In this paper, we introduced a novel storage architecture 'Unified Storage Network', which merges NAC( Network Attached Channel) and SAN( Storage Area Network) , and provides the file I/O services as NAS devic...In this paper, we introduced a novel storage architecture 'Unified Storage Network', which merges NAC( Network Attached Channel) and SAN( Storage Area Network) , and provides the file I/O services as NAS devices and provides the block I/O services as SAN. To overcome the drawbacks from FC, we employ iSCSI to implement the USN( Unified Storage Network) . To evaluate whether iSCSI is more suitable for implementing the USN, we analyze iSCSI protocol and compare it with FC protocol from several components of a network protocol which impact the performance of the network. From the analysis and comparison, we can conclude that the iSCSI is more suitable for implementing the storage network than the FC under condition of the wide-area network. At last, we designed two groups of experiments carefully.展开更多
The mismatch of ill-phase and quadrature channels in quadrature receiver affects and constrains radar detection performance in coherent radar. It is necessary to keep the in-phase and quadrature branches symmetrical. ...The mismatch of ill-phase and quadrature channels in quadrature receiver affects and constrains radar detection performance in coherent radar. It is necessary to keep the in-phase and quadrature branches symmetrical. In this letter, an adaptive method t.o detect imbalance parameters is derived by means of evaluating channel errors from the received signal .sequences. No matter how the bias degree of the gain and phase errors in I/Q channels are. the proposed adaptive scheme can obtain good calibration results. And the inquired calculations are only a few multiplications and additions. No need of a special test signal, the introduced method is simple to implement and easy to operate.展开更多
In this paper,we introduce the design and implementation of ParaVT,which is a visual performance analysis and parallel debugging tool.In ParaVT,we propose an automated instrumentation mechanism.Based on this mechanism...In this paper,we introduce the design and implementation of ParaVT,which is a visual performance analysis and parallel debugging tool.In ParaVT,we propose an automated instrumentation mechanism.Based on this mechanism,ParaVT automatically analyzes the performance bottleneck of parallel applications and provides a visual user interface to monitor and analyze the performance of parallel programs.In addition,it also supports certain extensions.展开更多
The architecture of a BioAccel (internal code) chip for RNA secondary structure prediction is described in the letter. The system is based on a BioBus (internal code), whose distinguishing features are: Two separated ...The architecture of a BioAccel (internal code) chip for RNA secondary structure prediction is described in the letter. The system is based on a BioBus (internal code), whose distinguishing features are: Two separated control and data channels, and a slave-associated arbitration scheme. Two reference systems based on the AMBA AHB bus and Coreconnect bus are introduced to evaluate the performance of the system. The simulation results are attractive. The average communication bandwidth of the chip is increased at severalfold, and the read and write latencies are reduced about 40 percent.展开更多
A class of modified parallel combined methods of real-time numerical simulation are presented for a stiff dynamic system. By combining the parallelism across the system with the parallelism across the method, and rela...A class of modified parallel combined methods of real-time numerical simulation are presented for a stiff dynamic system. By combining the parallelism across the system with the parallelism across the method, and relaxing the dependence of stage value computation on sampling time of input function, a class of modified real-time parallel combined methods are constructed. Stiff and nonstiff subsystems are solved in parallel on a parallel computer by a parallel Rosen-brock method and a parallel RK method, respectively. Their order conditions and convergences are discussed. The numerical simulation experiments show that this class of modified algorithms can get high speed and efficiency.展开更多
Appropriate schemata as a novel concept to characterize building blocks are introduced, and then, the traits of appropriate schemata are presented. The effects of building blocks by search operators are analyzed. Henc...Appropriate schemata as a novel concept to characterize building blocks are introduced, and then, the traits of appropriate schemata are presented. The effects of building blocks by search operators are analyzed. Hence, the experiments on RR-8X8 are employed to verify that appropriate schemata construct the building blocks. The validity of appropriate schemata and building blocks from the views of theory and practice is presented.展开更多
