Browsing by Author "Peng, Wen-Chih"

Now showing 1 - 20 of 93

A Framework for Discovering Maritime Traffic Conflict from AIS Network
(2017-01-01)
Lei, Po-Ruey
;
Tsai, Tzu-Hao
;
Wen, Yu-Ting
;
Peng, Wen-Chih
;
交大名義發表
Thanks for the common use of Automatic Identification System (AIS) network has made a large number of the maritime traffic data to be available. Ships equipped with AIS automatically exchange navigational information with nearby ships and terrestrial AIS receivers to facilitate the tracking and monitoring of ships' location and movement for collision avoidance and control. Obviously, with increasing amount of maritime shipping traffic, the navigational collisions are one of the growing safety concerns in maritime traffic situation awareness. To understand the collision situations can help the maritime traffic managers to improve the safety control of maritime traffic. However, it is difficult to statistically analyze such collision due to the number of collected real cases of collisions are relatively low within a short period of time. To overcome the problem of low sample size, we discover traffic conflict from data collected by AIS network to substitute the real collision. Given a set of maritime traffic data collected from AIS network, we try to discover ships' movements that have conflict behaviors and these behaviors may bring a possible collision if they do not take any evasive action. We propose a framework of Clustering-and-Detection to automatically discover the clusters of conflict trajectory from AIS trajectory data in an unsupervised way. Based on real AIS data, the experimental results show that the proposed framework is able to effectively discover sets of trajectory with conflict situation from maritime AIS traffic data. The statistical analysis on the discovered sets of conflict trajectory is able to provide useful knowledge for maritime traffic monitoring.
A Framework of Mining Semantic Regions from Trajectories
(2011)
Lu, Chun-Ta
;
Lei, Po-Ruey
;
Peng, Wen-Chih
;
Su, Ing-Jiunn
;
交大名義發表
With the pervasive use of mobile devices with location sensing and positioning functions, such as Wi-Fi and GPS, people now are able to acquire present locations and collect their movement. As the availability of trajectory data prospers, mining activities hidden in raw trajectories becomes a hot research problem. Given a set of trajectories, prior works either explore density-based approaches to extract regions with high density of GPS data points or utilize time thresholds to identify users' stay points. However, users may have different activities along with trajectories. Prior works only can extract one kind of activity by specifying thresholds, such as spatial density or temporal time threshold. In this paper, we explore both spatial and temporal relationships among data points of trajectories to extract semantic regions that refer to regions in where users are likely to have some kinds of activities. In order to extract semantic regions, we propose a sequential clustering approach to discover clusters as the semantic regions from individual trajectory according to the spatial-temporal density. Based on semantic region discovery, we develop a shared nearest neighbor (SNN) based clustering algorithm to discover the frequent semantic region where the moving object often stay, which consists of a group of similar semantic regions from multiple trajectories. Experimental results demonstrate that our techniques are more accurate than existing clustering schemes.
A Framework of Traveling Companion Discovery on Trajectory Data Streams
(2013-12-01)
Tang, Lu-An
;
Zheng, Yu
;
Yuan, Jing
;
Han, Jiawei
;
Leung, Alice
The advance of mobile technologies leads to huge volumes of spatio-temporal data collected in the form of trajectory data streams. In this study, we investigate the problem of discovering object groups that travel together (i.e., traveling companions) from trajectory data streams. Such technique has broad applications in the areas of scientific study, transportation management, and military surveillance. To discover traveling companions, the monitoring system should cluster the objects of each snapshot and intersect the clustering results to retrieve moving-together objects. Since both clustering and intersection steps involve high computational overhead, the key issue of companion discovery is to improve the efficiency of algorithms. We propose the models of closed companion candidates and smart intersection to accelerate data processing. A data structure termed traveling buddy is designed to facilitate scalable and flexible companion discovery from trajectory streams. The traveling buddies are microgroups of objects that are tightly bound together. By only storing the object relationships rather than their spatial coordinates, the buddies can be dynamically maintained along the trajectory stream with low cost. Based on traveling buddies, the system can discover companions without accessing the object details. In addition, we extend the proposed framework to discover companions on more complicated scenarios with spatial and temporal constraints, such as on the road network and battlefield. The proposed methods are evaluated with extensive experiments on both real and synthetic datasets. Experimental results show that our proposed buddy-based approach is an order of magnitude faster than the baselines and achieves higher accuracy in companion discovery.
