SVM is the structural risk minimization of statistical learning theory developed on the basis of a pattern recognition method, based on limited sample information and the complexity of the model to find the best compr...
详细信息
SVM is the structural risk minimization of statistical learning theory developed on the basis of a pattern recognition method, based on limited sample information and the complexity of the model to find the best compromise between the generalization ability. As there is a supervised learning method, the standard SVM classification requires supervised learning algorithm based on the principle: from a limited number of labeled samples to learn the rules and the rule extended to the unknown non-tag samples.
The paper proposes a learning method for an evolving Radial Basis Neural Network that makes it possible in an online mode to adjust not only synaptic weights but also parameters of the radial basis functions and the n...
详细信息
The paper proposes a learning method for an evolving Radial Basis Neural Network that makes it possible in an online mode to adjust not only synaptic weights but also parameters of the radial basis functions and the network architecture. A special feature of architecture learning is that a number of neurons in the network can both increase and decrease with a sequential stream of information at the system input. The implementation of the proposed algorithms has low computational complexity. The proposed evolving neural network can process data in an online mode.
This paper deals with an implementation of a distribution static compensator (DSTATCOM) using a learning-based anti-Hebbian control algorithm for compensation of linear/nonlinear loads. The proposed control algorithm ...
详细信息
This paper deals with an implementation of a distribution static compensator (DSTATCOM) using a learning-based anti-Hebbian control algorithm for compensation of linear/nonlinear loads. The proposed control algorithm is used for extraction of fundamental active and reactive power components of load currents for estimating reference supply currents. This control algorithm is implemented on a developed DSTATCOM using a digital signal processor for reactive power compensation, harmonics elimination, and load balancing. Simulation and test results demonstrate satisfactory performance of the proposed control algorithm for the control of DSTATCOM under varying loads.
The intensity of the flow accelerated corrosion (FAC) process depends on a great number of parameters with a complicated effect on each other. The use of an intellectual neural network (INN) to solve the FAC predictio...
详细信息
The intensity of the flow accelerated corrosion (FAC) process depends on a great number of parameters with a complicated effect on each other. The use of an intellectual neural network (INN) to solve the FAC prediction problem makes it possible to estimate the mutual effects from all the factors involved, to identify the essential properties of the information obtained, and, ultimately, to improve the accuracy of prediction without determining the whole range of dependences among a great deal of factors on which the FAC process depends. An approach is proposed to the creation and training of an optimal neural network for the NPP piping FAC rate prediction problem. Matlab software was used to develop an intellectual neural network to address the problem of the wall thinning prediction for a straight pipe with the VVER NPP single-phase secondary fluid. The network has been trained using an elastic back propagation algorithm, a number of the NS configurations have been studied, and the findings have been analyzed. A conceptual framework has been built for the intellectual system in the form of three NS types: a replicative NS, a Kohonen self-organizing NS, and a back-propagation NS.
Control by interconnection (CbI) is a dynamic output-feedback approach used to control port-Hamiltonian (PH) systems. Here, both the plant and the controller are modelled in PH form, in terms of their own Hamiltonians...
详细信息
ISBN:
(纸本)9781479977888
Control by interconnection (CbI) is a dynamic output-feedback approach used to control port-Hamiltonian (PH) systems. Here, both the plant and the controller are modelled in PH form, in terms of their own Hamiltonians. However, obtaining an appropriate controller Hamiltonian is generally difficult. In this paper, we address this issue by using reinforcement learning (RL). Additionally due to the semi-supervised optimization nature of the RL algorithms, a performance criterion can be readily included in CbI. We demonstrate the usefulness of the proposed learning algorithm for stabilization of a manipulator arm.
We introduce an atomic congestion game with two types of agents, namely, cars and trucks, to model the traffic flow on a road over various time intervals of the day. Cars maximize their utility by finding a tradeoff b...