The Godson project is the first attempt to design high performancegeneral-purpose microprocessors in China. This paper introduces the microarchitecture of theGodson-2 processor which is a 64-bit, 4-issue, out-of-order...The Godson project is the first attempt to design high performancegeneral-purpose microprocessors in China. This paper introduces the microarchitecture of theGodson-2 processor which is a 64-bit, 4-issue, out-of-order execution RISC processor that implementsthe 64-bit MIPS-like instruction set. The adoption of the aggressive out-of-order executiontechniques (such as register mapping, branch prediction, and dynamic scheduling) and cachetechniques (such as non-blocking cache, load speculation, dynamic memory disambiguation) helps theGodson-2 processor to achieve high performance even at not so high frequency. The Godson-2 processorhas been physically implemented on a 6-metal 0.18 μm CMOS technology based on the automaticplacing and routing flow with the help of some crafted library cells and macros. The area of thechip is 6,700 micrometers by 6,200 micrometers and the clock cycle at typical corner is 2.3 ns.展开更多
Domain-specific ontologies are greatly useful in knowledge acquisition,sharing and analysis. In this paper, botany-specific ontology for acquiring and analyzing botanicalknowledge is presented. The ontology is represe...Domain-specific ontologies are greatly useful in knowledge acquisition,sharing and analysis. In this paper, botany-specific ontology for acquiring and analyzing botanicalknowledge is presented. The ontology is represented in a set of well-defined categories, and eachconcept is viewed as an instance of certain category. The authors also introduce botany-specificaxioms, an integral part of the ontology, for checking and reasoning with the acquired knowledge.Consistency, completeness and redundancy of the axioms are discussed.展开更多
The current research progresses and problems of the semantic Web are analyzed in this paper, and the insufficiency of using description logic to act as logical foundation for the semantic Web is analyzed too. Accordin...The current research progresses and problems of the semantic Web are analyzed in this paper, and the insufficiency of using description logic to act as logical foundation for the semantic Web is analyzed too. According to the characteristics and requirement of the semantic Web, a kind of new dynamic description logic (DDL) framework is presented. The representation and reasoning of static knowledge and dynamic knowledge are integrated in this framework. Especially, a kind of action description method is proposed, and according to description logic theory, the action semantics is described, so DDL is a kind of formal logical framework which can process static knowledge and dynamic knowledge. The DDL has clear and formally defined semantics. It provides decidable reasoning services, and it can support effective representation and reasoning of the static knowledge, dynamic process and running mechanism (realization and subsumption relation of action). Therefore, the DDL provides reasonable logic foundation for the semantic Web, and overcomes the insufficiency of using description logic to act as logical foundation for the semantic Web.展开更多
Inherent heterogeneity and distribution of knowledge strongly prevent knowledge from sharing and reusing among different agents and software entities, and a formal ontology has been viewed as a promising means to tack...Inherent heterogeneity and distribution of knowledge strongly prevent knowledge from sharing and reusing among different agents and software entities, and a formal ontology has been viewed as a promising means to tackle this problem. In this paper, a domain-specific formal ontology of archaeology is presented. The ontology mainly consists of three parts: archaeological categories, their relationships and axioms. The ontology not only captures the semantics of archaeological knowledge, but also provides archaeology with an explicit and formal specification of a shared conceptualization, thus making archaeological knowledge shareable and reusable across humans and machines in a structured fashion. Further, we propose a method to verify ontology. correctness based on the individuals of categories. As applications of the ontology,we have developed an ontology-driven approach to knowledge acquisition from archaeological text and a question answering system for archaeological knowledge.展开更多