A Fusion-based Approach for User Activities Recognition on Smart Phones
(2015)
Njoo, Gunarto Sindoro
;
Ruan, Xiao Wen
;
Hsu, Kuo-Wei
;
Peng, Wen-Chih
;
交大名義發表
In the recent years, several research works have been conducted on collecting context data from various sensors for activity inference. We observe that users perform several actions in their mobile phones: taking photos, performing checkins, and accessing Wi-Fi networks. These actions generate spatial-temporal data that could be utilized to capture user activities. Spatial-temporal data could indicate that a user stays in a certain location at a particular time for a certain activity. In addition, by referring to social media data, one could also infer user activities. Three types of features are extracted for activity inference: 1) geographical feature, indicating where a user performs activities; 2) temporal feature, indicating when a user performs activities; and 3) semantic feature, showing the semantic concept of a place from location-based social networks. Here, we propose Spatial-Temporal Activity Inference Model (STAIM) to infer user activities from data with those three features. In addition, to determine the weight for each feature, we further propose three methods based on frequency, entropy, and entropy-frequency. Experimental results show that STAIM is able to effectively infer user activities, achieving 75% accuracy on average. Moreover, STAIM could infer user activities even when there is no training data (with some performance loss). Moreover, sensitive analysis of parameters is also conducted to select the most optimal parameter.
A Gamma-based Regression for Winning Price Estimation in Real-Time Bidding Advertising
(2017-01-01)
Zhu, Wen-Yuan
;
Shih, Wen-Yueh
;
Lee, Ying-Hsuan
;
Peng, Wen-Chih
;
Huang, Jiun-Long
In Real-Time Bidding (RTB) advertising, estimating the winning price is an important task in evaluating the bid cost of bid requests in Demand-Side Platforms (DSPs). The prior works utilize censored linear regression for winning price estimation by considering both winning and losing bid records. In the traditional regression models, the winning price of each bid request is based on Gaussian distribution. However, the property of Gaussian distribution is not suitable for the winning price of each bid request, and it is hard to link the physical meaning of Gaussian distribution and the winning price. Therefore, in this paper, based on our observation and analysis, the winning price of each bid request is modeled by a unique gamma distribution with respect to its features. Then we propose a gamma-based censored linear regression with regularization for winning price estimation. To derive the parameters of our proposed complicated model based on bid records, our approach is to divide this hard problem into two sub-problems, which are easier to solve. In practice, we also provide four heuristic initial parameter settings that are able to greatly reduce the computation cost when deriving the parameters. The experimental results demonstrate that our approach is highly effective for estimating the winning price compared with the state-of-the-art approaches in three real datasets.
A Novel System for Extracting Useful Correlation in Smart Home Environment
(2013)
Chen, Yi-Cheng
;
Peng, Wen-Chih
;
Lee, Wang-Chien
;
資訊工程學系
;
Department of Computer Science
Owing to the great advent of sensor technology, the usage data of appliances in a house can be logged and collected easily today. However, it is a challenge for the residents to visualize how these appliances are used. Thus, mining algorithms are much needed to discover appliance usage patterns. Most previous studies on usage pattern discovery are mainly focused on analyzing the patterns of single appliance rather than mining the usage correlation among appliances. In this paper, a novel system, namely, Correlation Pattern Mining System (CPMS), is developed to capture the usage patterns and correlations among appliances. With several new optimization techniques, CPMS can reduce the search space effectively and efficiently. Furthermore, the proposed algorithm is applied on a real-world dataset to show the practicability of correlation pattern mining.
A Profile-based Framework for Interaction Prediction
(2012)
Liao, Zhung-Xun
;
Peng, Wen-Chih
;
Yu, Philip S.
