Search results for " network"
showing 10 items of 6428 documents
Why Cortices? Neural Networks for Visual Information Processing
1989
Neural networks for the processing of sensory information show remarkable similarities between different species and across different sensory modalities. As an example, cortical organization found in the mamalian neopallium and in the optic tecta of most vertebrates appears to be equally appropriate as a substrate for visual, auditory, and somatosensory information processing. In this paper, we formulate three structural principles of the vertebrate visual cortex that allow to analyze structure and function of these neural networks on an intermediate level of complexity. Computational applications are taken from the field of early vision. The proposed principles are: (a) Average anatomy, i …
Improving the Competency of Classifiers through Data Generation
2001
This paper describes a hybrid approach in which sub-symbolic neural networks and symbolic machine learning algorithms are grouped into an ensemble of classifiers. Initially each classifier determines which portion of the data it is most competent in. The competency information is used to generated new data that are used for further training and prediction. The application of this approach in a difficult to learn domain shows an increase in the predictive power, in terms of the accuracy and level of competency of both the ensemble and the component classifiers.
Neural network prediction in a system for optimizing simulations
2002
Neural networks have been widely used for both prediction and classification. Back-propagation is commonly used for training neural networks, although the limitations associated with this technique are well documented. Global search techniques such as simulated annealing, genetic algorithms and tabu search have also been used for this purpose. The developers of these training methods, however, have focused on accuracy rather than training speed in order to assess the merit of new proposals. While speed is not important in settings where training can be done off-line, the situation changes when the neural network must be trained and used on-line. This is the situation when a neural network i…
Clustering Quality and Topology Preservation in Fast Learning SOMs
2008
The Self-Organizing Map (SOM) is a popular unsupervised neural network able to provide effective clustering and data visualization for data represented in multidimensional input spaces. In this paper, we describe Fast Learning SOM (FLSOM) which adopts a learning algorithm that improves the performance of the standard SOM with respect to the convergence time in the training phase. We show that FLSOM also improves the quality of the map by providing better clustering quality and topology preservation of multidimensional input data. Several tests have been carried out on different multidimensional datasets, which demonstrate better performances of the algorithm in comparison with the original …
A Feed-Forward Neural Network for Robust Segmentation of Color Images
1999
A novel approach for segmentation of color images is proposed. The approach is based on a feed-forward neural network that learns to recognize the hue range of meaningful objects. Experimental results showed that the proposed method is effective and robust even in presence of changing environmental conditions. The described technique has been tested in the framework of the Robot Soccer World Cup Initiative (RoboCup). The approach is fully general and it may be successfully employed in any intermediate level image-processing task, where the color is a meaningful descriptor.
ConvLSTM Neural Networks for seismic event prediction in Chile
2021
Predicting seismic risk is a challenging task in order to avoid catastrophic effects. In this work, two models based on Convolutional Network (CNN) and Long Short Term Memory (LSTM) networks are proposed to predict the seismic risk in Chile. In particular, a ConvLSTM and a Multi-column ConvLSTM network are used for the prediction of the average number of seismic events greater than 2,8 magnitude on the Richter scale, in the Chilean regions of Coquimbo and Araucania between the years 2010 and 2017. For this model, the values of the intensity function estimated through an ETAS model and the accumulated displacement prior to a the seismic events are used as inputs. In particular, given the spa…
Testing selected optimal descriptors with artificial neural networks
2013
Eleven properties have been modeled with the objective of checking the importance for model purposes of mixed descriptors made of empirical parameters, molecular connectivity indices and random numbers. The mixed descriptors with random indices have a descriptive character which is satisfactorily confirmed by the leave-one-out method of statistical analysis. The introduction of a partition of the set of compounds into training and evaluation sets decreases drastically the probability to find a mixed descriptor with random indices with good model quality. Two properties, the magnetic susceptibility and the elutropic values, insist on having optimal descriptors with random indices. The overal…
An Encrypted Traffic Classification Framework Based on Convolutional Neural Networks and Stacked Autoencoders
2020
In recent years, deep learning-based encrypted traffic classification has proven to be effective; especially, using neural networks to extract features from raw traffic to classify encrypted traffic. However, most of the neural networks need a fixed-sized input, so that the raw traffic need to be trimmed. This will cause the loss of some information; for example, we do not know the number of packets in a session. To solve these problems, a framework, which implements both a convolutional neural network (CNN) and a stacked autoencoder (SAE), is proposed in this paper. This framework uses a CNN to extract high-level features from raw network traffic and uses an SAE to encode the 26 statistica…
A Neural Architecture for 3D Segmentation
2003
An original neural scheme for segmentation of range data is presented, which is part of a more general 3D vision system for robotic applications. The entire process relies on a neural architecture aimed to perform first order image irradiance analysis, that is local estimation of magnitude and orientation of the image irradiance gradient.