The single-layer backpropagation algorithm is a gradient-descent method that adjusts the connection weights of a single-layer perceptron to minimize the mean-square error at the output. It is similar to the standard l...
详细信息
The single-layer backpropagation algorithm is a gradient-descent method that adjusts the connection weights of a single-layer perceptron to minimize the mean-square error at the output. It is similar to the standard least mean square algorithm, except the output of the linear combiner contains a differentiable nonlinearity. In this paper, we present a statistical analysis of the mean weight behavior of the single-layer backpropagation algorithm for Gaussian input signals. It is based on a nonlinear system identification model of the desired response which is capable of generating an arbitrary hyperplane decision boundary. It is demonstrated that, although the weights grow unbounded, the algorithm, on average, quickly learns the correct hyperplane associated with the system identification model.
The authors present a computer-aided designalgorithm for the analysis and design of an evanescent-mode bandpass filter with nontouching E-plane fins. The theoretical analysis is based on the generalized scattering ma...
详细信息
The authors present a computer-aided designalgorithm for the analysis and design of an evanescent-mode bandpass filter with nontouching E-plane fins. The theoretical analysis is based on the generalized scattering matrix technique in conjunction with the spectral-domain approach and mode-matching method. The technique used takes into account the dominant as well as the higher-order effects. The measured filter responses in the Ka-band are in good agreement with those obtained by this analysis.< >
This paper describes an algorithmic development for the design of stable infinite impulse response (IIR) digital filters with equiripple passbands and peak-constrained least-squares stopbands. Central to the developme...
详细信息
This paper describes an algorithmic development for the design of stable infinite impulse response (IIR) digital filters with equiripple passbands and peak-constrained least-squares stopbands. Central to the development is a re-formulation of the design problem as an iterative quadratic programming problem where the stability and equiripple passbands and peak-constrained requirements are met by imposing a system of linear inequality constraints while the LSS property is satisfied by minimizing a weighted least-squares type objective function.
Automated vehicle classification based on static images is highly practical and directly applicable for various operations such as traffic related investigations. An integrated vehicle detection and classification sys...
详细信息
Automated vehicle classification based on static images is highly practical and directly applicable for various operations such as traffic related investigations. An integrated vehicle detection and classification system is proposed in this paper. A multi-resolution vehicle detection scheme is introduced as an improvement over the cascade boosted classifiers proposed recently by Negri et al. 2008 in the literature. Building on solutions from previous works from Negri et al, the implementation of a new decision strategy renders current detection method to be robust to environmental changes. The vehicle classification is based on the Classified Vector Quantization (CVQ) proposed earlier by Zhang et al. 2009. The justification of choosing CVQ is its advantages in providing classification confidence by incorporating rejection option. The significance of rejection in enhancing the system's reliability is emphasized and evaluated. A database composed of more than 2800 images of four types of vehicles (cars, vans, light trucks and buses) was created using police surveillance cameras. The proposed scheme offers a performance accuracy of over 95% with a rejection rate of 8%, and reliability over 98% with a rejection rate of 20%. This exhibits promising potentials for implementations into real-world applications.
This paper presents the peak-constrained least-squares (PCLS) approach to designing IIR digital filters, PCLS IIR digital filters that meet simultaneous specifications on the frequency response magnitude and the group...
详细信息
This paper presents the peak-constrained least-squares (PCLS) approach to designing IIR digital filters, PCLS IIR digital filters that meet simultaneous specifications on the frequency response magnitude and the group delay are introduced, As a point of reference, we consider the IIR digital filter design problem that appears in Deczky's classic paper and in the popular textbook by Oppenheim and Schafer, In addition, the same design problem appears in the IIR filter design chapter by Higgins and Munson in the recent Handbook for Digital Signal Processing, By using our new algorithm with simultaneous optimization of the frequency response magnitude and the group delay, we obtain a dramatic improvement in the solution of this classic IIR digital filter design problem. Starting from the same filter structure and the same specifications for the frequency response magnitude as in the works of Deczky, Oppenheim and Schafer, and Higgins and Munson, we are able to reduce the group delay ripple by a factor of 35, In another design problem that originated in Deczky's work, we use PCLS optimization to reduce the group delay ripple by a factor of 40 at the same time we reduce the stopband energy by 6 dB, without sacrificing any other performance measure, The group delay ripple in this IIR digital filter example is reduced to only +/-0.002 samples.