;
資訊工程學系
;
Department of Computer Science
In this paper, we generalize the link prediction problem to an interaction prediction problem. Compared with links in social networks, interactions can occur several times repeatedly. Based on the observation, we formulate an event-triggered interaction prediction problem. For example, we may want to know when a user connects to a website (e.g. Facebook), who will also connect to the website. We propose a Profile-based Interaction Prediction Framework (PIPF) which can solve the event-triggered interaction prediction problem efficiently and effectively. In PIPF, we first transform the interaction log into a Sliding-window Evolving Graph (SEG) to reduce the data volume and incrementally update SEG as interaction log grows. Then, we build profiles designed to present users' behavior by extracting the static and surprising features from SEG. The static (respectively, surprising) feature reflects the regularity of users' behavior (respectively, the temporal behavior). When an event occurs, we compute the similarity between the event and each candidate link. We propose two similarity functions for static and surprising features and an automatic selection strategy to control the influence of the two features. We use a real dataset that records Internet connections to evaluate the scalability, efficiency, and effectiveness of PIPF. The experimental results show that PIPF is far more scalable and efficient than the previous methods to perform real-time prediction.
A regression-based approach for mining user movement patterns from random sample data
(2011-01-01)
Hung, Chih-Chieh
;
Peng, Wen-Chih
;
資訊工程學系
;
Department of Computer Science
Mobile computing systems usually express a user movement trajectory as a sequence of areas that capture the user movement trace. Given a set of user movement trajectories, user movement patterns refer to the sequences of areas through which a user frequently travels. In an attempt to obtain user movement patterns for mobile applications, prior studies explore the problem of mining user movement patterns from the movement logs of mobile users. These movement logs generate a data record whenever a mobile user crosses base station coverage areas. However, this type of movement log does not exist in the system and thus generates extra overheads. By exploiting an existing log, namely, call detail records, this article proposes a Regression-based approach for mining User Movement Patterns (abbreviated as RUMP). This approach views call detail records as random sample trajectory data, and thus, user movement patterns are represented as movement functions in this article. We propose algorithm LS (standing for Large Sequence) to extract the call detail records that capture frequent user movement behaviors. By exploring the spatio-temporal locality of continuous movements (i.e., a mobile user is likely to be in nearby areas if the time interval between consecutive calls is small), we develop algorithm TC (standing for Time Clustering) to cluster call detail records. Then, by utilizing regression analysis, we develop algorithm MF (standing for Movement Function) to derive movement functions. Experimental studies involving both synthetic and real datasets show that RUMP is able to derive user movement functions close to the frequent movement behaviors of mobile users. (C) 2010 Elsevier B.V. All rights reserved.
A storage management for mining object moving patterns in object tracking sensor networks
(2006)
Hung, Chih-Chieh
;
Peng, Wen-Chih
;
資訊工程學系
;
Department of Computer Science
One promising application of sensor networks is object tracking. Because the movements of the tracked objects usually show repeating patterns, we propose a heterogeneous tracking model, referred to as HTM, to efficiently mine object moving patterns and track objects. To ensure the quality of moving patterns, we develop a storage management to facilitate mining object moving patterns. Specifically, we explore load-balance feature to store more moving data for mining moving patterns. Once a storage of a cluster head is occupied by moving data, we devise a replacement strategy to replace the less informative patterns. Simulation results show that HTM with storage management is able not only to increase the accuracy of predition but also to save more energy in tracking objects.
A Study on Multiple Wearable Sensors for Activity Recognition
(2017-01-01)
Huang, Yu-Chuan
;
Yi, Chih-Wei
;
Peng, Wen-Chih
;
Lin, Hsing-Chen
;
Huang, Ching-Yu
In the past few years, human activity recognition is an active area of machine learning. The possible applications include daily activity monitoring for elders, exercise and fitness workout assistant systems, life style analysis, etc. In this work, tri-axial accelerometers were worn at the right wrist, left wrist and waist to collect motion data for activity recognition. Three supervised machine learning algorithms including random forests, decision trees and support vector machines were implemented to classify daily activities into running, walking, standing, sitting and dining from inertial data. The purposes of this study are to understand how good the machine learning algorithms can achieve and how the wearing location and number of sensors impact the recognition accuracy. Our results showed that the multi-sensors achieve the accuracy of 81%, and dominant hand sensor achieves the accuracy of 80%, which is 7% higher than non-dominant hand sensor.
