Browsing by Author "Fu, Hsin-Chia"

Now showing 1 - 20 of 20

A polygon description based similarity measurement of stock market behavior
(2010-03-01)
Lai, Por-Shen
;
Fu, Hsin-Chia
;
資訊工程學系
;
Department of Computer Science
In this paper. we propose and implement a data modeling system to measure the similarity of stock market behavior in different time periods. To implement this system, we also propose a data distribution approximation model, Polygon descriptor, and a shape difference measurement. called Deforming distance method The Polygon descriptor can model a data distribution by characterizing the data dependency relationship among the data variables, and the Deforming distance is designed to be translation, scale, and rotation Invariant. The Polygon descriptor and Deforming distance method are combined to measure the similarity between data sets based on the shape of data distribution. Stock price and EPS (earn per share) data during different time periods were selected to verify the modeling and measuring power of the proposed model and method. The Polygon descriptor is used to model the collected Stock price and EPS data for their dependency relationship Then, difference between Polygon descriptors can be measured by using the Deforming distance Measurement method for their Deforming distance Values Based oil the Deforming difference Values. the similarity of stock market behaviors in two time periods can be assessed, and the similar stock market behavior of different time periods can be identified To demonstrate the capabilities of the proposed method and system. real-world data of Taiwan stock market during the period from 1986 to 2006 were collected and used. Experimental results show that the Polygon descriptor successfully captures the data dependencies between stock price and EPS, and the Deforming difference based similarity measurement estimates the changes of market behavior better than some commonly used methods. A web-based prototype system is available at http://www.csie.cntu.edu tw/-pslar /TWStockSldx/ for public trial. evaluation, comments and suggestions (C) 2009 Elsevier Ltcl All rights reserved.
A prototypes-embedded genetic K-means algorithm
(2006)
Cheng, Shih-Sian
;
Chao, Yi-Hsiang
;
Wang, Hsin-Min
;
Fu, Hsin-Chia
;
資訊工程學系
This paper presents a genetic algorithm (GA) for K-means clustering. Instead of the widely applied string-of-group-numbers encoding, we encode the prototypes of the clusters into the chromosomes. The crossover operator is designed to exchange prototypes between two chromosomes. The one-step K-means algorithm is used as the mutation operator. Hence, the proposed GA is called the prototypes-embedded genetic K-means algorithm (PGKA). With the inherent evolution process of evolutionary algorithms, PGKA has superior performance than the classical K-means algorithm, while comparing to other GA-based approaches, PGKA is more efficient and suitable for large scale data sets.
An EM based multiple instance learning method for image classification
(2008-10-01)
Pao, H. T.
;
Chuang, S. C.
;
Xu, Y. Y.
;
Fu, Hsin-Chia
;
資訊工程學系
In this paper, we propose an EM based learning algorithm to provide a comprehensive procedure for maximizing the measurement of diverse density on given multiple Instances. Furthermore, the new EM based learning framework converts an MI problem into a single-instance treatment by using EM to maximize the instance responsibility for the corresponding label of each bag. To learn a desired image class, a user may select a set of exemplar images and label them to be conceptual related (positive) or conceptual unrelated (negative) images. A positive image consists of at least one object that the user may be interested, and a negative image should not contain any object that the user may be interested. By using the proposed EM based learning algorithm, an image retrieval prototype system is implemented. Experimental results show that for only a few times of relearning cycles, the prototype system can retrieve user's favor images from WWW over Internet. (C) 2007 Published by Elsevier Ltd.
