques have been restricted for protecting CMMs due to their own disadvantages. As a green, safe and economic strategy for the preservation of CMMs, "couplet medicine" technique based on the principle of "protecting CMM with another CMM" has been developed: two kinds of CMMs are stored together and fight against each other to prevent mildew metamorphism, exhibiting no obvious changes in color, smell and quality. Nowadays, certain application results have been obtained for the "antagonistic storage" method based on the above mode and principle. In this paper, we would review and discuss the mechanism, practical application and the problems of "couplet medicine" technique, and provide scientific evidences for developing safe and effective tools to protect CMMs from being mouldy.

Recently a few systems for automatically solving math word problems have reported promising results. However, the datasets used for evaluation have limitations in both scale and diversity. In this paper, we build a large-scale dataset which is more than 9 times the size of previous ones, and contains many more problem types. Problems in the dataset are semi-automatically obtained from community question-answering (CQA) web pages. A ranking SVM model is trained to automatically extract problem answers from the answer text provided by CQA users, which significantly reduces human annotation cost. Experiments conducted on the new dataset lead to interesting and surprising results.

Labelled image datasets are the backbone for high-level image understanding tasks with wide application scenarios, and continuously drive and evaluate the progress of feature designing and supervised learning models. Recently, the million scale labelled image dataset further contributes to the rebirth of deep convolutional neural network and bypass manual designing handcraft features. However, the construction process of image dataset is mainly manual-based and quite labor intensive, which often take years' efforts to construct a million scale dataset with high quality. In this paper, we propose a deep learning based method to construct large scale image dataset in an automatic way. Specifically, word representation and image representation are learned in a deep neural network from large amount of click-through logs, and further used to define word-word similarity and image-word similarity. These two similarities are used to automatize the two labor intensive steps in manual-based image dataset construction: query formation and noisy image removal. With a new proposed cross convolutional filter regularizer, we can construct a million scale image dataset in one week. Finally, two image datasets are constructed to verify the effectiveness of the method. In addition to scale, the automatically constructed dataset has comparable accuracy, diversity and cross-dataset generalization with manually labelled image datasets.

Networks in real world usually have multiple layers. Most of the current works concentrate on the single-layer network. Visualization of double-layer networks using 2.5D method is proposed in this paper to reveal the network structure and interdependent relations within network layers. A context-preserving layout method and matrix transformation is adopted to preserve node positions and reduce visual clutters within network layers. Visual analysis is conducted based on the 2.5D visualization, focusing on the interdependent relations and node importance evaluation. The proposed method ensures satisfying readability of the network structural features and demonstrates the relations in the network.

The increasing availability of location-acquisition technologies (GPS, GSM networks, etc.) enables people to log the location histories with spatio-temporal data. Such real-world location histories imply, to some extent, users' interests in places, and bring us opportunities to understand the correlation between users and locations. In this article, we move towards this direction and report on a personalized friend and location recommender for the geographical information systems (GIS) on the Web. First, in this recommender system, a particular individual's visits to a geospatial region in the real world are used as their implicit ratings on that region. Second, we measure the similarity between users in terms of their location histories and recommend to each user a group of potential friends in a GIS community. Third, we estimate an individual's interests in a set of unvisited regions by involving his/her location history and those of other users. Some unvisited locations that might match their tastes can be recommended to the individual. A framework, referred to as a hierarchical-graph-based similarity measurement (HGSM), is proposed to uniformly model each individual's location history, and effectively measure the similarity among users. In this framework, we take into account three factors: 1) the sequence property of people's outdoor movements, 2) the visited popularity of a geospatial region, and 3) the hierarchical property of geographic spaces. Further, we incorporated a content-based method into a user-based collaborative filtering algorithm, which uses HGSM as the user similarity measure, to estimate the rating of a user on an item. We evaluated this recommender system based on the GPS data collected by 75 subjects over a period of 1 year in the real world. As a result, HGSM outperforms related similarity measures, namely similarity-by-count, cosine similarity, and Pearson similarity measures. Moreover, beyond the item-based CF method and random recommendations, our system provides users with more attractive locations and better user experiences of recommendation.

Map search engines, such as Google Maps, Yahoo! Maps, and Microsoft Live Maps, allow users to explicitly specify a target geographic location, either in keywords or on the map, and to search businesses, people, and other information of that location. In this article, we report a first study on a million-entry map search log. We identify three key attributes of a map search record—the keyword query, the target location and the user location, and examine the characteristics of these three dimensions separately as well as the associations between them. Comparing our results with those previously reported on logs of general search engines and mobile search engines, including those for geographic queries, we discover the following unique features of map search: (1) People use longer queries and modify queries more frequently in a session than in general search and mobile search; People view fewer result pages per query than in general search; (2) The popular query topics in map search are different from those in general search and mobile search; (3) The target locations in a session change within 50 kilometers for almost 80% of the sessions; (4) Queries, search target locations and user locations (both at the city level) all follow the power law distribution; (5) One third of queries are issued for target locations within 50 kilometers from the user locations; (6) The distribution of a query over target locations appears to follow the geographic location of the queried entity.

User mobility has given rise to a variety of Web applications, in which the global positioning system (GPS) plays many important roles in bridging between these applications and end users. As a kind of human behavior, transportation modes, such as walking and driving, can provide pervasive computing systems with more contextual information and enrich a user's mobility with informative knowledge. In this article, we report on an approach based on supervised learning to automatically infer users' transportation modes, including driving, walking, taking a bus and riding a bike, from raw GPS logs. Our approach consists of three parts: a change point-based segmentation method, an inference model and a graph-based post-processing algorithm. First, we propose a change point-based segmentation method to partition each GPS trajectory into separate segments of different transportation modes. Second, from each segment, we identify a set of sophisticated features, which are not affected by differing traffic conditions (e.g., a person's direction when in a car is constrained more by the road than any change in traffic conditions). Later, these features are fed to a generative inference model to classify the segments of different modes. Third, we conduct graph-based postprocessing to further improve the inference performance. This postprocessing algorithm considers both the commonsense constraints of the real world and typical user behaviors based on locations in a probabilistic manner. The advantages of our method over the related works include three aspects. (1) Our approach can effectively segment trajectories containing multiple transportation modes. (2) Our work mined the location constraints from user-generated GPS logs, while being independent of additional sensor data and map information like road networks and bus stops. (3) The model learned from the dataset of some users can be applied to infer GPS data from others. Using the GPS logs collected by 65 people over a period of 10 months, we evaluated our approach via a set of experiments. As a result, based on the change-point-based segmentation method and Decision Tree-based inference model, we achieved prediction accuracy greater than 71 percent. Further, using the graph-based post-processing algorithm, the performance attained a 4-percent enhancement.

Urban air pollution has caused public concern globally because it seriously affects human life. Modern monitoring systems providing pollution information with high spatio-temporal resolution have been developed to identify personal exposures. However, these systems’ hardware specifications and configurations are usually fixed according to the applications. They can be inconvenient to maintain, and difficult to reconfigure and expand with respect to sensing capabilities. This paper aims at tackling these issues by adopting the proposed Modular Sensor System (MSS) architecture and Universal Sensor Interface (USI), and modular design in a sensor node. A compact MSS sensor node is implemented and evaluated. It has expandable sensor modules with plug-and-play feature and supports multiple Wireless Sensor Networks (WSNs). Evaluation results show that MSS sensor nodes can easily fit in different scenarios, adapt to reconfigurations dynamically, and detect low concentration air pollution with high energy efficiency and good data accuracy. We anticipate that the efforts on system maintenance, adaptation, and evolution can be significantly reduced when deploying the system in the field.