This paper presents a new technique of unified probabilistic models for facerecognition from only one single example image per person. The unified models, trained on anobtained training set with multiple samples per p...This paper presents a new technique of unified probabilistic models for facerecognition from only one single example image per person. The unified models, trained on anobtained training set with multiple samples per person, are used to recognize facial images fromanother disjoint database with a single sample per person. Variations between facial images aremodeled as two unified probabilistic models: within-class variations and between-class variations.Gaussian Mixture Models are used to approximate the distributions of the two variations and exploita classifier combination method to improve the performance. Extensive experimental results on theORL face database and the authors'' database (the ICT-JDL database) including totally 1,750 facialimages of 350 individuals demonstrate that the proposed technique, compared with traditionaleigenface method and some well-known traditional algorithms, is a significantly more effective androbust approach for face recognition.展开更多
The automatic segmentation of news items is a key for implementing the automatic cataloging system of news video. This paper presents an approach which manages audio and video feature information to automatically segm...The automatic segmentation of news items is a key for implementing the automatic cataloging system of news video. This paper presents an approach which manages audio and video feature information to automatically segment news items. The integration of audio and visual analyses can overcome the weakness of the approach using only image analysis techniques. It makes the approach more adaptable to various situations of news items. The proposed approach detects silence segments in accompanying audio, and integrates them with shot segmentation results, as well as anchor shot detection results, to determine the boundaries among news items. Experimental results show that the integration of audio and video features is an effective approach to solving the problem of automatic segmentation of news items.展开更多
This paper presents a test resource partitioning technique based on anefficient response compaction design called quotient compactor(q-Compactor). Because q-Compactor isa single-output compactor, high compaction ratio...This paper presents a test resource partitioning technique based on anefficient response compaction design called quotient compactor(q-Compactor). Because q-Compactor isa single-output compactor, high compaction ratios can be obtained even for chips with a small numberof outputs. Some theorems for the design of q-Compactor are presented to achieve full diagnosticability, minimize error cancellation and handle unknown bits in the outputs of the circuit undertest (CUT). The q-Compactor can also be moved to the load-board, so as to compact the outputresponse of the CUT even during functional testing. Therefore, the number of tester channelsrequired to test the chip is significantly reduced. The experimental results on the ISCAS ''89benchmark circuits and an MPEG 2 decoder SoC show that the proposed compaction scheme is veryefficient.展开更多
An orthogonal scaling function (?)(t) can realize perfect A/D (Analogue/Digital) and D/A if and only if (?)(t) is cardinal in the case of scalar wavelet. But it is not true when it comes to multiwavelets. Even if a mu...An orthogonal scaling function (?)(t) can realize perfect A/D (Analogue/Digital) and D/A if and only if (?)(t) is cardinal in the case of scalar wavelet. But it is not true when it comes to multiwavelets. Even if a multiscaling function ?(t) is not cardinal, it also holds for perfect A/D and D/A. This property shows the limitation of Selesnick's sampling theorem. In this paper, we present a general sampling theorem for multiwavelet subspaces by Zak transform and make a large family of multiwavelets with some good properties (orthogonality, compact support, symmetry, high approximation order, etc.), but not necessarily with cardinal property, realize perfect A/D and D/A. Moreover, Selesnick's result is just the special case of our theorem. And our theorem is suitable for some symmetrical or nonorthogonal multiwavelets.展开更多
Novelty detection is to retrieve new information and filter redundancy fromgiven sentences that are relevant to a specific topic. In TREC2003, the authors tried an approach tonovelty detection with semantic distance c...Novelty detection is to retrieve new information and filter redundancy fromgiven sentences that are relevant to a specific topic. In TREC2003, the authors tried an approach tonovelty detection with semantic distance computation. The motivation is to expand a sentence byintroducing semantic information. Computation on semantic distance between sentences incorporatesWordNet with statistical information. The novelty detection is treated as a binary classificationproblem: new sentence or not. The feature vector, used in the vector space model for classification,consists of various factors, including the semantic distance from the sentence to the topic and thedistance from the sentence to the previous relevant context occurring before it. New sentences arethen detected with Winnow and support vector machine classifiers, respectively. Several experimentsare conducted to survey the relationship between different factors and performance. It is provedthat semantic computation is promising in novelty detection. The ratio of new sentence size torelevant size is further studied given different relevant document sizes. It is found that the ratioreduced with a certain speed (about 0.86). Then another group of experiments is performedsupervised with the ratio. It is demonstrated that the ratio is helpful to improve the noveltydetection performance.展开更多