Automatic construction of regression class tree for MLLR via model-based hierarchical clustering
(2006)
Cheng, Shih-Sian
;
Xu, Yeong-Yuh
;
Wang, Hsin-Min
;
Fu, Hsin-Chia
;
資訊工程學系
In this paper, we propose a model-based hierarchical clustering algorithm that automatically builds a regression class tree for the well-known speaker adaptation technique - Maximum Likelihood Linear Regression (MLLR). When building a regression class tree, the mean vectors of the Gaussian components of the model set of a speaker independent CDHMM-based speech recognition system are collected as the input data for clustering. The proposed algorithm comprises two stages. First, the input data (i.e., all the Gaussian mean vectors of the CDHMMs) is iteratively partitioned by a divisive hierarchical clustering strategy, and the Bayesian Information Criterion (BIC) is applied to determine the number of clusters (i.e., the base classes of the regression class tree). Then, the regression class tree is built by iteratively merging these base clusters using an agglomerative hierarchical clustering strategy, which also uses BIC as the merging criterion. We evaluated the proposed regression class tree construction algorithm on a Mandarin Chinese continuous speech recognition task. Compared to the regression class tree implementation in HTK, the proposed algorithm is more effective in building the regression class tree and can determine the number of regression classes automatically.
BIC-based audio segmentation by divide-and-conquer
(2008)
Cheng, Shih-Sian
;
Wang, Hsin-Min
;
Fu, Hsin-Chia
;
資訊工程學系
;
Department of Computer Science
Audio segmentation has received increasing attention in recent years for its potential applications in automatic indexing and transcription of audio data. Among existing audio segmentation approaches, the BIC-based approach proposed by Chen and Gopalakrishnan is most well-known for its high accuracy. However, this window-growingbased segmentation approach suffers from the high computation cost. In this paper, we propose using the efficient divide-and-conquer strategy in audio segmentation. Our approaches detect acoustic changes by recursively partitioning an analysis window into two sub-windows using Delta BIC. The results of experiments conducted on the broadcast news data demonstrate that our approaches not only have a lower computation cost but also achieve a higher segmentation accuracy than window-growing-based segmentation.
BIC-Based Speaker Segmentation Using Divide-and-Conquer Strategies With Application to Speaker Diarization
(2010-01-01)
Cheng, Shih-Sian
;
Wang, Hsin-Min
;
Fu, Hsin-Chia
;
資訊工程學系
;
Department of Computer Science
In this paper, we propose three divide-and-conquer approaches for Bayesian information criterion (BIC)-based speaker segmentation. The approaches detect speaker changes by recursively partitioning a large analysis window into two sub-windows and recursively verifying the merging of two adjacent audio segments using Delta BIC, a widely-adopted distance measure of two audio segments. We compare our approaches to three popular distance-based approaches, namely, Chen and Gopalakrishnan's window-growing-based approach, Siegler et al.'s fixed-size sliding window approach, and Delacourt and Wellekens's DISTBIC approach, by performing computational cost analysis and conducting speaker change detection experiments on two broadcast news data sets. The results show that the proposed approaches are more efficient and achieve higher segmentation accuracy than the compared distance-based approaches. In addition, we apply the segmentation approaches discussed in this paper to the speaker diarization task. The experiment results show that a more effective segmentation approach leads to better diarization accuracy.
Clean energy, non-clean energy, and economic growth in the MIST countries
(2014-04-01)
Pao, Hsiao-Tien
;
Li, Yi-Ying
;
Fu, Hsin-Chia
;
管理科學系
;
Department of Management Science
This paper explores the causal relationship between clean (renewable/nuclear) and non-clean energy consumption and economic growth in emerging economies of the MIST (Mexico, Indonesia, South Korea, and Turkey) countries. The panel co-integration tests reveal that there is a long-term equilibrium relationship among GDP, capital formation, labor force, renewable/nuclear, and fossil fuel energy consumption. The panel causality results indicate that (1) there is a positive unidirectional short-run causality from fossil fuel energy consumption to economic growth with a bidirectional long-run causality; (2) there is a unidirectional long-run causality from renewable energy consumption to economic growth with positive bidirectional short-run causality, and a long-run causality from renewable to fossil fuel energy consumption with negative short-run feedback effects; and (3) there is a bidirectional long-run causality between nuclear energy consumption and economic growth and a long-run causality from fossil fuel energy consumption to nuclear energy consumption with positive short-run feedback effects. These suggest that MIST countries should be energy-dependent economies and that energy conservation policies may depress their economic development. However, developing renewable and nuclear energy is a viable solution for addressing energy security and climate change issues, and creating clean and fossil fuel energy partnerships could enhance a sustainable energy economy. (c) 2014 Elsevier Ltd. All rights reserved.