A Temporal Probabilistic Model for Dynamic Circle Recommendation in Mobile Applications
(2013)
Chou, Fan-Kai
;
Chiang, Meng-Fen
;
Peng, Wen-Chih
;
資訊工程學系
;
Department of Computer Science
This paper presents a novel framework for dynamic circle recommendation for a query user at a given time point from historical communication logs. We identify the fundamental factors that govern interactions and aim to automatically form friend circles for scenarios, such as, who should I share the photo with in the early morning? Whose post should be listed on top of my Facebook Wall feed at night? We develop a temporal probabilistic model that not only captures temporal tendencies between the query user and each friend candidate but also blends frequency and recency into circle formation. Experimental results on Enron dataset and Call Detail Records prove the effectiveness of dynamic circle formation with proposed temporal probabilistic model.
Activity Sense Organs: Energy-efficient Activity Sensing with Adaptive Duty Cycle Control
(2016-01-01)
Wu, Fang-Jing
;
Tseng, Yu-Chee
;
Peng, Wen-Chih
;
資訊工程學系
;
Department of Computer Science
Recently, smartphones with built-in sensors are exploited to detect human activities. Since sensing human activities may exhaust energy of a smartphone, this work brings up the concept of "activity sense organs" to the uprising smartphone sensing paradigm. As the human sense organs are the most important sensors to perceive real-world information, this work identifies the most important sensors to perceive a particular human activity based on the "quality of the features" contributed by sensors. To optimize the energy usage of activity sensing, we exploit the importance of sensors to profile the duty-cycles of sensors for activity detection. Specifically, a smartphone will operate in a duty-cycled sensing mode, where sensors in the activity sense organs will collect data at the higher sampling rate than those sensors not in the activity sense organs. We design a knowledge lifecycle to find out the activity sense organs, profile the sensing modes of activity detection, and adjust the sensing mode for real-time activity detection. The experimental results indicate that the activity sense organs provide sufficient information for activity detection.
An Efficient Geometry Data Allocation Algorithm in Cloud Computing Environments
(2012-01-01)
Wang, Kun-Wei
;
Huang, Bo-Wei
;
Peng, Wen-Chih
;
資訊工程學系
;
Department of Computer Science
The number of location-based services is growing and developing. Usually, these services put a huge amount of effort into geometry data computation. Thus, their workload is generally high. By exploring cloud computing techniques, one could utilize a number of computing nodes to distribute the workload of the systems. However, the workload is usually not equally balanced across computing nodes, if data is not well-distributed. To make the best use of computing nodes, we propose a sophisticated data distribution technology for geometry computation processing. Intuitively, one can simply divide geometry data into tiles so that the geometry data in each tile can be stored on one computing node. Unfortunately, since data in a tile shares spatial-proximity, processing a geometry computation on spatial-proximity data still incurs a huge workload. To address this issue, we propose a new data distribution approach, Reversed K-means, to distribute geometry data that shares spatial-proximity across different computing nodes. In this way, we can use more computing nodes to process geometry computation and get better performance. To evaluate the performance of our proposed algorithm, we evaluate the utility of computing nodes and the response time when performing geometry computations. The experimental results show that the utility of the computing nodes is higher than existing methods, and the response time is the fastest of all methods.
