In this paper, we consider a distributed resource allocation problem with communication limitation. We propose a gradient-descent algorithm to solve the distributed resource allocation problem with quantization mechan...
详细信息
In this paper, we consider a distributed resource allocation problem with communication limitation. We propose a gradient-descent algorithm to solve the distributed resource allocation problem with quantization mechanism due to the communication limitations or in order to reduce the communication cost in the network. With carefully selected parameters, the convergence and correctness of the quantized algorithm can be obtained for fixed communication topologies and moreover, extended to switching jointly connected topologies. The exact optimal value can be obtained, which is different from some existing works, whose convergence accuracies were limited by the quantization bandwidth. In addition, the data rate of the proposed algorithm is also analyzed for both fixed and switching communication networks. Particularly, in the fixed topology case, we obtain that 1 bit is sufficient to solve the distributed problem with the proposed algorithm.
An improved pattern recovery approach that integrates the Hopfield neural network (HNN) with the iterative signal reconstruction and Parzen window-based classification is proposed. The HNN is observed as a form of ass...
详细信息
An improved pattern recovery approach that integrates the Hopfield neural network (HNN) with the iterative signal reconstruction and Parzen window-based classification is proposed. The HNN is observed as a form of associative memory network, used for various pattern recognition and optimization tasks. However, when the input pattern is highly damaged with a very limited set of available samples, the Hopfield network fails to perform the retrieval. The convex optimization-based gradientdescentalgorithm is considered for pattern recovery of damaged inputs in order to provide an improved pattern approximation for further processing within the HNN, enabling successful network performance. Additionally, in the case of grayscale images, the Parzen window approach is used to classify the probability density functions (pdfs) of the training set and to choose those being comparable to the pdf of the input pattern, therefore refining the selection of patterns and providing better convergence to the exact retrieval. The theoretical considerations are verified experimentally, showing the high performance of the proposed approach when only 10 % of the pixels are available for binary patterns and 40 % of pixels for grayscale patterns.
Since a conceptor, achieving direction-selective damping of high-dimensional network signals, usually takes the form of a projection matrix and is deduced analytically, a conceptor-based neural network is thought to b...
详细信息
Since a conceptor, achieving direction-selective damping of high-dimensional network signals, usually takes the form of a projection matrix and is deduced analytically, a conceptor-based neural network is thought to be untrainable with backpropagation and gradient-descent algorithms from end to end. It limits the application of conceptors. To address this issue, an algorithm is proposed to train conceptor-based neural networks from end to end with gradient-descent algorithms. To the best of the authors' knowledge, it is the first work of such an end-to-end training algorithm. To develop this algorithm, a softmax-like loss function involved with conceptors is constructed empirically. Based on this loss function, corresponding gradients are deduced by using backpropagation method so that it is possible to train conceptor neural networks from end to end with a gradient-descent algorithm. Several experiments are conducted to show the feasibility and effectiveness of the proposed training algorithm.
Noise originating from several sources in a RF environment degrades the performance of communication systems. In wideband systems, such as cognitive radios, noise at the receiver can also originate from non-linearity ...
详细信息
ISBN:
(纸本)9781509014965
Noise originating from several sources in a RF environment degrades the performance of communication systems. In wideband systems, such as cognitive radios, noise at the receiver can also originate from non-linearity present in the RF front end, time-varying thermal noise within the receiver radio system, and noise from adjacent network nodes. Several denoising techniques have been proposed for cognitive radios, some of which are applied during spectrum sensing and others to received noisy signals. Examples of these techniques include least mean square (LMS) and its variants. However, these algorithms have low performance with non-linear signals and cannot locate the global optimum solution for noise cancellation. Therefore, application of global search optimization techniques, such as evolutionary algorithms, is considered for noise cancellation. In this paper, particle swarm optimization (PSO) and LMS algorithms are implemented and their performances compared. Extensive simulations were performed where Gaussian and nonlinear random noise were added to the transmitted signals. The performance comparison was done using two metrics: bit error rate and mean square error. The results show that PSO outperforms LMS under both Gaussian and non-linear random noise.
On-chip Wavelength Division Multiplexing (WDM) devices have been widely used in optical communications and signal processing to increase data throughput. As the number of wavelength channels increases, the performance...