Competition and stability analyses among emissions, energy, and economy: Application for Mexico
(2015-03-15)
Pao, Hsiao-Tien
;
Fu, Hsin-Chia
;
管理科學系
;
Department of Management Science
In view of limited natural resources on Earth, linkage among environment, energy, and economy (3Es) becomes important perspectives for sustainable development. This paper proposes to use Lotka-Volterra, model for SUstainable Development (LV-SUD) to analyse the interspecific interactions, equilibria and their stabilities among emissions, different types of energy consumption (renewable, nuclear, and fossil fuel), and real GDP, the main factors of 3Es issues. Modelling these interactions provides a useful multivariate framework for prediction outcomes. Interaction between 3Es, namely competition, symbiosis, or predation, plays an important role in policy development to achieve a balanced use of energy resources and to strengthen the green economy. Applying LV-SUD in Mexico, an emerging markets country, analysing results show that there is a mutualism between fossil fuel consumption and GDP; prey-predator relationships that fossil fuel and GDP enhance the growth of emissions, but emissions inhibit the growth of the others; and commensalisms that GDP benefits from nuclear power, and renewable power benefits from fossil fuel. It is suggested that national energy policies should remain committed to decoupling the relevance between non-clean energy and GDP, to actively developing clean energy and thereby to properly reducing fossil fuel consumption and emissions without harming economic growth. (C) 2015 Elsevier Ltd. All rights reserved.
Constructing and application of multimedia TV-news archives
(2008-10-01)
Pao, H. T.
;
Chen, Y. H.
;
Lai, P. S.
;
Xu, Y. Y.
;
Fu, Hsin-Chia
This paper addresses an integrated information mining techniques for broadcasting TV-news. This utilizes technique from the fields of acoustic, image, and video analysis, for information on news story title, newsman and scene identification. The goal is to construct a compact yet meaningful abstraction of broadcast TV-news, allowing users to browse through large amounts of data in a non-linear fashion with flexibility and efficiency. By adding acoustic analysis, a news program can be partitioned into news and commercial clips, with 90% accuracy on a data set of 400 It TV-news recorded off the air from July 2005 to August 2006. By applying speaker identification and/ or image detection techniques, each news stories can be segmented with a better accuracy of 95.92%. On-screen captions or subtitles are recognized by OCR techniques to produce the text title of each news stories. The extracted title words can be used to link or to navigate more related news contents on the WWW. In cooperation with facial and scene analysis and recognition techniques, OCR results can provide users with multimodal query on specific news stories. Some experimental results are presented and discussed for the system reliability, performance evaluation and comparison. (C) 2007 Published by Elsevier Ltd.
Day-and-Night Video Based Face Identification
(2015)
Chang, Jyh-Yeong
;
Chan, Tzu-Rou
;
Fu, Hsin-Chia
;
電子工程學系及電子研究所
;
Department of Electronics Engineering and Institute of Electronics
Human face recognition system is a desired technique in our daily life. It is a widely well-come technique that can all-day-long and on-line recognize a person from video cameras. To this end, we use a near infrared (NIR) camera to capture day-and-night video images for on-line human recognition. In this paper, we adopt human face sub-image attraction package in OpenCV, which is based on Haar cascade classifier. The package is a feature-based algorithm and works much faster than the pixel-based algorithm. It is to be noted that the image contrast color tones of video frames in the night is worse than that in the day, thus we employ multi-scale retinex to enhance video frames in the night before OpenCV face extraction routine. The extracted face sub-image is first transformed to a new space by eigenspace and canonical space transformation. The recognition is finally done in canonical space. Despite OpenCV\'s popularity to date, extracting face sub-images from taken videos are still not reliable enough. Namely, we can obtain many non-face sub-images among the extracted face sub-images. We judiciously classify the sub-images that are far away from the centroids of persons to be classified as non-face sub-images. This may remedy the shortcoming of OpenCV package, and greatly increase the face recognition accuracy. Furthermore, we consider the most recent three consecutive face image recognitions from video, and use majority vote to recognize a person to enhance the accuracy. Besides, we have tested face image recognition to reject intruders successfully.
