Browsing by Author "Chang, Ruei-Chuan"

Now showing 1 - 8 of 8

A kernel-based discrimination framework for solving hypothesis testing problems with application to speaker verification
(2006)
Chao, Yi-Hsiang
;
Tsai, Wei-Ho
;
Wang, Hsin-Min
;
Chang, Ruei-Chuan
;
資訊工程學系
Real-word applications often involve a binary hypothesis testing problem with one of the two hypotheses ill-defined and hard to be characterized precisely by a single measure. In this paper, we develop a framework that integrates multiple hypothesis testing measures into a unified decision basis, and apply kernel-based classification techniques, namely, Kernel Fisher Discriminant (KFD) and Support Vector Machine (SVM), to optimize the integration. Experiments conducted on speaker verification demonstrate the superiority of our approaches over the predominant approaches.
A novel alternative hypothesis characterization using kernel classifiers for LLR-based speaker verification
(2006)
Chao, Yi-Hsiang
;
Wang, Hsin-Min
;
Chang, Ruei-Chuan
;
資訊工程學系
;
Department of Computer Science
In a log-likelihood ratio (LLR)-based speaker verification system, the alternative hypothesis is usually ill-defined and hard to characterize a priori, since it should cover the space of all possible impostors. In this paper, we propose a new LLR measure in an attempt to characterize the alternative hypothesis in a more effective and robust way than conventional methods. This LLR measure can be further formulated as a non-linear discriminant classifier and solved by kernel-based techniques, such as the Kernel Fisher Discriminant (KFD) and Support Vector Machine (SVM). The results of experiments on two speaker verification tasks show that the proposed methods outperform classical LLR-based approaches.
GSR: A global seek-optimizing real-time disk-scheduling algorithm
(2007-02-01)
Chang, Hsung-Pin
;
Chang, Ray-I
;
Shih, Wei-Kuan
;
Chang, Ruei-Chuan
;
資訊工程學系
Earliest-deadline-first (EDF) is good for scheduling real-time tasks in order to meet timing constraint. However, it is not good enough for scheduling real-time disk tasks to achieve high disk throughput. In contrast, although SCAN can maximize disk throughput, its schedule results may violate real-time requirements. Thus, during the past few years, various approaches were proposed to combine EDF and SCAN (e.g., SCAN-EDF and RG-SCAN) to resolve the real-time disk-scheduling problem. However, in previous schemes, real-time tasks can only be rescheduled by SCAN within a local group. Such restriction limited the obtained data throughput. In this paper, we proposed a new globally rescheduling scheme for real-time disk scheduling. First, we formulate the relations between the EDF schedule and the SCAN schedule of input tasks as EDF-to-SCAN mapping (ESM). Then, on the basis of ESM, we propose a new real-time disk-scheduling algorithm: globally seek-optimizing rescheduling (GSR) scheme. Different from previous approaches, a task in GSR may be rescheduled to anywhere in the input schedule to optimize data throughput. Owing to such a globally rescheduling characteristic, GSR obtains a higher disk throughput than previous approaches. Furthermore, we also extend the GSR to serve fairly non-real-time tasks. Experiments show that given 15 real-time tasks, our data throughput is 1.1 times that of RG-SCAN. In addition, in a mixed workload, compared with RG-SCAN, our GSR achieves over 7% improvement in data throughput and 33% improvement in average response time. (C) 2006 Elsevier Inc. All rights reserved.
Improved methods for characterizing the alternative hypothesis using minimum verification error training for LLR-based speaker verification
(2007)
Chao, Yi-Hsiang
;
Tsai, Wei-Ho
;
Wang, Hsin-Min
;
Chang, Ruei-Chuan
;
資訊工程學系
Speaker verification based on the log-likelihood ratio (LLR) is essentially a task of modeling and testing two hypotheses: the null hypothesis and the alternative hypothesis. Since the alternative hypothesis involves unknown imposters, it is usually hard to characterize a priori. In this paper, we propose a framework to better characterize the alternative hypothesis with the goal of optimally separating client speakers from imposters. The proposed framework is built on either a weighted arithmetic combination or a weighted geometric combination of useful information extracted from a set of pre-trained anti-speaker models. The parameters associated with the combinations are then optimized using Minimum Verification Error training such that both the false acceptance probability and the false rejection probability are minimized. Our experiment results show that the proposed framework outperforms conventional LLR-based approaches.