An incremental algorithm for clustering spatial data streams: exploring temporal locality
(2013-11-01)
Wei, Ling-Yin
;
Peng, Wen-Chih
;
資訊工程學系
;
Department of Computer Science
Clustering sensor data discovers useful information hidden in sensor networks. In sensor networks, a sensor has two types of attributes: a geographic attribute (i.e, its spatial location) and non-geographic attributes (e.g., sensed readings). Sensor data are periodically collected and viewed as spatial data streams, where a spatial data stream consists of a sequence of data points exhibiting attributes in both the geographic and non-geographic domains. Previous studies have developed a dual clustering problem for spatial data by considering similarity-connected relationships in both geographic and non-geographic domains. However, the clustering processes in stream environments are time-sensitive because of frequently updated sensor data. For sensor data, the readings from one sensor are similar for a period, and the readings refer to temporal locality features. Using the temporal locality features of the sensor data, this study proposes an incremental clustering (IC) algorithm to discover clusters efficiently. The IC algorithm comprises two phases: cluster prediction and cluster refinement. The first phase estimates the probability of two sensors belonging to a cluster from the previous clustering results. According to the estimation, a coarse clustering result is derived. The cluster refinement phase then refines the coarse result. This study evaluates the performance of the IC algorithm using synthetic and real datasets. Experimental results show that the IC algorithm outperforms exiting approaches confirming the scalability of the IC algorithm. In addition, the effect of temporal locality features on the IC algorithm is analyzed and thoroughly examined in the experiments.
An Intelligent System for Mining Usage Patterns from Appliance Data in Smart Home Environment
(2012)
Chen, Yi-Cheng
;
Ko, Yu-Lun
;
Peng, Wen-Chih
;
資訊工程學系
;
Department of Computer Science
In the last decade, considerable concern has arisen over the electricity saving due to the issue of reducing greenhouse gases. Previous studies on usage pattern utilization mainly are focused on power disaggregation and appliance recognition. Little attention has been paid to utilizing pattern mining for the target of energy saving. In this paper, we develop an intelligent system which analyzes appliance usage to extract users' behavior patterns in a smart home environment. With the proposed system, users can acquire the electricity consumption of each appliance for energy saving easily. In advance, if the electricity cost is high, users can observe the abnormal usage of appliances from the proposed system. Furthermore, we also apply our system on real-world dataset to show the practicability of mining usage pattern in smart home environment.
Anomaly secure detection methods by analyzing dynamic characteristics of the network traffic in cloud communication?
(2014-02-10)
Xiong, Wei
;
Hu, Hanping
;
Xiong, Naixue
;
Yang, Laurence T.
;
Peng, Wen-Chih
Cloud computing represents a new paradigm where computing resources are offered as services in the world via communication Internet. As many new types of attacks are arising at a high frequency, the cloud computing services are exposed to an increasing amount of security threats. To reduce security risks, two approaches of the network traffic anomaly detection in cloud communications have been presented, which analyze dynamic characteristics of the network traffic based on the synergetic neural networks and the catastrophe theory. In the former approach, a synergetic dynamic equation with a group of the order parameters is used to describe the complex behaviors of the network traffic system in cloud communications. When this equation is evolved, only the order parameter determined by the primary factors can converge to 1. Then, the anomaly can be detected. In the latter approach; a catastrophe potential function is introduced to describe the catastrophe dynamic process of the network traffic in cloud communications. When anomalies occur, the state of the network traffic will deviate from the normal one. To assess the deviation, an index named as catastrophe distance is defined. The network traffic anomaly can be detected by the value of this index. We evaluate the performance of these two approaches using the standard Defense Advanced Research Projects Agency data sets. Experimental results show that our approaches can effectively detect the network traffic anomaly and achieve the high detection probability and the low false alarms rate. (C) 2013 Published by Elsevier Inc.
AppNow: Predicting Usages of Mobile Applications on Smart Phones
(2012)
Liao, Zhung-Xun
;
Lei, Po-Ruey
;
Shen, Tsu-Jou
;
Li, Shou-Chung
;
Peng, Wen-Chih
Due to the proliferation of mobile applications (abbreviated as Apps) on smart phones, users can install many Apps to facilitate their life. Usually, users browse their Apps by swiping touch screen on smart phones, and are likely to spend much time on browsing Apps. In this paper, we design an AppNow widget that is able to predict users' Apps usage. Therefore, users could simply execute Apps from the widget. The main theme of this paper is to construct the temporal profiles which identify the relation between Apps and their usage times. In light of the temporal profiles of Apps, the AppNow widget predicts a list of Apps which are most likely to be used at the current time. In our experiments, we collected real usage traces to show that the accuracy of AppNow could reach 86% for identifying temporal profiles and 90% for predicting App usage.