详细信息
ISBN:
(纸本)9781665481557
On-chip Wavelength Division Multiplexing (WDM) devices have been widely used in optical communications and signal processing to increase data throughput. As the number of wavelength channels increases, the performance of WDM devices is susceptible to various types of errors from fabrication, design and environmental changes. To address this issue, we demonstrate an automated optimization control method based on the gradient-descent algorithm and examine the performance of applying this algorithm to a silicon photonic 16-channel WDM device utilizing 4-level cascaded tunable MachZehnder interferometers. The extinction ratio and crosstalk of all 16 channels are successfully optimized with different initial temperatures in simulation. This procedure provides a general approach to automated control and optimize complex photonic systems.
This paper attempts to solve the optimal power allocation (OPA) problem for smart grid system in a new distinguished way. Conventionally the numerical optimization approaches including traditional convex optimization ...
详细信息
ISBN:
(纸本)9781728158556
This paper attempts to solve the optimal power allocation (OPA) problem for smart grid system in a new distinguished way. Conventionally the numerical optimization approaches including traditional convex optimization and heuristic search methods almost occupy the addressing of such problem. However, these optimization algorithms may suffer from high computational complexity when the system scales up, which would inevitably create a gap between the theoretical algorithm design and real-time algorithm implementation. In this paper, we aim to provide a new learning based approach to handle the real-time OPA problem in smart grid system. The key idea behind this approach is to treat the input and output of traditional OPA optimization algorithm as an unknown nonlinear mapping, which is then approximated by recent popular learning based tools such as deep neural network (DNN). As long as the constructed DNN can accurately learn such nonlinear mapping, then the OPA problem can be solved in real time. Our main contribution is to theoretically show that the traditional decentralized gradient-based optimization algorithm for OPA problem can be accurately approximated by a well-constructed DNN. Furthermore, experimental case studies validate the effectiveness and advantages of our proposed method.
Many real life problems require the classification of items into naturally ordered classes. These problems are traditionally handled by conventional methods intended for the classification of nominal classes, where th...
详细信息
ISBN:
(纸本)9783642289309;9783642289316
Many real life problems require the classification of items into naturally ordered classes. These problems are traditionally handled by conventional methods intended for the classification of nominal classes, where the order relation is ignored. This paper proposes a hybrid neural network model applied to ordinal classification using a possible combination of projection functions (product unit, PU) and kernel functions (radial basis function, RBF) in the hidden layer of a feed-forward neural network. A combination of an evolutionary and a gradient-descent algorithms is adapted to this model and applied to obtain an optimal architecture, weights and node typology of the model. This combined basis function model is compared to the corresponding pure models: PU neural network, and the RBF neural network. Combined functions using projection and kernel functions are found to be better than pure basis functions for the task of ordinal classification in several datasets.
We propose a block equalization algorithm using optimal step size. The algorithm shows fast convergence, low steady-state error and good tracking capacity in comparison to standard equalizers operating on a sample-by-...
详细信息
ISBN:
(纸本)9781424485369;9781424485345
We propose a block equalization algorithm using optimal step size. The algorithm shows fast convergence, low steady-state error and good tracking capacity in comparison to standard equalizers operating on a sample-by-sample basis.
A Takagi-Sugeno fuzzy controller with reinforcement learning part is proposed in this paper, which is used to control a real inverted cart-pendulum system. The fuzzy controller part is a zero-order Takagi-Sugeno syste...
详细信息
ISBN:
(纸本)0780377028
A Takagi-Sugeno fuzzy controller with reinforcement learning part is proposed in this paper, which is used to control a real inverted cart-pendulum system. The fuzzy controller part is a zero-order Takagi-Sugeno system with four inputs and one output. The learning part is based on the gradient-descent algorithm which modifies the consequent parameters of the fuzzy rules. Because the expected output values are unknown, a reinforcement signal instead of the output error is used in learning process. The reinforcement signal is decided by the judgment of whether the action should be punished or rewarded and the degree of punishments or rewards. The performance of controlling a real inverted cart-pendulum system proves the validity and the superiority of the proposed fuzzy controller with reinforcement learning part.
The soft-sensor of ozone concentration is introduced. Six secondary variables are chosen in the soft-sensor model. The model is built based on the radial basis function neural network and its parameters are confirmed ...
详细信息
The soft-sensor of ozone concentration is introduced. Six secondary variables are chosen in the soft-sensor model. The model is built based on the radial basis function neural network and its parameters are confirmed by the gradient-descent algorithm. The model is implemented on the basis of the monitoring system that is necessary for ozone generation;thus, it requires no additional hardware cost. The response time of the model is less than 0.6 seconds. The experimental results indicate that the relative errors of the soft-sensor are less than 5%.
暂无评论