At GCC 2003 in Shanghai in December 2003, a panel discussion was held on thefuture of grid computing and on the role of the Globus Toolkit in future grid standards. Panelistsinclude Andrew Chien (UCSD, USA), Wolfgang ...At GCC 2003 in Shanghai in December 2003, a panel discussion was held on thefuture of grid computing and on the role of the Globus Toolkit in future grid standards. Panelistsinclude Andrew Chien (UCSD, USA), Wolfgang Gentzsch (Sun), Francis Lau (HKU, China), Carl Kesselman(USC, USA), Satoshi Matsuoka (TIT, Japan), Xian-He Sun (IIT, USA), Richard Wirt (Intel), Liang-JieZhang (IBM Research), Song-Nian Zhou (Platform Computing), and Zhi-Wei Xu (ICT, China), with Hai Jin(HUST, China) served as the coordinator. The panel talks were stimulating and well received. Threeof the panel talk notes are selected and included in this viewpoint.展开更多
A fault-tolerant and heuristic routing algorithm for faulty hypercube sys-tems is described. To improve the efficiency, the algorithm adopts a heuristic backtracking strategy and each node has an array to record its a...A fault-tolerant and heuristic routing algorithm for faulty hypercube sys-tems is described. To improve the efficiency, the algorithm adopts a heuristic backtracking strategy and each node has an array to record its all neighbors'faulty link information to avoid unnecessary searching for the known faulty links. Furthermore, the faulty link information is dynamically accumulated and the technique of heuristically searching for optimal link is used. The algo rithm routes messages through the minimum feasible path between the sender and receiver if at Ieast one such path ekists, and ta.kes the optimal path with higher probability when faulty links exist in the faulty hypercube.展开更多
Tools for pair-wise bio-sequence alignment have for long played a central role in computation biology. Several algorithms for bio-sequence alignment have been developed. The Smith-Waterman algorithm, based on dynamic ...Tools for pair-wise bio-sequence alignment have for long played a central role in computation biology. Several algorithms for bio-sequence alignment have been developed. The Smith-Waterman algorithm, based on dynamic programming, is considered the most fundamental alignment algorithm in bioinformatics. However the existing parallel Smith-Waterman algorithm needs large memory space, and this disadvantage limits the size of a sequence to be handled. As the data of biological sequences expand rapidly, the memory requirement of the existing parallel Smith- Waterman algorithm has become a critical problem. For solving this problem, we develop a new parallel bio-sequence alignment algorithm, using the strategy of divide and conquer, named PSW-DC algorithm. In our algorithm, first, we partition the query sequence into several subsequences and distribute them to every processor respectively, then compare each subsequence with the whole subject sequence in parallel, using the Smith-Waterman algorithm, and get an interim result, finally obtain the optimal alignment between the query sequence and subject sequence, through the special combination and extension method. Memory space required in our algorithm is reduced significantly in comparison with existing ones. We also develop a key technique of combination and extension, named the C&E method, to manipulate the interim results and obtain the final sequences alignment. We implement the new parallel bio-sequences alignment algorithm, the PSW-DC, in a cluster parallel system.展开更多
Leakage current of CMOS circuit increases dramatically with the technologyscaling down and has become a critical issue of high performance system. Subthreshold, gate andreverse biased junction band-to-band tunneling (...Leakage current of CMOS circuit increases dramatically with the technologyscaling down and has become a critical issue of high performance system. Subthreshold, gate andreverse biased junction band-to-band tunneling (BTBT) leakages are considered three maindeterminants of total leakage current. Up to now, how to accurately estimate leakage current oflarge-scale circuits within endurable time remains unsolved, even though accurate leakage modelshave been widely discussed. In this paper, the authors first dip into the stack effect of CMOStechnology and propose a new simple gate-level leakage current model. Then, a table-lookup basedtotal leakage current simulator is built up according to the model. To validate the simulator,accurate leakage current is simulated at circuit level using popular simulator HSPICE forcomparison. Some further studies such as maximum leakage current estimation, minimum leakage currentgeneration and a high-level average leakage current macromodel are introduced in detail.Experiments on ISCAS85 and ISCAS89 benchmarks demonstrate that the two proposed leakage currentestimation methods are very accurate and efficient.展开更多