Forecasting of CO2 emissions, energy consumption and economic growth in China using an improved grey model
(2012-04-01)
Pao, Hsiao-Tien
;
Fu, Hsin-Chia
;
Tseng, Cheng-Lung
;
資訊工程學系
;
管理科學系
Analyses and forecasts of carbon emissions, energy consumption and real outputs are key requirements for clean energy economy and climate change in rapid growth market such as China. This paper employs the nonlinear grey Bernoulli model (NGBM) to predict these three indicators and proposes a numerical iterative method to optimize the parameter of NGBM. The forecasting ability of NGBM with optimal parameter model, namely NGBM-OP has remarkably improved, compared to the GM and ARIMA. The MAPEs of NGBM-OP for out-of-sample (2004-2009) are ranging from 1.10 to 6.26. The prediction results show that China's compound annual emissions, energy consumption and real GDP growth is set to 4.47%, -0.06% and 6.67%, respectively between 2011 and 2020. The co-integration results show that the long-run equilibrium relationship exists among these three indicators and emissions appear to be real output inelastic and energy consumption elastic. The estimated values cannot support an EKC hypothesis, and real output is significantly negative impact on emissions. In order to promote economic and environmental quality, the results suggest that China should adopt the dual strategy of increasing energy efficiency, reducing the loss in power transmission and distribution and stepping up energy conservation policies to reduce any unnecessary wastage of energy. (C) 2012 Elsevier Ltd. All rights reserved.
Intention deduction from demonstrated trajectory for tool-handling task
(2013)
Chan, Hoa-Yu
;
Young, Kuu-young
;
Fu, Hsin-Chia
;
資訊工程學系
;
電子工程學系及電子研究所
When the robot comes to a home-like environment, its programming becomes very demanding. The concept of learning by demonstration is thus introduced, which may remove the load of detailed analysis and programming from the user. Following this concept, in this article, we propose a novel approach for the robot to deduce the intention of the demonstrator from the trajectories during task execution. We focus on the tool-handling task, which is common in the home environment, but complicated for analysis. The proposed approach does not pre-define-motions or put constraints on motion speed, while allowing the event order to be altered and allowing for the presence of redundant operations during the demonstration. We apply the concept of cross-validation to locate the portions of the trajectory that correspond to delicate and skillful maneuvering, and apply an algorithm based on dynamic programming previously developed to search for the most probable intention. In experiments, we applied the proposed approach for two different kinds of tasks, the pouring and coffee-making tasks, with the number of objects and their locations varied during demonstrations. To further investigate our method's scalability and generality, we also performed intensive analysis on the parameters involved in the tasks.
Intention Learning From Human Demonstration
(2011-05-01)
Chan, Hoa-Yu
;
Young, Kuu-Young
;
Fu, Hsin-Chia
;
資訊工程學系
;
電機工程學系
Equipped with better sensing and learning capabilities, robots nowadays are meant to perform versatile tasks. To remove the load of detailed analysis and programming from the engineer, a concept has been proposed that the robot may learn how to execute the task from human demonstration by itself Following the idea, in this paper, we propose an approach for the robot to learn the intention of the demonstrator from the resultant trajectory during task execution. The proposed approach, identifies the portions of the trajectory that correspond to delicate and skillful maneuvering. Those portions, referred to as motion features, may implicate the intention of the demonstrator. As the trajectory may result from so many possible intentions, it poses a severe challenge on finding the correct one's. We first formulate the problem into a realizable mathematical form and then employ the method of dynamic programming for the search. Experiments based on the pouring and also fruit jam tasks are performed to demonstrate the proposed approach, in which the derived intention is used to execute the same task under different experimental settings.
