The hierarchical network proposed (Multi-RBFNN), is composed of complete Radial Basis Function Neural Networks (RBFNNs) that are in charge of a reduced set of input variables with the property of which every Sub-RBFNN...
详细信息
The hierarchical network proposed (Multi-RBFNN), is composed of complete Radial Basis Function Neural Networks (RBFNNs) that are in charge of a reduced set of input variables with the property of which every Sub-RBFNN can take charge of a set of input variables and not of all. For the optimization of the whole net, we propose a new method to select the more important input variables, which is capable of deciding which of the chosen variables go alone or together to a Sub-RBFNN to build the hierarchic structure Multi-RBFNN, thus reducing the dimension of the input variable space for each RBFNN. We also provide an algorithm which automatically finds the most suitable topology of the proposed hierarchical structure and selects the more important input variables for it.
It is well known the relationship between source separation and blind deconvolution: If a filtered version of an unknown i.i.d. signal is observed, temporal independence between samples can be used to retrieve the ori...
详细信息
Transitivity is an important property used in logical reasoning. However, Part-Whole (PW) relations usually do not satisfy the property. Wiston and Odell partitioned PW relations into six sub-PW relations respectively...
详细信息
Transitivity is an important property used in logical reasoning. However, Part-Whole (PW) relations usually do not satisfy the property. Wiston and Odell partitioned PW relations into six sub-PW relations respectively, such that the transitivity property is held in each sub-class and each different sub-PW denotes a kind of PW relations with special meanings. This paper studies the composition of different sub-PW relations given by Wiston and Odell in following three aspects: (I) Does the composition relation of two different sub- PW relations make sense?(2) Is the transitivity still held for the composition relations of two different sub- PW relations; (3) which sub-PW class does the composition relation belong to. Different from former study, a formal discussion is carried for above issues.
Monte Carlo Go is a promising method to improve the performance of computer Go programs. This approach determines the next move to play based on many Monte Carlo samples. This paper examines the relative advantages of...
详细信息
Recent parallel systems use multiple injection ports and various injection policies, but little is known about their impact on network performance. This paper evaluates the influence that these injection interfaces ha...
详细信息
Recent parallel systems use multiple injection ports and various injection policies, but little is known about their impact on network performance. This paper evaluates the influence that these injection interfaces have on maximum sustained throughput in adaptive cut-through torus networks by modeling the number of injection queues (1 or 4), and the allocation of new packets to those queues. Network evaluations for medium to large size 2D tori show that designs with multiple injection ports do not improve performance under uniform traffic. On the contrary, they result in more pressure from the injection interface to acquire the scarce network resources of an already clogged system. Interestingly, for small networks, a single injection FIFO queue, with the HOLE it entails, indirectly provides the much needed injection control. For networks with thousands of nodes and multiple injection channels, as those being implemented in current massively parallel processors, this implicit form of congestion control is not enough. In such systems, restrictive injection policies are required to prevent routers from being flooded with new packets for loads beyond saturation.
作者：
Prof. Jian-Xin XuProf. Leonid FridmanDepartment of Electrical and Computer Eng. National University of Singapore 4 Engineering Drive 3 Singapore 117576 Tel +65 6874-2566
Fax +65 6779-1103 Dr Jian-Xin Xu received his Bachelor degree from Zhejiang University
China in 1982. He attended the University of Tokyo Japan where he received his Master's and Ph.D. degrees in 1986 and 1989 respectively. All his degrees are in Electrical Engineering. He worked for one year in the Hitachi research Laboratory Japan and for more than one year in Ohio State University U.S.A. as a Visiting Scholar. In 1991 he joined the National University of Singapore and is currently an associate professor in the Department of Electrical Engineering. His research interests lie in the fields of learning control variable structure control fuzzy logic control discontinuous signal processing and applications to motion control and process control problems. He is the associate editor of Asian Journal of Control member of TC on variable structure systems and sliding mode control of IEEE Control Systems Society and a senior member of IEEE. He has produced more than 90 peer-refereed journal papers near 160 technical papers in conference proceedings and authored/edited 4 books. Division de Estudios de Posgrado Facultad de Ingenieria National Autonomous University of Mexico DEP-FI
UNAM Edificio “A” Circuito Exterior Ciudad Universitaria A. P. 70–256 C.P.04510 Mexico D.F. Mexico Tel +52 55 56223014 Fax +52 55 56161719 Dr. Leonid M. Fridman received his M.S in mathematics from Kuibyshev (Samara) State University
Russia Ph.D. in Applied Mathematics from Institute of Control Science (Moscow) and Dr. of Science degrees in Control Science from Moscow State University of Mathematics and Electronics in 1976 1988 and 1998 respectively. In 1976–1999 Dr. Fridman was with the Department of Mathematics at the Samara State Architecture and Civil Engineering Academy Samara Russia. In 2000–2002 he was with the Department of Postgraduate Study and Investigations at the Chihuahu
Nature has optimized the processing of visual information, especially in primates. The estimation of optical flow is a complex task that gives information about ego-motion, and permits tracking objects from a given sc...
