Radial basis function (RBF) networks of Gaussian activation functions have been widely used in many applications due to its simplicity, robustness, good approximation and generalization ability, etc.. However, the tra...
详细信息
ISBN:
(纸本)9783540723820
Radial basis function (RBF) networks of Gaussian activation functions have been widely used in many applications due to its simplicity, robustness, good approximation and generalization ability, etc.. However, the training of such a RBF network is still a rather difficult task in the general case and the main crucial problem is how to select the number and locations of the hidden units appropriately. In this paper, we utilize a new kind of Bayesian Ying-Yang (BYY) automated model selection (AMS) learning algorithm to select the appropriate number and initial locations of the hidden units or Gaussians automatically for an input data set. It is demonstrated well by the experiments that this BYY-AMS training method is quite efficient and considerably outperforms the typical existing training methods on the training of RBF networks for both clustering analysis and nonlinear time series prediction.
Cerebellar Model Articulation Controller Neural Networks (CMAC NN) is one of the intelligent systems used for modeling, identification, classification, and controlling of nonlinear systems. In this paper, the mathemat...
详细信息
ISBN:
(纸本)9789077381595
Cerebellar Model Articulation Controller Neural Networks (CMAC NN) is one of the intelligent systems used for modeling, identification, classification, and controlling of nonlinear systems. In this paper, the mathematical model of CMAC is presented. CMAC is implemented using Simulink environment and its parameters are tuned to get the best CMAC control action. Three different learning algorithms are tested, using a constant learning rate, a variable learning rate, and learning by the control action of the conjugate conventional controller. The effect of varying CMAC parameters is studied and discussed. The simulation results showed that the learning algorithm based on constant learning rate gives the best performance.
In a finite game the Stochastically Stable States (SSSs) of adaptive play are contained in the set of minimizers of resistance trees. Also, in potential games, the SSSs of the log-linear learning algorithm are the min...
详细信息
ISBN:
(纸本)9783319675404;9783319675398
In a finite game the Stochastically Stable States (SSSs) of adaptive play are contained in the set of minimizers of resistance trees. Also, in potential games, the SSSs of the log-linear learning algorithm are the minimizers of the potential function. The SSSs can be characterized using the resistance trees of a Perturbed Markov Chain (PMC), they are the roots of minimum resistance tree. Therefore, computing the resistance of trees in PMC is important to analyze the SSSs of learning algorithms. A learning algorithm defines the Transition Probability Function (TPF) of the induced PMC on the action space of the game. Depending on the characteristics of the algorithm the TPF may become composite and intricate. Resistance computation of intricate functions is difficult and may even be infeasible. Moreover, there are no rules or tools available to simplify the resistance computations. In this paper, we propose novel rules that simplify the computation of resistance. We first, give a generalized definition of resistance that allows us to overcome the limitations of the existing definition. Then, using this new definition we develop the rules that reduce the resistance computation of composite TPF into resistance computation of simple functions. We illustrate their strength by efficiently computing the resistance in log-linear and payoff-based learning algorithms. They provide an efficient tool for characterizing SSSs of learning algorithms in finite games.
This paper should contribute to a structured and theoretical view of the backpropagation algorithm and some of its well-known extensions. Basing on a mathematical investigation of the algorithms conditions for structu...
详细信息
ISBN:
(纸本)0780327683;0780327691
This paper should contribute to a structured and theoretical view of the backpropagation algorithm and some of its well-known extensions. Basing on a mathematical investigation of the algorithms conditions for structured improvements and developments of these techniques will be described. The construction of adaptive parameter regulations for learning and momentum rate will follow. These parameter regulations allow the presentation of adaptive learning techniques. It will be shown that off-line versions of these techniques represent minimization methods which are exact in mathematical sense. Under consideration of complexity conditions on-line algorithms will be preferred and described in detail. Finally their numerical behaviour will be investigated and simulation results will be presented in comparison with standard algorithms.
Reinforcement learning (RL) is a machine learning method that can learn an optimal strategy for a system without knowing the mathematical model of the system. Many RL algorithms are successfully applied in various fie...
详细信息
ISBN:
(纸本)9780780394902
Reinforcement learning (RL) is a machine learning method that can learn an optimal strategy for a system without knowing the mathematical model of the system. Many RL algorithms are successfully applied in various fields. However, each algorithm has its advantages and disadvantages. With the increasing complexity of environments and tasks, it is difficult for a single learning algorithm to cope with complicated learning problems with high performance. This motivated us to combine some learning algorithms to improve the learning quality. This paper proposes a new multiple learning architecture, "Aggregated Multiple Reinforcement learning System (AMRLS)". AMRLS adopts three different learning algorithms to learn individually and then combines their results with aggregation methods. To evaluate its performance, AMRLS is tested on two different environments: a Cart-pole System and a Maze environment. The presented simulation results reveal that aggregation not only provides robustness and fault tolerance ability, but also produces more smooth learning curves and needs fewer learning steps than individual learning algorithms.