Associative Classification for Human Activity Inference on Smart Phones
(2014-01-01)
Peng, Yu-Hsiang
;
Njoo, Gunarto Sindoro
;
Li, Shou-Chun
;
Peng, Wen-Chih
;
資訊工程學系
With the population of smart phones, the general trend of human activity inference is prospering under a powerful computation capabilities on modern phones. Such an assistant make users life more convenient and help them prevent from unnecessary interferences. In conventional research, the activity inference problem is considered a classification instance, so in this paper we propose an association-based classifier framework (ACF) that aims at exploring the correlation among collected sensor data. Each data consists of multiple sensor readings with a label, e.g., dining, shopping, working, driving, sporting, and entertaining. Note that ACF caters to the discrete data; as a consequence, the continuous sensor readings are needed to be transformed to some discrete groups. Therefore, we propose an Interval Length-Gini Discretization (LGD) method which considers the groups and misclassified cases to obtain the best hypothesis for a given set of data. After an appropriate discretization, we propose one-cut and memory-iteration-based approach to select a set of useful sensor-value pairs for reducing the model size by removing redundant features and guaranteeing an acceptable accuracy. In the experiments our framework has a good performance on real data set collected from 50 participants in eight months, and a smaller size than the existing classifications.
CAPatternMiner: Mining Ship Collision Avoidance Behavior from AIS Trajectory Data
(2018-01-01)
Lei, Po-Ruey
;
Xiao, Li-Pin
;
Wen, Yu-Ting
;
Peng, Wen-Chih
;
交大名義發表
The improvement of collision avoidance for ship navigation in encounter situation is an important topic in maritime traffic safety. Most research on maritime collision avoidance has focused on planning a safe path for a ship to keep away from the approaching ship under the requirements of the International Regulations for Preventing Collision at Sea (COLREGs). However, the specific anticollision actions are actually carried out by the navigators' own experience according to the local encounter situation. In this paper, different from the existing works, we discover the collision avoidance behavior from real ships' movement, i.e., AIS trajectory data. However, the uncertainty of maritime trajectory data brings the challenge of collision avoidance behavior mining. To achieve our goal, we propose CAPatternMiner to provide a framework to discover the ships' anti-collision behavior, which is effective in the encounter situation, and generate the discovered behavior in form of collision avoidance pattern. Furthermore, a prototype of CAPatternMiner is built for pattern analysis and visualization and also benefits a deeper understanding of collision avoidance behavior on maritime traffic. The proposed framework will be applied to the developing of pattern-aware collision avoidance system to improve the maritime traffic safety.
CIM: Community-Based Influence Maximization in Social Networks
(2014-04-01)
Chen, Yi-Cheng
;
Zhu, Wen-Yuan
;
Peng, Wen-Chih
;
Lee, Wang-Chien
;
Lee, Suh-Yin
Given a social graph, the problem of influence maximization is to determine a set of nodes that maximizes the spread of influences. While some recent research has studied the problem of influence maximization, these works are generally too time consuming for practical use in a large-scale social network. In this article, we develop a new framework, community-based influence maximization (CIM), to tackle the influence maximization problem with an emphasis on the time efficiency issue. Our proposed framework, CIM, comprises three phases: (i) community detection, (ii) candidate generation, and (iii) seed selection. Specifically, phase (i) discovers the community structure of the network; phase (ii) uses the information of communities to narrow down the possible seed candidates; and phase (iii) finalizes the seed nodes from the candidate set. By exploiting the properties of the community structures, we are able to avoid overlapped information and thus efficiently select the number of seeds to maximize information spreads. The experimental results on both synthetic and real datasets show that the proposed CIM algorithm significantly outperforms the state-of-the-art algorithms in terms of efficiency and scalability, with almost no compromise of effectiveness.