It is an effective approach to learn the influence of environmental parameters, such as additive noise and channel distortions, from training data for robust speech recognition. Most of the previous methods are based ...It is an effective approach to learn the influence of environmental parameters, such as additive noise and channel distortions, from training data for robust speech recognition. Most of the previous methods are based on maximum likelihood estimation criterion. However, these methods do not lead to a minimum error rate result. In this paper, a novel discrimina-tive learning method of environmental parameters, which is based on Minimum Classification Error (MCE) criterion, is proposed. In the method, a simple classifier and the Generalized Probabilistic Descent (GPD) algorithm are adopted to iteratively learn the environmental pa-rameters. Consequently, the clean speech features are estimated from the noisy speech features with the estimated environmental parameters, and then the estimations of clean speech features are utilized in the back-end HMM classifier. Experiments show that the best error rate reduction of 32.1% is obtained, tested on a task of 18 isolated confusion Korean words, relative to a conventional HMM system.展开更多
文摘In this paper, we introduced a novel storage architecture 'Unified Storage Network', which merges NAC( Network Attached Channel) and SAN( Storage Area Network) , and provides the file I/O services as NAS devices and provides the block I/O services as SAN. To overcome the drawbacks from FC, we employ iSCSI to implement the USN( Unified Storage Network) . To evaluate whether iSCSI is more suitable for implementing the USN, we analyze iSCSI protocol and compare it with FC protocol from several components of a network protocol which impact the performance of the network. From the analysis and comparison, we can conclude that the iSCSI is more suitable for implementing the storage network than the FC under condition of the wide-area network. At last, we designed two groups of experiments carefully.
文摘The mismatch of ill-phase and quadrature channels in quadrature receiver affects and constrains radar detection performance in coherent radar. It is necessary to keep the in-phase and quadrature branches symmetrical. In this letter, an adaptive method t.o detect imbalance parameters is derived by means of evaluating channel errors from the received signal .sequences. No matter how the bias degree of the gain and phase errors in I/Q channels are. the proposed adaptive scheme can obtain good calibration results. And the inquired calculations are only a few multiplications and additions. No need of a special test signal, the introduced method is simple to implement and easy to operate.
文摘In this paper,we introduce the design and implementation of ParaVT,which is a visual performance analysis and parallel debugging tool.In ParaVT,we propose an automated instrumentation mechanism.Based on this mechanism,ParaVT automatically analyzes the performance bottleneck of parallel applications and provides a visual user interface to monitor and analyze the performance of parallel programs.In addition,it also supports certain extensions.
基金Supported by the National Natrual Science Foundation of China (No.60373044) and Knowl-edge Innovative Project of CAS (No.KSCX2-SW-233).
文摘The architecture of a BioAccel (internal code) chip for RNA secondary structure prediction is described in the letter. The system is based on a BioBus (internal code), whose distinguishing features are: Two separated control and data channels, and a slave-associated arbitration scheme. Two reference systems based on the AMBA AHB bus and Coreconnect bus are introduced to evaluate the performance of the system. The simulation results are attractive. The average communication bandwidth of the chip is increased at severalfold, and the read and write latencies are reduced about 40 percent.
基金This project was supported by the National Natural Science Foundation of China (19871080).
文摘A class of modified parallel combined methods of real-time numerical simulation are presented for a stiff dynamic system. By combining the parallelism across the system with the parallelism across the method, and relaxing the dependence of stage value computation on sampling time of input function, a class of modified real-time parallel combined methods are constructed. Stiff and nonstiff subsystems are solved in parallel on a parallel computer by a parallel Rosen-brock method and a parallel RK method, respectively. Their order conditions and convergences are discussed. The numerical simulation experiments show that this class of modified algorithms can get high speed and efficiency.