In this paper, we present an evaluation of learning algorithms of a novel rule evaluation support method for post processing of mined results with rule evaluation models based on objective indices. Post-processing of ...
详细信息
ISBN:
(纸本)9781424409907
In this paper, we present an evaluation of learning algorithms of a novel rule evaluation support method for post processing of mined results with rule evaluation models based on objective indices. Post-processing of mined results is one of the key processes in a data mining process. However, it is difficult for human experts to completely evaluate several thousands of rules from a large dataset with noises. To reduce the costs in such rule evaluation task, we have developed the rule evaluation support method with rule evaluation models, which learn from objective indices for mined classification rules and evaluations by a human expert for each rule. To enhance adaptability of rule evaluation models, we introduced a constructive meta-learning system to choose proper learning algorithms. Then, we have done the case study on the meningitis data mining as an actual problem.
In this paper, we consider a two-dimensional (2-D) formation problem for multi-agent systems subject to switching topologies that dynamically change along both a finite time axis and an infinite iteration axis. We pre...
详细信息
ISBN:
(纸本)9781479901784
In this paper, we consider a two-dimensional (2-D) formation problem for multi-agent systems subject to switching topologies that dynamically change along both a finite time axis and an infinite iteration axis. We present a distributed iterative learning control (ILC) algorithm via the nearest neighbor rules. By employing the 2-D approach, we develop both the asymptotic and exponentially fast convergence of our formation ILC, which can be guaranteed by conditions in terms of the spectral radius and the matrix norms, respectively.
Wearables have emerged as a revolutionary technology in many application domains including healthcare and fitness. Machine learning algorithms, which form the core intelligence of wearables, traditionally deduce a com...
详细信息
ISBN:
(纸本)9781467390057
Wearables have emerged as a revolutionary technology in many application domains including healthcare and fitness. Machine learning algorithms, which form the core intelligence of wearables, traditionally deduce a computational model from a set of training examples to detect events of interest (e.g. activity type). However, in the dynamic environment in which wearables typically operate in, the accuracy of a computational model drops whenever changes in configuration of the system (such as device type and sensor orientation) occur. Therefore, there is a need to develop systems which can adapt to the new configuration autonomously. In this paper, using transfer learning as an organizing principle, we develop several algorithms for data mapping. The data mapping algorithms employ effective signal similarity methods and are used to adapt the system to the new configuration. We demonstrate the efficacy of the data mapping algorithms using a publicly available dataset on human activity recognition.
This paper analyses the performance and efficiency of reinforcement learning algorithms for matching the availability of uncertain renewable energy sources (RES) with flexible loads. More specifically, this paper prop...
详细信息
ISBN:
(纸本)9781665467612
This paper analyses the performance and efficiency of reinforcement learning algorithms for matching the availability of uncertain renewable energy sources (RES) with flexible loads. More specifically, this paper proposes a novel scalable (with the number of customers) and efficient learning-based energy matching solution for maximizing social welfare in dynamic matching markets. The key features of the proposed solution is combining a simple rule-based function and a learnable component to achieve the aforementioned properties. The output of the learnable component is a probability distribution over the matching decisions for the individual customers. The proposed hybrid model enables the learning algorithm to find an effective matching policy that simultaneously satisfies the customers' servicing preferences. Extensive simulations are presented to show that the learning algorithm learns an effective matching policy for different generation-consumption profiles despite of the complexity reduction. The proposed solution exhibits significantly better performance compared to standard online matching heuristics such as Match on Arrival, Match to the Highest, and Match to the Earliest Deadline policies.
Identification of fuzzy rules is an important issue in designing of a fuzzy neural network (FNN). However, there is no systematic design procedure at present. In this paper we present a genetic algorithm (G,4) based l...
详细信息
ISBN:
(纸本)9780769529769
Identification of fuzzy rules is an important issue in designing of a fuzzy neural network (FNN). However, there is no systematic design procedure at present. In this paper we present a genetic algorithm (G,4) based learning algorithm to make use of the known member-ship function to identify the fuzzy rules form a large set of all possible rules. The proposed learning algorithm initially considers all possible rules then uses the training data and the fitness function to perform rule selection. The proposed GA based learning algorithm has been tested with two different sets of training data. The results obtained from the experiments are promising and demonstrate that the proposed G,4 based learning algorithm can provide a reliable mechanism for fuzzy rule selection.
暂无评论