The traditional framework for back-end comp.lation is based on the scope of functions, which is a convenient way to partition an entire application program. However, the sizes and structures of functions may not be th...
详细信息
The traditional framework for back-end comp.lation is based on the scope of functions, which is a convenient way to partition an entire application program. However, the sizes and structures of functions may not be the best scope for program analysis and transformations when considering comp.lation resources (such as time and memory usage), performance, and functionality. This problem is particularly critical when the modern comp.lers resort to sophisticated and expensive algorithms to exploit inst.uction-level parallelism. When a function has a large size, an expensive algorithm based on the function scope often makes the usage of comp.lation time and space usage unacceptable. Some earlier work proposed a comp.lation framework to allow some control on the size and structure of an optimization scope. This paper proposes a new region-based comp.lation framework driven by the considerations of comp.lation resources and performance opportunities. And some optimization-directed attributes can be passed and observed from optimization phase to phase on a region basis. This region-based framework is being implemented in a comp.ler ORC (Open Research comp.ler) targeting at the Itanium Processor Family (IPF).
One possible approach to web personalization is to mine typical user profiles from the vast amount of historical data stored in access logs. Clustering techniques have been used to automatically discover typical user ...
详细信息
ISBN:
(纸本)1932415076
One possible approach to web personalization is to mine typical user profiles from the vast amount of historical data stored in access logs. Clustering techniques have been used to automatically discover typical user profiles recently. But it is a challenging problem to design effective similarity measure between the session vectors which are usually high dimensional and sparse. A new approach based on non-negative matrix factorization (NMF) is presented. We apply non-negative matrix factorization to dimensionality reduction of the session-URL matrix, and the projecting vectors of the user session vectors are clustered into typical user session profiles using the spherical k-means algorithm. The results of experiment show that our algorithm can mine interesting user profiles effectively.
Cluster of high performance PC is a popular comp.ting platform today. Unfortunately, traditional network interface architecture introduces so much software overhead that such clusters are unsuitable for shared virtual...
详细信息
Cluster of high performance PC is a popular comp.ting platform today. Unfortunately, traditional network interface architecture introduces so much software overhead that such clusters are unsuitable for shared virtual memory parallel applications which need short latency, high bandwidth and quick synchronization. Virtual Interface Architecture (VIA) specification gives a new approach to solve this problem for its low software overhead, by passing OS intervention, high overlapping between communication and comp.tation and zero copy protocol. By evaluating the performance of JIAJIA software DSM system over VIA and traditional network interface, it is shown that VIA can improve the performance and scalability of software DSMs. The specification also give us new ideas to design hardware support for SVM systems.
Tow-color reflection hypothesis were utilized to derive solid subject surface diffusion reflection vector and mirror reflection vector in the condition that illuminative light color is unknown by dominant comp.sition ...
详细信息
Tow-color reflection hypothesis were utilized to derive solid subject surface diffusion reflection vector and mirror reflection vector in the condition that illuminative light color is unknown by dominant comp.sition analysis method according to the distribution of subject color histogram. Based on this, the reflection characteristic vector of the subject surface was estimated, and the validity of this method was proved by the experiment.
As a formal tool for knowledge representation and reasoning, description logic provides the decidable reasoning service. This paper proposes a new description logic with default reasoning, which integrates monotonic r...
详细信息
As a formal tool for knowledge representation and reasoning, description logic provides the decidable reasoning service. This paper proposes a new description logic with default reasoning, which integrates monotonic reasoning and non-monotonic reasoning while avoiding the difficulty of Reiter's default logic. Based on the description logic with default reasoning, we construct a knowledge base system that incorporates Tbox, Abox and default rules. Then we discuss several reasoning problems including concept satisfiability, default satisfiability, concept subsumption, default subsumption, and inst.nce checking. Furthermore, we present a new algorithm called Tableau-D that is used to check the satisfiability and default satisfiability. Finally a transformation theorem of default satisfiability and default subsumption is also attained.
