Serwis Infona wykorzystuje pliki cookies (ciasteczka). Są to wartości tekstowe, zapamiętywane przez przeglądarkę na urządzeniu użytkownika. Nasz serwis ma dostęp do tych wartości oraz wykorzystuje je do zapamiętania danych dotyczących użytkownika, takich jak np. ustawienia (typu widok ekranu, wybór języka interfejsu), zapamiętanie zalogowania. Korzystanie z serwisu Infona oznacza zgodę na zapis informacji i ich wykorzystanie dla celów korzytania z serwisu. Więcej informacji można znaleźć w Polityce prywatności oraz Regulaminie serwisu. Zamknięcie tego okienka potwierdza zapoznanie się z informacją o plikach cookies, akceptację polityki prywatności i regulaminu oraz sposobu wykorzystywania plików cookies w serwisie. Możesz zmienić ustawienia obsługi cookies w swojej przeglądarce.
Through building up the functional relationship between the error E and the learning rate η, we propose one kind of new improved learning rate BP algorithm. This improved BP algorithm adopts serial dynamic adaptive learning rate, thus according to different error E to determine different learning rates. Compared with VLBP, the simulation result shows this improved variable learning rate BP algorithm...
Injecting weight noise during training has been proposed for almost two decades as a simple technique to improve fault tolerance and generalization of a multilayer perceptron (MLP). However, little has been done regarding their convergence behaviors. Therefore, we presents in this paper the convergence proofs of two of these algorithms for MLPs. One is based on combining injecting multiplicative weight...
Improving fault tolerance of a neural network is an important issue that has been studied for more than two decades. Various algorithms have been proposed in sequel and many of them have succeeded in attaining a fault tolerant neural network. Amongst all, on-line node fault injection-based algorithms are one type of these algorithms. Despite its simple implementation, theoretical analyses on these...
BP algorithm is a very important and classic learning algorithm. It have a wide range of applications in pattern recognition, image processing and analysis and control areas. However, in practice, we found that BP algorithm still have inadequates, such as the algorithm's convergence is slowly, easy to converge to local minimum points, but not the overall optimal, Multiple iterations, numerical stability...
This paper studied the accelerating convergence of the vector sequences generated by BP algorithm with vector epsilon algorithm, and presented the conclusion that the algorithms is not only convergent but also accelerated. Finally, we tested them for three classical artificial neural network problems. By numerical experiments, results shown that can reduce CPU time for computation and improve the...
We propose a new algorithm for learning kernels for variants of the Normalized Cuts (NCuts) objective - i.e., given a set of training examples with known partitions, how should a basis set of similarity functions be combined to induce NCuts favorable distributions. Such a procedure facilitates design of good affinity matrices. It also helps assess the importance of different feature types for discrimination...
Several object categorization algorithms use kernel methods over multiple cues, as they offer a principled approach to combine multiple cues, and to obtain state-of-the-art performance. A general drawback of these strategies is the high computational cost during training, that prevents their application to large-scale problems. They also do not provide theoretical guarantees on their convergence rate...
The production period of the crystalline aluminium chloride is considerably long. However, the offline assay of AlCl3??6H2O content has large time delay. Thus soft sensor modeling is needed to analyze its content, and estimate the value to improve the product quality. The conventional back-propagation (BP) neural network training is easily trapped to the local minimum, To overcome this embarrassment,...
Neural Network is an effective tool in the field of pattern recognition. The neural network classifies the pattern from the training data and recognizes if the testing data holds that pattern. The classical Back propagation (BP) algorithm is generally used to train the neural network for its simplicity. The basic drawback of this algorithm is its uncertainty and long training time and it searches...
Personal Credit Scoring is of great significance for commercial banks to circumvent credit consumption, the original BP algorithm's convergence rate is slow, learning precision is low, the training process is easy to fall into local minimum, this paper presents an improved algorithm with variable learning rate based on BP algorithm, and applied to simulate personal credit scoring. After comparing...
The notion of using context information for solving high-level vision and medical image segmentation problems has been increasingly realized in the field. However, how to learn an effective and efficient context model, together with an image appearance model, remains mostly unknown. The current literature using Markov Random Fields (MRFs) and Conditional Random Fields (CRFs) often involves specific...
This paper proposes a new global optimization technique in which combines population migration algorithm (PMA) and radial basis function (RBF) neural networks learning algorithm for training RBF neural network. Compared with the traditional RBF training algorithm, the simulation results show that the method has a higher accuracy in a stringency and works well in avoiding sticking in local minima.
In order to improve the convergence speed of reinforcement learning and avoid the local optimum for multirobot systems, a new method of cooperative Q-learning based on maturity of the policy is presented. The learning process is executed at the blackboard architecture making use of all the robots in the training scenario to explore the learning space and collect experiences. The reinforcement learning...
In this paper we present a new continuous-time recurrent neurofuzzy network structure for modeling and identification of a class of nonlinear systems, using a training algorithm motivated from previous works in adaptive observers. Using only output measurements and the knowledge of an excitation input signal, the proposed network is trained by generating estimates of an ideal network and jointly identifying...
Recently, cost-sensitive data mining has been an area of extensive research interests. Intelligent ant colony classification algorithm is introduced in cost-sensitive data mining method in order to obtain satisfied classification results by interaction of ant individuals. The convergence rate of classification is increased by using of metacost's meta-learning theory. Moreover, Boosting theory is investigated...
Reinforcement learning is learning what to do - how to map situations to actions - so as to maximize a numerical reward signal. In allusion to the problem that Q-Learning, which uses discount reward as the evaluation criterion, cannot show the affect of the action to the next situation, the paper puts forward AR-Q-Learning based on the average reward and Q-Learning. In allusion to the Curse Of Dimensionality,...
An optimal design approach of Hilbert convertor is researched in detail based on the neural-network algorithm. The main idea is to minimize the sum of the square errors between the amplitude-frequency response of the desired Hilbert convertor and that of the designed by training the weight vector of neural-network, then obtains the impulse response of Hilbert convertor. The convergence theorem of...
In this paper, an improved Levenberg-Marquardt learning algorithm based on terminal attractors for feedforward neural networks is proposed. The effectiveness of the proposed algorithm in improving learning speed is shown by the simulation results.
The classification performance of nearest prototype classifiers largely relies on the prototype learning algorithms, such as the learning vector quantization (LVQ) and the minimum classification error (MCE). This paper proposes a new prototype learning algorithm based on the minimization of a conditional log-likelihood loss (CLL), called log-likelihood of margin (LOGM). A regularization term is added...
Learning from data is one of the basic ways humans perceive the world and acquire the knowledge. Support vector machine (SVM for short) has emerged as a good classification technique and achieved excellent generalization performance in a variety of applications. Training SVM on a dataset of huge size with millions of data is a challenging problem since it is computationally expensive and the memory...
Podaj zakres dat dla filtrowania wyświetlonych wyników. Możesz podać datę początkową, końcową lub obie daty. Daty możesz wpisać ręcznie lub wybrać za pomocą kalendarza.