文摘Appropriate schemata as a novel concept to characterize building blocks are introduced, and then, the traits of appropriate schemata are presented. The effects of building blocks by search operators are analyzed. Hence, the experiments on RR-8X8 are employed to verify that appropriate schemata construct the building blocks. The validity of appropriate schemata and building blocks from the views of theory and practice is presented.
文摘The Godson project is the first attempt to design high performancegeneral-purpose microprocessors in China. This paper introduces the microarchitecture of theGodson-2 processor which is a 64-bit, 4-issue, out-of-order execution RISC processor that implementsthe 64-bit MIPS-like instruction set. The adoption of the aggressive out-of-order executiontechniques (such as register mapping, branch prediction, and dynamic scheduling) and cachetechniques (such as non-blocking cache, load speculation, dynamic memory disambiguation) helps theGodson-2 processor to achieve high performance even at not so high frequency. The Godson-2 processorhas been physically implemented on a 6-metal 0.18 μm CMOS technology based on the automaticplacing and routing flow with the help of some crafted library cells and macros. The area of thechip is 6,700 micrometers by 6,200 micrometers and the clock cycle at typical corner is 2.3 ns.
文摘Domain-specific ontologies are greatly useful in knowledge acquisition,sharing and analysis. In this paper, botany-specific ontology for acquiring and analyzing botanicalknowledge is presented. The ontology is represented in a set of well-defined categories, and eachconcept is viewed as an instance of certain category. The authors also introduce botany-specificaxioms, an integral part of the ontology, for checking and reasoning with the acquired knowledge.Consistency, completeness and redundancy of the axioms are discussed.
基金This work was supported by the 863 High Tech Programme(Grant No.2001AA113121)the National Natural Science Foundation of China(Grant No.90104021).
文摘The current research progresses and problems of the semantic Web are analyzed in this paper, and the insufficiency of using description logic to act as logical foundation for the semantic Web is analyzed too. According to the characteristics and requirement of the semantic Web, a kind of new dynamic description logic (DDL) framework is presented. The representation and reasoning of static knowledge and dynamic knowledge are integrated in this framework. Especially, a kind of action description method is proposed, and according to description logic theory, the action semantics is described, so DDL is a kind of formal logical framework which can process static knowledge and dynamic knowledge. The DDL has clear and formally defined semantics. It provides decidable reasoning services, and it can support effective representation and reasoning of the static knowledge, dynamic process and running mechanism (realization and subsumption relation of action). Therefore, the DDL provides reasonable logic foundation for the semantic Web, and overcomes the insufficiency of using description logic to act as logical foundation for the semantic Web.
文摘Inherent heterogeneity and distribution of knowledge strongly prevent knowledge from sharing and reusing among different agents and software entities, and a formal ontology has been viewed as a promising means to tackle this problem. In this paper, a domain-specific formal ontology of archaeology is presented. The ontology mainly consists of three parts: archaeological categories, their relationships and axioms. The ontology not only captures the semantics of archaeological knowledge, but also provides archaeology with an explicit and formal specification of a shared conceptualization, thus making archaeological knowledge shareable and reusable across humans and machines in a structured fashion. Further, we propose a method to verify ontology. correctness based on the individuals of categories. As applications of the ontology,we have developed an ontology-driven approach to knowledge acquisition from archaeological text and a question answering system for archaeological knowledge.
文摘This paper presents a new technique of unified probabilistic models for facerecognition from only one single example image per person. The unified models, trained on anobtained training set with multiple samples per person, are used to recognize facial images fromanother disjoint database with a single sample per person. Variations between facial images aremodeled as two unified probabilistic models: within-class variations and between-class variations.Gaussian Mixture Models are used to approximate the distributions of the two variations and exploita classifier combination method to improve the performance. Extensive experimental results on theORL face database and the authors'' database (the ICT-JDL database) including totally 1,750 facialimages of 350 individuals demonstrate that the proposed technique, compared with traditionaleigenface method and some well-known traditional algorithms, is a significantly more effective androbust approach for face recognition.
