Abstract—A novel approach is presented in this paper for improving the performance of neural-network classifiers in image recognition, segmentation, or coding applications, based on a retraining procedure at the user level. The procedure includes: 1) a training algorithm for adapting the network weights to the current condition; 2) a maximum a posteriori (MAP) estimation procedure for optimally selecting the most representative data of the current environment as retraining data; and 3) a decision mechanism for determining when network retraining should be activated. The training algorithm takes into consideration both the former and the current network knowledge in order to achieve good generalization. The MAP estimation procedure models the network output as a Markov random field (MRF) and optimally selects the set of training inputs and corresponding desired outputs. Results are presented which illustrate the theoretical developments as well as the performance of the proposed approach in real-life experiments. Index Terms—Image analysis, MPEG-4, neural-network retraining, segmentation, weight adaptation.

This paper presents a constructive algorithm for training cooperative neural-network ensembles (CNNEs). CNNE combines ensemble architecture design with cooperative training for individual neural networks (NNs) in ensembles. Unlike most previous studies on training ensembles, CNNE puts emphasis on both accuracy and diversity among individual NNs in an ensemble. In order to maintain accuracy among individual NNs, the number of hidden nodes in individual NNs are also determined by a constructive approach. Incremental training based on negative correlation is used in CNNE to train individual NNs for different numbers of training epochs. The use of negative correlation learning and different training epochs for training individual NNs reflect CNNEs emphasis on diversity among individual NNs in an ensemble. CNNE has been tested extensively on a number of benchmark problems in machine learning and neural networks, including Australian credit card assessment, breast cancer, diabetes, glass, heart disease, letter recognition, soybean, and Mackey--Glass time series prediction problems. The experimental results show that CNNE can produce NN ensembles with good generalization ability.

Abstract not found

Neural networks have been successfully applied to solve a variety of application problems including classification and function approximation. They are especially useful as function approximators because they do not require prior knowledge of the input data distribution and they have been shown to be universal approximators. In many applications, it is desirable to extract knowledge that can explain how the problems are solved by the networks. Most existing approaches have focused on extracting symbolic rules for classification. Few methods have been devised to extract rules from trained neural networks for regression. This article presents an approach for extracting rules from trained neural networks for regression. Each rule in the extracted rule set corresponds to a subregion of the input space and a linear function involving the relevant input attributes of the data approximates the network output for all data samples in this subregion. Extensive experimental results on 32 benchmark data sets demonstrate the effectiveness of the proposed approach in generating accurate regression rules.

This paper presents a model selection procedure which stresses the importance of the classic polynomial models as tools for evaluating the complexity of a given modeling problem, and for removing non-significant input variables. If the complexity of the problem makes a neural network necessary, the selection among neural candidates can be performed in two phases. In an additive phase, the most important one, candidate neural networks with an increasing number of hidden neurons are trained. The addition of hidden neurons is stopped when the effect of the round-off errors becomes significant, so that, for instance, confidence intervals cannot be accurately estimated. This phase leads to a set of approved candidate networks. In a subsequent subtractive phase, a selection among approved networks is performed using statistical Fisher tests. The series of tests starts from a possibly too large unbiased network (the full network), and ends with the smallest unbiased network whose input variables and hidden neurons all have a significant contribution to the regression estimate. This method was successfully tested against the real-world regression problems proposed at the NIPS2000 Unlabeled Data Supervised Learning Competition; two of them are included here as illustrative examples.

In this paper, an unsupervised video object (VO) segmentation and tracking algorithm is proposed based on an adaptable neural-network architecture. The proposed scheme comprises: 1) a VO tracking module and 2) an initial VO estimation module. Object tracking is handled as a classification problem and implemented through an adaptive network classifier, which provides better results compared to conventional motion-based tracking algorithms. Network adaptation is accomplished through an efficient and cost effective weight updating algorithm, providing a minimum degradation of the previous network knowledge and taking into account the current content conditions. A retraining set is constructed and used for this purpose based on initial VO estimation results. Two different scenarios are investigated. The first concerns extraction of human entities in video conferencing applications, while the second exploits depth information to identify generic VOs in stereoscopic video sequences. Human face/ body detection based on Gaussian distributions is accomplished in the first scenario, while segmentation fusion is obtained using color and depth information in the second scenario. A decision mechanism is also incorporated to detect time instances for weight updating. Experimental results and comparisons indicate the good performance of the proposed scheme even in sequences with complicated content (object bending, occlusion).

The problem of determining the architecture of a multilayer perceptron together with the disadvantages of the standard backpropagation algorithm, directed the research towards algorithms that determine not only the weights but also the structure of the network necessary for learning the data. We propose a Constructive Algorithm with Multiple Operators using Statistical Test (MOST) for determining the architecture. The networks that are constructed by MOST can have multiple hidden layers with multiple hidden units in each layer. The algorithm uses node removal, addition and layer addition and determines the number of nodes in layers by heuristics. It applies a statistical test to compare different architectures. The results are promising and near optimal.

Neural networks have been widely used as a tool for regression. They are capable of approximating any function and they do not require any assumption about the distribution of the data. The most commonly used architectures for regression are the feedforward neural networks with one or more hidden layers. In this paper, we present a network pruning algorithm which determines the number of units in the input and hidden layers of the networks. We compare the performance of the pruned networks to four regression methods namely, linear regression (LR), Naive Bayes (NB), k-nearest-neighbor (kNN), and a decision tree predictor M5 0 . On 32 publicly available data sets tested, the neural network method outperforms NB and kNN if the prediction errors are computed in terms of the root mean squared errors. Under this measurement metric, it also performs as well as LR and M5 0 . On the other hand, using the mean absolute error as the measurement metric, the neural network metho...

Abstract — We present a new approach to cope with unknown redundant systems. For this we present i) an online algorithm that learns general input-output restrictions and, ii) a method that, given a partial set of input-output variables, provides an estimate of the remaining ones, using the learned restrictions. We show applications of the algorithm using examples of direct and inverse robot kinematics. Index Terms — manifold learning, humanoid robots, redundancy, sensory-motor coordination I.

This dissertation explores the modular learning in artificial neural networks that mainly driven by the inspiration from the neurobiological basis of the human learning. The presented modularization approaches to the neural network design and learning are inspired by the engineering, complexity, psychological and neurobiological aspects. The main theme of this dissertation is to explore the organization and functioning of the brain to discover new structural and learning inspirations that can be subsequently utilized to design artificial neural network.