详细信息
We introduce an atomic congestion game with two types of agents, namely, cars and trucks, to model the traffic flow on a road over various time intervals of the day. Cars maximize their utility by finding a tradeoff between the time they choose to use the road, the average velocity of the flow at that time, and the dynamic congestion tax that they pay for using the road. In addition to these terms, the trucks have an incentive for using the road at the same time as their peers because they have platooning capabilities, which allow them to save fuel. The dynamics and equilibria of this game-theoretic model for the interaction between car traffic and truck platooning incentives are investigated. We use traffic data from Stockholm, Sweden, to validate parts of the modeling assumptions and extract reasonable parameters for the simulations. We use joint strategy fictitious play and average strategy fictitious play to learn a pure strategy Nash equilibrium of this game. We perform a comprehensive simulation study to understand the influence of various factors, such as the drivers' value of time and the percentage of the trucks that are equipped with platooning devices, on the properties of the Nash equilibrium.
Extreme learning machine (ELM) is originally proposed for single-hidden layer feed-forward neural networks (SLFN). From the functional equivalence of fuzzy logic systems and SLFN, the fuzzy logic systems can be interp...
详细信息
ISBN:
(纸本)9781450333771
Extreme learning machine (ELM) is originally proposed for single-hidden layer feed-forward neural networks (SLFN). From the functional equivalence of fuzzy logic systems and SLFN, the fuzzy logic systems can be interpreted as a special case of SLFN under some mild conditions. Hence the fuzzy logic systems can be trained using SLFN's learning algorithms. Considering the same equivalence, ELM is utilized here to train interval type-2 fuzzy logic systems (IT2FLSs). Based on the working principle of the ELM, the parameters of the antecedent of IT2FLSs are randomly generated while the consequent part of IT2FLSs is optimized using Moore-Penrose generalized inverse of ELM. Application of the developed model to electricity load forecasting is another novelty of the research work. Experimental results shows better forecasting performance of the proposed model over the two frequently used forecasting models.
In cognitive radio networks, secondary users (SUs) face two conflicting objectives. Each SU seeks to minimize the sensing duration while maximizing the detection probability of primary users (PU) to avoid interfering ...
详细信息
In cognitive radio networks, secondary users (SUs) face two conflicting objectives. Each SU seeks to minimize the sensing duration while maximizing the detection probability of primary users (PU) to avoid interfering with their transmissions. Both objectives have a substantial effect on energy efficiency. This paper investigates a noncooperative setting for selecting the sensing duration when multiple SUs operate in the same network. Here, each SU has a certain throughput requirement. The interaction among SUs is captured via a satisfaction strategic game with explicitly stated throughput demands. We prove that depending on the throughput requirements, either zero, one or two Satisfaction Equilibria (SE) exist. We then provide a fully distributed learning algorithm (SELA) to discover them. Extensive simulation results show the validity of the proposed SELA and illustrate the relationship between the throughput demand and the sensing duration.
Occupant dynamic presence and characteristics associated with lighting loads/usage in residential buildings are not replicated in most practices currently adopted in modelling lighting profile. This study involves the...
详细信息
Occupant dynamic presence and characteristics associated with lighting loads/usage in residential buildings are not replicated in most practices currently adopted in modelling lighting profile. This study involves the use of adaptive neural fuzzy inference system (ANFIS) for lighting load profile prediction. Natural light, occupancy (active) and income level are the characterization (variables) factors considered in this investigation. The accuracy of the developed prediction models in relation to various income earners groups were analyzed using statistical measures;correlation output of the ANFIS approach and the impact of the characteristics on the lighting profile development in relation to trend analysis were also employed. Results obtained after validation of the developed models using investigative data, metering data and regression model showed a better correlation and root mean square error (RMSE) in comparison with actual values. The intelligence approach showed a better correlation of fit and good learning predictive accuracy in terms of behavioural and environmental variableness;and presents its output according to the complex nature of lighting usage in relation to the variables. The efficacy of the method was also validated. (C) 2015 Elsevier B.V. All rights reserved.
暂无评论