Improving the characterization of the alternative hypothesis via minimum verification error training with applications to speaker verification
(2009-07-01)
Chao, Yi-Hsiang
;
Tsai, Wei-Ho
;
Wang, Hsin-Min
;
Chang, Ruei-Chuan
;
資訊工程學系
Speaker verification is usually formulated as a statistical hypothesis testing problem and solved by a log-likelihood ratio (LLR) test. A speaker verification system's performance is highly dependent on modeling the target speaker's voice (the null hypothesis) and characterizing non-target speakers' voices (the alternative hypothesis). However, since the alternative hypothesis involves unknown impostors, it is usually difficult to characterize a priori. in this paper, we propose a framework to better characterize the alternative hypothesis with the goal of optimally distinguishing the target speaker from impostors. The proposed framework is built on a weighted arithmetic combination (WAC) or a weighted geometric combination (WGC) of useful information extracted from a set of pre-trained background models. The parameters associated with WAC or WGC are then optimized using two discriminative training methods, namely, the minimum verification error (MVE) training method and the proposed evolutionary, MVE (EMVE) training method, such that both the false acceptance probability and the false rejection probability are minimized. Our experiment results show that the proposed framework Outperforms conventional LLR-based approaches. (C) 2008 Elsevier Ltd. All rights reserved.
nDriver: Online driver replacement for increasing operating system availability
(2007-07-01)
Chang, Da-Wei
;
Huang, Zhi-Yuan
;
Chang, Ruei-Chuan
;
資訊工程學系
;
Department of Computer Science
Device drivers are the most unreliable part of an operating system. In this paper, we propose a framework called nDriver. Based on the design diversity concept, it uses multiple implementations of a device driver to survive from driver faults. Once a fault happens in a driver, nDriver can dynamically replace the faulty driver with another implementation, instead of allowing the faulty driver to crash the system. The unique features of nDriver are as follows. First, it can detect two major kinds of driver faults, the exception and blocking faults. Second, the requests issued to the driver will not be lost due to the driver replacement. Third, the driver replacement is transparent to all the other kernel subsystems. Fourth, nDriver requires no modification to the existing operating system or driver codes. The major contribution of this work is that nDriver implements the concept of design diversity at the device driver layer. Moreover, it achieves the goal of seamless driver replacement and improves operating system availability without modifying the existing operating system or driver codes. We implemented nDriver as a kernel module in Linux. Currently, it can recover the system from faults in network device drivers. However, the mechanisms can be adapted to other module-based device drivers with a slight extension. According to the performance evaluation, the overhead of nDriver is no more than 3.5% and the recovery time is quite small. This indicates that nDriver is an efficient mechanism to increase the availability of an operating system.
Using Kernel Discriminant Analysis to Improve the Characterization of the Alternative Hypothesis for Speaker Verification
(2008-11-01)
Chao, Yi-Hsiang
;
Tsai, Wei-Ho
;
Wang, Hsin-Min
;
Chang, Ruei-Chuan
;
資訊工程學系
Speaker verification can be viewed as a task of modeling and testing two hypotheses: the null hypothesis and the alternative hypothesis. Since the alternative hypothesis involves unknown impostors, it is usually hard to characterize a priori. In this paper, we propose improving the characterization of the alternative hypothesis by designing two decision functions based, respectively, on a weighted arithmetic combination and a weighted geometric combination of discriminative information derived from a set of pretrained background models. The parameters associated with the combinations are then optimized using two kernel discriminant analysis techniques, namely, the kernel Fisher discriminant (KFD) and support vector machine (SVM). The proposed approaches have two advantages over existing methods. The first is that they embed a trainable mechanism in the decision functions. The second is that they convert variable-length utterances into fixed-dimension characteristic vectors, which are easily processed by kernel discriminant analysis. The results of speaker-verification experiments conducted on two speech corpora show that the proposed methods outperform conventional likelihood ratio-based approaches.
Wireless LAN medium access controller supporting mode change and mode change method thereof
(2006-04-27)
Lin, Yen-Liang
;
Chang, Hsung-Pin
;
Chang, Ruei-Chuan
A wireless LAN medium access controller supporting mode change and the mode change method thereof are proposed. A physical layer detects signal strength in the environment and writes a mode index value of the stronger signal into a register. When the written mode index value is different from a mode index value in the communication system, a medium access controller is informed to perform mode change. The physical layer then sends out an interrupt signal to a central processor and sends the mode index value to the central processor for comparison via a DMA module and a memory. An interface module is used to access the mode index value in the memory for providing an external interface for the medium access control layer. When the mode index value doesn't conform to the signal in the environment, a software layer in the medium access controller is used for mode change.