Interactive Relevance Visual Learning for Image Retrieval
(2015-01-01)
Fu, Hsin-Chia
;
Wang, Z. H.
;
Wang, W. J.
;
Pao, Hsiao-Tien
;
管理科學系
This paper proposes mixture Gaussian neural networks (MGNN) to learn visual features from user specified query image objects or regions for relevance image retrieval. Instead of segmenting query image regions from sample images, relevance feedback feature learning is performed by the proposed MGNN to extract query visual features. After feature learning, the MGNN can be used to measure the appearance difference between the query features and images for image retrieval. The proposed methods were tested on COREL image gallery and the WWW image collections, and testing results were compared with currently leading approaches. From the experimental results, that the extracted and learned query visual features by MGNN can be very close to users\' mind and/or desire, and the closeness is somewhat related to the number of feature leaning iterations. Since any dimensional data can be approximated by mixture Gaussian distributions, thus using MGNN to query and to retrieve similar and/or relevance high dimensional data or images will be a new area of research for future works.
Model-Based Clustering by Probabilistic Self-Organizing Maps
(2009-05-01)
Cheng, Shih-Sian
;
Fu, Hsin-Chia
;
Wang, Hsin-Min
;
資訊工程學系
;
Department of Computer Science
In this paper, we consider the learning process of a probabilistic self-organizing map (PbSOM) as a model-based data clustering procedure that preserves the topological relationships between data clusters in a neural network. Based on this concept, we develop a coupling-likelihood mixture model for the PbSOM that extends the reference vectors in Kohonen's self-organizing map (SOM) to multivariate Gaussian distributions. We also derive three expectation-maximization (EM)-type algorithms, called the SOCEM, SOEM, and SODAEM algorithms, for learning the model (PbSOM) based on the maximum-likelihood criterion. SOCEM is derived by using the classification EM (CEM) algorithm to maximize the classification likelihood; SOEM is derived by using the EM algorithm to maximize the mixture likelihood; and SODAEM is a deterministic annealing (DA) variant of SOCEM and SOEM. Moreover, by shrinking the neighborhood size, SOCEM and SOEM can be interpreted, respectively, as DA variants of the CEM and EM algorithms for Gaussian model-based clustering. The experimental results show that the proposed PbSOM learning algorithms achieve comparable data clustering performance to that of the deterministic annealing EM (DAEM) approach, while maintaining the topology-preserving property.
Renewable energy, non-renewable energy and economic growth in Brazil
(2013-09-01)
Pao, Hsiao-Tien
;
Fu, Hsin-Chia
;
管理科學系
;
Department of Management Science
This study employs Brazil's yearly statistics from 1980 to 2010 to explore the causal relationships between the real GDP and four types of energy consumption: non-hydroelectric renewable energy consumption (NHREC), total renewable energy consumption (TREC), non-renewable energy consumption (NREC), and the total primary energy consumption (TEC). The cointegration test reveals a long-run equilibrium among Brazil's real GDP, labour, capital, and each of the four types of consumption. The development of the Brazilian economy has close ties with capital formation and labour force. The influence of NHREC/TREC on real output is positive and significant, while the impacts by NREC/TEC are insignificant. The results from the vector error correction models reveal a unidirectional causality from NHREC to economic growth, a bidirectional causality between economic growth and TREC, and a unidirectional causality from economic growth to NREC or TEC without feedback in the long-run. These findings suggest that Brazil is an energy-independent economy and that economic growth is crucial in providing the necessary resources for sustainable development. Expanding renewable energy would not only enhance Brazil's economic growth and curb the deterioration of the environment but also create an opportunity for a leadership role in the international system and improve Brazil's competition with more developed countries. (C) 2013 Elsevier Ltd. All rights reserved.
Texture recognition by generalized probabilistic decision-based neural networks
(2011-05-01)
Xu, Yeong-Yuh
;
Tseng, C. -L.