The technologies of information warfare encomp.ss intelligence, surveillance, and reconnaissance (ISR);command, control, communications, coordination, intelligence, interoperability and precision force application. Of...
详细信息
The technologies of information warfare encomp.ss intelligence, surveillance, and reconnaissance (ISR);command, control, communications, coordination, intelligence, interoperability and precision force application. Of these, the ISR technologies are the most visible in the current combat zone. A brief survey of the status and the general direction of these technologies as pursued by the major military powers of the world has been attempted in the paper. The gap between the developed and the developing nations is quite evident. Nevertheless, all the major developing military powers are keenly pursuing the path of modernization in this field. However, the economical and technological infrastructures of these nations are nowhere near the requisite standards in this field. At the same time, the teeming millions of these countries are desperately vying for a better standard of life. Economic strength and happy population form important ingredients of the national security in modern times. Herein lies the dilemma of the developing nations. A viable model has been suggested for the developing nations to attain a minimum level of hi-tech military strength in this field in a cost-effective manner so that other aspects of national security are not ignored.
With great demand for digital signal transmission and big losses incurred from illegal data access, signal security has become an extremely critical issue. In this paper, a new signal security system, along with its V...
详细信息
With great demand for digital signal transmission and big losses incurred from illegal data access, signal security has become an extremely critical issue. In this paper, a new signal security system, along with its VLSI architecture, is proposed. Based on the simple idea of overturning dominoes, each signal value is randomly transformed according to a binary sequence generated from a chaotic system, and therefore, a comp.etely disordered signal can be obtained. The features of the system are as follows: 1) no distortion, 2) high security, and 3) suitability for system integration. In order to implement the system, its VLSI architecture is designed with low hardware comp.exity and high comp.ting speed. Moreover, the FPGA realization results indicate that the data processing rate of the proposed design achieves 800 Mbps, fast enough for MPEG2 in HDTV applications. Finally, the simulation and comp.rison results are given to demonstrate the proposed system's effectiveness.
Shared-Memory parallel program is difficult to debug because of data races. Most of the previous work in dynamic data race detection is based on Happen-Before Order. The tools based on Happen-before Order have a signi...
详细信息
Shared-Memory parallel program is difficult to debug because of data races. Most of the previous work in dynamic data race detection is based on Happen-Before Order. The tools based on Happen-before Order have a significant drawback that for an input set, a detection of the execution of a program can miss some feasible data races. This article describes a lockset-based dynamic data race detection technique to solve this problem. Our dynamic data race detection is targeted on a software distributed shared memory (DSM) system called JIAJIA which supports Scope Consistency. By using the frame model of memory consistency model, Enhanced-Happen-Before Order for Scope Consistency has been defined and in further step, a lockset-based dynamic data race algorithm is suggested. This article describes the implementation of lockset-based algorithm in JIAJIA and presents an experimental evaluation of lockset-based technique by looking for data races in five common shared-memory programs. In TSP program, some read-write data races have been correctly found. The slowdown of this lockset-based technique averages less that 3.14 for our applications, which means that our lockset-based technique has gotten good performance.
We solved the liveness check problem using a Support Vector Machine based eye movement model. We trained the eye movement model using many eye state examples. And in the liveness check procedure, the user has to blink...
详细信息
We solved the liveness check problem using a Support Vector Machine based eye movement model. We trained the eye movement model using many eye state examples. And in the liveness check procedure, the user has to blink his/her eyes to get through the check. The photos, which can not blink, are successfully stopped before entering the recognition stage.
We present a learning algorithm for incremental support vector machine (SVM). We analysis the possible changes of support vector set after new samples are added to training set. Based on the analysis result, an algori...
详细信息
We present a learning algorithm for incremental support vector machine (SVM). We analysis the possible changes of support vector set after new samples are added to training set. Based on the analysis result, an algorithm is presented. This algorithm is applied for the image recognition of two special animal fibers. With the algorithm, the useless sample is discarded and knowledge is accumulated, The experiment result shows that this algorithm is more effective than the traditional SVM while the classification precision is guaranteed.
暂无评论