详细信息
Nature has optimized the processing of visual information, especially in primates. The estimation of optical flow is a complex task that gives information about ego-motion, and permits tracking objects from a given scene. The multi-channel spatio-temporal filtering required to detect motion is suitable for a parallel implementation on reconfigurable circuitry. We detail here the design of a neuromorphic FPGA implementation of the pre-processing stages for optical flow estimation that permits highly parallel real-time filtering
In this paper we present how face tracking can be implemented on mobile devices. Our main contribution is to present how face tracking on mobile systems can be used as a multi-dimensional input technique and to demons...
详细信息
ISBN:
(纸本)1595932984
In this paper we present how face tracking can be implemented on mobile devices. Our main contribution is to present how face tracking on mobile systems can be used as a multi-dimensional input technique and to demonstrate how this can be used in different mobile applications. We present at set of different applications based on the tracking, and discuss current and future advantages, challenges and problems with face tracking as input device for mobile systems.
Performance uncertainty is a barrier to implementation of innovative technologies. This research investigates the potential of flexible design - one that enables future change - to improve the economic performance of ...
详细信息
ISBN:
(纸本)089553178X
Performance uncertainty is a barrier to implementation of innovative technologies. This research investigates the potential of flexible design - one that enables future change - to improve the economic performance of a naturally ventilated building. The flexible design of the naturally ventilated building enables future installation of a mechanical cooling system by including features such as space for pipes and chillers. The benefits of the flexible design are energy savings, delay of capital costs and capability of mitigating the risk of a failed building (by installing the mechanical cooling system). To evaluate the flexible design, building energy simulation is conducted over a multi-year time period with stochastic outdoor temperature variables. One result is a probability distribution of the time when the maximum allowable indoor temperature under natural ventilation is exceeded, which may be "never." Probability distributions are also obtained for energy savings and cost savings as compared to a mechanically cooled building. Together, these results allow decision-makers to evaluate the long-term performance risks and opportunities afforded by a flexible implementation strategy for natural ventilation. It is shown that the likelihood of future installation of mechanical cooling is most sensitive to design parameters. The impact of increased climate variability depends on the local climate. The probability of installing the mechanical system also depends on the comfort criteria. The results show that capital costs for cooling equipment are much greater than the present value of 10 years of cooling energy costs. This result motivates consideration of flexible design as opposed to hybrid cooling designs (which have immediate installation of mechanical cooling). Future work will study the impact of uncertain energy prices on investment attractiveness of naturally ventilated buildings. Other applications of the framework presented herein include replacing the build
We present an algorithm implementing the failure detector class omega (ω) in the crash-recovery model to coordinate data aggregation in wireless sensor networks. The algorithm ensures the agreement on a common aggreg...
详细信息
ISBN:
(纸本)1595934871
We present an algorithm implementing the failure detector class omega (ω) in the crash-recovery model to coordinate data aggregation in wireless sensor networks. The algorithm ensures the agreement on a common aggregator by all sensors of a region, as well as on a common super-aggregator among the set of aggregators of the network, hence providing a hierarchical energy-efficient data aggregation mechanism. We also introduce a battery depletion threshold to enhance the quality of service of the wireless sensor network. Copyright 2006 ACM.
暂无评论