A protocol design system is a collection of software tools for assisting protocol designers to specify, validate, and implement communication protocols. In this paper, we propose a computer-aided protocol design syste...
详细信息
A protocol design system is a collection of software tools for assisting protocol designers to specify, validate, and implement communication protocols. In this paper, we propose a computer-aided protocol design system based on the OPS5 production system approach. In reality, communication protocols are rule-based and data-driven without a fixed order in which the submodules can follow and computations in communication protocols are mainly symbolic with a few numerical computations. These characteristics fall into the applicable problem domain of OPS5. Using the OPS5 production system approach, communication rules (state transitions) are specified as triples of "object-attribute-value" and the modeling of state transitions are specified by production rules. For protocol validation, the modeling of global states, global state transitions, all logical errors and logical properties can be formally defined in terms fo production rules. This paper also presents an incremental validation algorithm to facilitate protocol design. Based on a globally shared dataspace (working memory) in which different types and levels of information are all represented in a uniform structure (element), the OPS5 production systeme integrates both rule-based and procedure-based computations. Using this characteristic, the machine-dependent part can be abstractly specified through through external procedure calls, the details of which can be coded in a procedure language until the implementation phase. Since computations in OPS5 are based on pattern matching, all of the attributes of elements that are in production rules or in external procedure calls can act as data templates for generic data types. This capability enhances the generic specification that allows different realization for various implementation environments. In this way, our computer-aided protocol design system can be used not only as a rapid prototyping tool for simulation but also as a real implementation tool for communicat
The MediaEval Multimedia Benchmark leveraged community cooperation and crowdsourcing to develop a large Internet video dataset for its Genre Tagging and Rich Speech Retrieval tasks.
The MediaEval Multimedia Benchmark leveraged community cooperation and crowdsourcing to develop a large Internet video dataset for its Genre Tagging and Rich Speech Retrieval tasks.
In this paper, a noisy slotted channel is considered. It is assumed that channel feedback might be misinterpreted due to the existence of noise on the channel. Furthermore, this disturbance is dependent on the channel...
详细信息
In this paper, a noisy slotted channel is considered. It is assumed that channel feedback might be misinterpreted due to the existence of noise on the channel. Furthermore, this disturbance is dependent on the channel state (either good or bad) which varies from slot to slot according to a Markov chain. Consequently, the occurrence of the a channel feedback error is dependent on previous occurrences of errors (i.e., with error memory). Under this assumption, the throughput performance of a random multiple-access algorithm, called the Two-Cell algorithm, is analyzed and the results are compared with the throughputs of the Capetanakis tree-splitting algorithm operating over the same channels. It is shown that the Two-Cell algorithm retains positive throughputs for all possible values of channel state parameters, and for all practical purposes, it outperforms the Capetanakis algorithm in terms of insensitivity to channel feedback errors.
A new algorithm for discrete tracking of slow frequency hopping spread-spectrum (FHSS) signals is described and analyzed. In this system, each Mth hopping interval is completely used for transmission of synchronizatio...
详细信息
A new algorithm for discrete tracking of slow frequency hopping spread-spectrum (FHSS) signals is described and analyzed. In this system, each Mth hopping interval is completely used for transmission of synchronization data. The motivation for introducing this algorithm is that it allows a large degree of flexibility with respect to trading off system performance, complexity and the redundancy introduced for tracking purposes. The analysis of the system performance is based on the utilization of results from discrete renewal process theory. However, because of the difficulty in obtaining the exact values for the transition probabilities, approximate results are presented.
暂无评论