文摘The automatic segmentation of news items is a key for implementing the automatic cataloging system of news video. This paper presents an approach which manages audio and video feature information to automatically segment news items. The integration of audio and visual analyses can overcome the weakness of the approach using only image analysis techniques. It makes the approach more adaptable to various situations of news items. The proposed approach detects silence segments in accompanying audio, and integrates them with shot segmentation results, as well as anchor shot detection results, to determine the boundaries among news items. Experimental results show that the integration of audio and video features is an effective approach to solving the problem of automatic segmentation of news items.
基金国家自然科学基金，the Sci. & Technol. Project of Beijing，中国科学院资助项目，Synopsys公司资助项目
文摘This paper presents a test resource partitioning technique based on anefficient response compaction design called quotient compactor(q-Compactor). Because q-Compactor isa single-output compactor, high compaction ratios can be obtained even for chips with a small numberof outputs. Some theorems for the design of q-Compactor are presented to achieve full diagnosticability, minimize error cancellation and handle unknown bits in the outputs of the circuit undertest (CUT). The q-Compactor can also be moved to the load-board, so as to compact the outputresponse of the CUT even during functional testing. Therefore, the number of tester channelsrequired to test the chip is significantly reduced. The experimental results on the ISCAS ''89benchmark circuits and an MPEG 2 decoder SoC show that the proposed compaction scheme is veryefficient.
基金This work was supported by the National Natural Science Foundation of China(Grant No. 69875014) and the Foundation of University Key Teacher by the Ministry of Education (Grant No. GG-520-10530-1022).
文摘An orthogonal scaling function (?)(t) can realize perfect A/D (Analogue/Digital) and D/A if and only if (?)(t) is cardinal in the case of scalar wavelet. But it is not true when it comes to multiwavelets. Even if a multiscaling function ?(t) is not cardinal, it also holds for perfect A/D and D/A. This property shows the limitation of Selesnick's sampling theorem. In this paper, we present a general sampling theorem for multiwavelet subspaces by Zak transform and make a large family of multiwavelets with some good properties (orthogonality, compact support, symmetry, high approximation order, etc.), but not necessarily with cardinal property, realize perfect A/D and D/A. Moreover, Selesnick's result is just the special case of our theorem. And our theorem is suitable for some symmetrical or nonorthogonal multiwavelets.
文摘Novelty detection is to retrieve new information and filter redundancy fromgiven sentences that are relevant to a specific topic. In TREC2003, the authors tried an approach tonovelty detection with semantic distance computation. The motivation is to expand a sentence byintroducing semantic information. Computation on semantic distance between sentences incorporatesWordNet with statistical information. The novelty detection is treated as a binary classificationproblem: new sentence or not. The feature vector, used in the vector space model for classification,consists of various factors, including the semantic distance from the sentence to the topic and thedistance from the sentence to the previous relevant context occurring before it. New sentences arethen detected with Winnow and support vector machine classifiers, respectively. Several experimentsare conducted to survey the relationship between different factors and performance. It is provedthat semantic computation is promising in novelty detection. The ratio of new sentence size torelevant size is further studied given different relevant document sizes. It is found that the ratioreduced with a certain speed (about 0.86). Then another group of experiments is performedsupervised with the ratio. It is demonstrated that the ratio is helpful to improve the noveltydetection performance.
文摘At GCC 2003 in Shanghai in December 2003, a panel discussion was held on thefuture of grid computing and on the role of the Globus Toolkit in future grid standards. Panelistsinclude Andrew Chien (UCSD, USA), Wolfgang Gentzsch (Sun), Francis Lau (HKU, China), Carl Kesselman(USC, USA), Satoshi Matsuoka (TIT, Japan), Xian-He Sun (IIT, USA), Richard Wirt (Intel), Liang-JieZhang (IBM Research), Song-Nian Zhou (Platform Computing), and Zhi-Wei Xu (ICT, China), with Hai Jin(HUST, China) served as the coordinator. The panel talks were stimulating and well received. Threeof the panel talk notes are selected and included in this viewpoint.
