Air pollution has become an extremely serious problem, with particulate matter having a significantly greater impact on human health than other contaminants. The small diameter of fine (PM2.5) allows it to penetrate deep into the alveoli as far bronchioles, interfering gas exchange within lungs. Long-term exposure been shown cause cardiovascular disease, respiratory and increase risk lung cancers. Therefore, forecasting air quality also important help guide individual actions. This paper...

In the data stream environment, patterns generated at different time instances are due to evolution. As progresses, behavior and members of clusters usually change. Hence, clustering continuous streams allows us observe changes group behavior. order support flexible requirements, we devise in this paper a Clustering on Demand framework, abbreviated as COD dynamically cluster multiple streams. While providing general framework streams, has two advantageous features, namely, one scan for...

This paper investigates the transportation and vehicular modes classification by using big data from smartphone sensors. The three types of sensors used in this include accelerometer, magnetometer, gyroscope. study proposes improved features uses machine learning algorithms including decision trees, K-nearest neighbor, support vector to classify user's modes. In experiments, we discussed compared performance different perspectives accuracy for both modes, executive time, model size. Results...

In recent years, waveform-mapping-based speech enhancement (SE) methods have garnered significant attention. These generally use a deep learning model to directly process and reconstruct waveforms. Because both the input output are in waveform format, SE can overcome distortion caused by imperfect phase estimation, which may be encountered spectral-mapping-based systems. So far, most focused on single-channel tasks. this article, we propose novel fully convolutional network (FCN) with Sinc...

Although there have been many recent studies on the mining of sequential patterns in a static database and with increasing data, these works, general, do not fully explore effect deleting old data from sequences database. When are generated, newly arriving may be identified as frequent due to existence sequences. Even worse, obsolete that recently stay reported results. In practice, users usually more interested than ones. To capture dynamic nature addition deletion, we propose general model...

Instead of finding clusters in the full feature space, subspace clustering is an emergent task which aims at detecting embedded subspaces. Most previous works literature are density-based approaches, where a cluster regarded as high-density region subspace. However, identification dense regions lacks considering critical problem, called "the density divergence problemrdquo this paper, refers to phenomenon that densities vary different cardinalities. Without utilize threshold discover all...

Generally speaking, to implement Apriori-based association rule mining in hardware, one has load candidate itemsets and a database into the hardware. Since capacity of hardware architecture is fixed, if number or items larger than capacity, are loaded separately. The time complexity those steps that need proportion multiplied by database. Too many large would create performance bottleneck. In this paper, we propose HAsh-based Pipelined (abbreviated as HAPPI) for hardware- enhanced mining. We...

We investigate the general model of mining associations in a temporal database, where exhibition periods items are allowed to be different from one another. The database is divided into partitions according time granularity imposed. Such association rules allow us observe short-term but interesting patterns that absent when whole range evaluated altogether. Prior work may omit some and thus have limited practicability. To remedy this give more precise frequent itemsets, we devise an...

Outlier detection in data streams is crucial to successful mining. However, this task made increasingly difficult by the enormous growth quantity of generated expansion Internet Things (IoT). Recent advances outlier based on density-based local factor (LOF) algorithms do not consider variations that change over time. For example, there may appear a new cluster points time stream. Therefore, we present novel algorithm for streaming data, referred as time-aware incremental (TADILOF) overcome...

In the data stream environment, patterns generated by mining techniques are usually distinct at different time because of evolution data. order to deal with various types multiple streams and support flexible requirements, we devise in this paper a clustering on demand framework, abbreviated as COD dynamically cluster streams. While providing general framework streams, has two major features, namely one scan for online statistics collection compact multiresolution approximations, which...

This study explores the spatial-temporal patterns of particulate matter (PM) in Taiwan. Probability map PM and daily are discussed this study. Data mining provides more detailed information for variations trends. The proposed model will show that data a relatively high goodness fit sufficient space-time explanatory power, particularly air pollution frequency affect areas. In model, method using Dynamic Time Warping is to analyse temporal similarity between stations. can eliminate global...

Identifying motifs in promoter regions is crucial to our understanding of transcription regulation. Researchers commonly use known features a variety species predict motifs. However the results are not particularly useful. Different rarely have similar binding sites. In this study, we adopt sequence analysis techniques find possible sites among different species. We sought improve existing algorithm suit task mining sequential patterns with specific number gaps. Moreover, discuss...

Most of the prior research works in data broadcasting are based on assumption that disseminated items independent one another. Since many applications, a mobile user will be interested more than item simultaneously, we discuss this paper issue dependency generating broadcast program. Algorithm PBA, standing for Placement-Based Allocation, is proposed to generate program with high quality and low complexity dependent environment. The experimental results show placement-based allocation...