The Infona portal uses cookies, i.e. strings of text saved by a browser on the user's device. The portal can access those files and use them to remember the user's data, such as their chosen settings (screen view, interface language, etc.), or their login data. By using the Infona portal the user accepts automatic saving and using this information for portal operation purposes. More information on the subject can be found in the Privacy Policy and Terms of Service. By closing this window the user confirms that they have read the information on cookie usage, and they accept the privacy policy and the way cookies are used by the portal. You can change the cookie settings in your browser.
Local scour around bridge abutment is a time-dependent complex phenomenon encountered world-wide. It is difficult to establish a general empirical model that can be applied to all abutment conditions. In this paper, Radial basis function (RBF) Network has been used to predict the maximum scour depth around bridge abutment. An appropriate model is identified using experimental data from literature...
In the presentation major difficulties of designing neural networks are shown. It turn out that popular MLP (Multi Layer Perceptron) networks in most cases produces far from satisfactory results. Also, popular EBP (Error Back Propagation) algorithm is very slow and often is not capable to train best neural network architectures. Very powerful and fast LM (Levenberg- Marquardt) algorithm was unfortunately...
An important issue in design and implementation a neural network is that perturbations of training pattern pairs may cause some disadvantages to outputs. How the perturbations of training pattern pairs in Morphological Bidirectional Associative Memories (MBAMs) influence on the outputs is discussed in this paper. We define the outputs' max error to evaluate the robustness of the MBAMs. The related...
Training of parameters in RBF neural network, this article proposes an optimization of RBF neural network parameters algorithm, which can overcome the disadvantages of select of data center and weights in RBF neural network. The algorithm process input data normalization and compute network output and hidden layer output angle cosine firstly, a set of data being established as the network center when...
Since the blog service brings a wealth of information resources, blog search and classification are showing their great research value. This paper focuses on the blog classification on the personal vs. official facet. Our system adopts a two-stage strategy; in training model, lexicons are built automatically; in classification model, scoring and ranking are carried out orderly. Our experimental results...
Real life databases contain many features. Many of these features may be irrelevant or redundant. For example, data recording the age of each teacher in a school is unlikely to help in assessing the success of students' results in the school. Hence, relevant analysis is needed to be performed on the data in order to identify and remove any such irrelevant or redundant attributes from the learning...
In real world classification tasks, the original instances are represented by raw features. Usually domain related algorithms are needed to extract discriminative features. But the algorithms selection and additional parameters tuning are difficult for people with little domain knowledge and experience. In this paper, a new machine learning framework called "decompose learning" is proposed...
Support vector regression (SVR) is a common learning method for machines which is developed these years. Comparing with the other regression models, this method has the advantages of structural risk minimization and strong generalization ability. It is widely used in the prediction field and acquires good effects. The training characters of SVR model are very important to SVR. To solve the problem,...
The generalization ability of neural network is an important aspect affecting its application. Meanwhile, the selection of training samples has a great impact on this ability. In order to improve the completeness of training samples, a method of samples self-learning of BP neural network based on clustering is put forward in this paper. By using the method of clustering, new samples can be collected...
The capability to visually discern possible obstacles from the sky would be a valuable asset to a UAV for avoiding both other flying vehicles and static obstacles in its environment. The main contribution of this article is the presentation of a feasible approach to obstacle avoidance based on the segmentation of camera images into sky and non-sky regions. The approach is named the Sky Segmentation...
In this paper, we propose a novel method of label propagation for one-class learning. For binary (positive/negative) classification, the proposed method simultaneously measures the pair-wise similarity between samples and the negativity at every sample based on a cone-based model of local neighborhoods. Relying only on positive labeled samples as in one-class learning, the method estimates the labels...
Many problems in machine learning and computer vision consist of predicting multi-dimensional output vectors given a specific set of input features. In many of these problems, there exist inherent temporal and spacial dependencies between the output vectors, as well as repeating output patterns and input-output associations, that can provide more robust and accurate predictors when modelled properly...
In this paper, we present a new learning algorithm for self-constructing fuzzy neural networks (FNN). First, an initial network starts with no hidden neurons and grows neurons based on the growth criteria. After the generation process, a neuron pruning algorithm based on optimal brain surgeon (OBS) is employed to reduce the size of the FNN. After the structure design process, weight adjustment method...
The goal of ranking is to learn a real-valued ranking function that induces a ranking or ordering over an instance space. A learning algorithm is stable if the algorithm satisfies the hypothesis that the output of the algorithm varies in a limited way in response to small changes made to the training set. This paper studies the `almost everywhere' stability of ranking algorithms, notions of strong...
Improving the diversity of Neural Network Ensembles (NNE) plays an important role in creating robust classification systems in many fields. Several methods have been proposed in the literature to create such diversity using different sets of classifiers or using different portions of training/feature sets. Neural networks are often used as base classifiers in multiple classifier systems as they adapt...
Artificial Neural Networks (ANNs) have been used as a promising tools for many applications. In recent years, a computer-aided design approach based on ANNs has been introduced to microwave modeling, simulation and optimization. In this work, the characteristics parameters of the conductor-backed asymmetric coplanar waveguide (CB - ACPW) with one lateral ground plane have been determined with the...
In nowadays, hand gestures can be used as a more natural and convenient way for human computer interaction. The direct interface of hand gestures provides us a new way for communicating with the virtual environment. In this paper, we propose a new hand gesture recognition method using self-organizing map (SOM) with datagloves. The SOM method is a type of machine learning algorithm. It deals with the...
A new method of SRG Position Sensorless based on Self-adaptive Fuzzy Neural Network by using Finite Element Analysis is proposed in this paper. Through current and magnetic linkage to get angle of SRG rotor position, so the nonlinear mapping of current-magnetic linkage-angle can be built. By training these sample data, the angle of SRG rotor position can replaced by the output of ANFIS, to achieve...
In recent years, the use of machine learning algorithms (classifiers) has proven to be of great value in solving a variety of problems in software engineering including software faults prediction. This paper extends the idea of predicting software faults by using an ensemble of classifiers which has been shown to improve classification performance in other research fields. Benchmarking results on...
The traditional neural network is unavoidable to present local extreme value question, may result in failing training. On the basis of quantization of weapon system safe index, it has adopted neural network based on improved genetic algorithm to set up the systematic safety evaluation model of the weapon. It utilizes improved genetic algorithm to optimize the weight of neural network and get the final...
Set the date range to filter the displayed results. You can set a starting date, ending date or both. You can enter the dates manually or choose them from the calendar.