文摘A fault-tolerant and heuristic routing algorithm for faulty hypercube sys-tems is described. To improve the efficiency, the algorithm adopts a heuristic backtracking strategy and each node has an array to record its all neighbors'faulty link information to avoid unnecessary searching for the known faulty links. Furthermore, the faulty link information is dynamically accumulated and the technique of heuristically searching for optimal link is used. The algo rithm routes messages through the minimum feasible path between the sender and receiver if at Ieast one such path ekists, and ta.kes the optimal path with higher probability when faulty links exist in the faulty hypercube.
文摘Tools for pair-wise bio-sequence alignment have for long played a central role in computation biology. Several algorithms for bio-sequence alignment have been developed. The Smith-Waterman algorithm, based on dynamic programming, is considered the most fundamental alignment algorithm in bioinformatics. However the existing parallel Smith-Waterman algorithm needs large memory space, and this disadvantage limits the size of a sequence to be handled. As the data of biological sequences expand rapidly, the memory requirement of the existing parallel Smith- Waterman algorithm has become a critical problem. For solving this problem, we develop a new parallel bio-sequence alignment algorithm, using the strategy of divide and conquer, named PSW-DC algorithm. In our algorithm, first, we partition the query sequence into several subsequences and distribute them to every processor respectively, then compare each subsequence with the whole subject sequence in parallel, using the Smith-Waterman algorithm, and get an interim result, finally obtain the optimal alignment between the query sequence and subject sequence, through the special combination and extension method. Memory space required in our algorithm is reduced significantly in comparison with existing ones. We also develop a key technique of combination and extension, named the C&E method, to manipulate the interim results and obtain the final sequences alignment. We implement the new parallel bio-sequences alignment algorithm, the PSW-DC, in a cluster parallel system.
文摘Leakage current of CMOS circuit increases dramatically with the technologyscaling down and has become a critical issue of high performance system. Subthreshold, gate andreverse biased junction band-to-band tunneling (BTBT) leakages are considered three maindeterminants of total leakage current. Up to now, how to accurately estimate leakage current oflarge-scale circuits within endurable time remains unsolved, even though accurate leakage modelshave been widely discussed. In this paper, the authors first dip into the stack effect of CMOStechnology and propose a new simple gate-level leakage current model. Then, a table-lookup basedtotal leakage current simulator is built up according to the model. To validate the simulator,accurate leakage current is simulated at circuit level using popular simulator HSPICE forcomparison. Some further studies such as maximum leakage current estimation, minimum leakage currentgeneration and a high-level average leakage current macromodel are introduced in detail.Experiments on ISCAS85 and ISCAS89 benchmarks demonstrate that the two proposed leakage currentestimation methods are very accurate and efficient.
基金the '863' High-Tech Programme of China (No. 863-306ZT03-02-3) and partially by the National Natural Science Foundation of China
文摘It is an effective approach to learn the influence of environmental parameters, such as additive noise and channel distortions, from training data for robust speech recognition. Most of the previous methods are based on maximum likelihood estimation criterion. However, these methods do not lead to a minimum error rate result. In this paper, a novel discrimina-tive learning method of environmental parameters, which is based on Minimum Classification Error (MCE) criterion, is proposed. In the method, a simple classifier and the Generalized Probabilistic Descent (GPD) algorithm are adopted to iteratively learn the environmental pa-rameters. Consequently, the clean speech features are estimated from the noisy speech features with the estimated environmental parameters, and then the estimations of clean speech features are utilized in the back-end HMM classifier. Experiments show that the best error rate reduction of 32.1% is obtained, tested on a task of 18 isolated confusion Korean words, relative to a conventional HMM system.