;
Fu, Hsin-Chia
;
資訊工程學系
;
Department of Computer Science
Texture recognition have received tremendous attentions in the past decades, due to its wide applications in computer vision and pattern recognition. For various applications, formulating texture features in distributional forms can sometimes provide meaningful representation than in numerical forms. In this paper, a generalized probabilistic decision-based neural network (GPDNN), based on a novel methodology for the measurement of the difference between two distributions, is proposed for texture recognition. Based on a two-layer pyramid-type network structure, the proposed GPDNN receives texture data via 2-D grid input nodes, and outputs the classification and/or retrieval results at the top layer node. Our prototype system demonstrates a successful utilization of GPDNN to the texture recognition on 40 texture images selected from the MIT Vision Texture (VisTex) database. Regarding the performance, experiment results show that (1) based on the proposed distribution difference measurement method, the texture retrieval accuracy is improved from 77% to 82% by comparing with some recently published leading methods, and (2) the proposed GPDNN has significant improvements in classification accuracy from 82.2% to 90.1% and retrieval accuracy from 79.9% to 88.6% by comparing with traditional approaches. (C) 2010 Elsevier Ltd. All rights reserved.
The causal relationship between energy resources and economic growth in Brazil
(2013-10-01)
Pao, Hsiao-Tien
;
Fu, Hsin-Chia
;
管理科學系
;
Department of Management Science
This study investigates the causal relationship between clean and non-clean energy consumption and economic growth in Brazil over the period of 1980-2009. Clean energy consumption at aggregated level of total renewable energy consumption and disaggregated levels of hydroelectric, new renewables, and nuclear energy consumption are tested within a production function framework. A cointegration test reveals a long-term equilibrium relationship between real output, capital, labor, and renewable and non-renewable energy consumption at aggregated level, and a long-term equilibrium relationship between real output, capital, labor, and hydroelectric/new renewables/nuclear and fossil fuel energy consumption at disaggregated level. The capital, labor, and new renewables elasticities of real output are positive and statistically significant, other energy consumption item's elasticities are insignificant. The results from error correction model reveal the interdependencies between new renewables, nuclear, fossil fuel, and total non-renewable energy consumption and economic growth, the unidirectional causality from hydroelectric/total renewable consumption to economic growth, the substitutability between new renewables and fossil fuel consumption, and the substitutability between new renewables and nuclear energy consumption. Additionally, nuclear and new renewables energy consumption responds to bring the system back to equilibrium. Overall, aggregated analysis may obscure the relationship between different types of clean energy consumption and economic growth. (C) 2013 Elsevier Ltd. All rights reserved.
Variance enhanced K-medoid clustering
(2011-01-01)
Lai, Por-Shen
;
Fu, Hsin-Chia
;
資訊工程學系
;
Department of Computer Science
This paper proposes new variance enhanced clustering methods to improve the popular K-medoid algorithm by adapting variance information in data clustering. Since measuring similarity between data objects is simpler than mapping data objects to data points in feature space, these pairwise similarity based clustering algorithms can greatly reduce the difficulty in developing clustering based pattern recognition applications. A web-based image clustering system has been developed to demonstrate and show the clustering power and significance of the proposed methods. Synthetic numerical data and real-world image collection are applied to evaluate the performance of the proposed methods on the prototype system. As shown as the web-demonstration, the proposed method, variance enhanced K-medoid model, groups similar images in clusters with various variances according to the distribution of image similarity values. (C) 2010 Elsevier Ltd. All rights reserved.
Weighted bootstrap for neural model selection
(2008)
Chuang, Shun-Chin
;
Hung, Wen-Liang
;
Fu, Hsin-Chia
;
資訊工程學系
;
Department of Computer Science
This article proposes a weighted bootstrap procedure, which is an efficient bootstrap technique for neural model selection. Our primary interest in reducing computer effort is to not resample (in the original bootstrap procedure) uniformly from the original sample, but to modify this distribution in order to obtain variance reduction. The performance of the weighted bootstrap is demonstrated on two artificial data sets and one real dataset. Experimental results show that the weighted bootstrap procedure permits an approximately 2 to 1 reduction in replication size.