Results 1  10
of
34
Problem Solving With Reinforcement Learning
, 1995
"... This dissertation is submitted for consideration for the dwree of Doctor' of Philosophy at the Uziver'sity of Cambr'idge Summary This thesis is concerned with practical issues surrounding the application of reinforcement lear'ning techniques to tasks that take place in high dimensional continuous ..."
Abstract

Cited by 45 (0 self)
 Add to MetaCart
This dissertation is submitted for consideration for the dwree of Doctor' of Philosophy at the Uziver'sity of Cambr'idge Summary This thesis is concerned with practical issues surrounding the application of reinforcement lear'ning techniques to tasks that take place in high dimensional continuous statespace environments. In particular, the extension of online updating methods is considered, where the term implies systems that learn as each experience arrives, rather than storing the experiences for use in a separate offline learning phase. Firstly, the use of alternative update rules in place of standard Qlearning (Watkins 1989) is examined to provide faster convergence rates. Secondly, the use of multilayer perceptton (MLP) neural networks (Rumelhart, Hinton and Williams 1986) is investigated to provide suitable generalising function approximators. Finally, consideration is given to the combination of Adaptive Heuristic Critic (AHC) methods and Qlearning to produce systems combining the benefits of realvalued actions and discrete switching
Structural adaptation and generalization in supervised feedforward networks, d
 Artif. Neural Networks
, 1994
"... This work explores diverse techniques for improving the generalization ability of supervised feedforward neural networks via structural adaptation, and introduces a new network structure with sparse connectivity. Pruning methods which start from a large network and proceed in trimming it until a sa ..."
Abstract

Cited by 31 (22 self)
 Add to MetaCart
This work explores diverse techniques for improving the generalization ability of supervised feedforward neural networks via structural adaptation, and introduces a new network structure with sparse connectivity. Pruning methods which start from a large network and proceed in trimming it until a satisfactory solution is reached, are studied first. Then, construction methods, which build a network from a simple initial configuration, are presented. A survey of related results from the disciplines of function approximation theory, nonparametric statistical inference and estimation theory leads to methods for principled architecture selection and estimation of prediction error. A network based on sparse connectivity is proposed as an alternative approach to adaptive networks. The generalization ability of this network is improved by partly decoupling the outputs. We perform numerical simulations and provide comparative results for both classification and regression problems to show the generalization abilities of the sparse network. 1
Fast pruning using principal components
 ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS, MORGANKAUFMANN
"... We present a new algorithm for eliminating excess parameters and improving network generalization after supervised training. The method, "Principal Components Pruning (PCP)", is based on principal component analysis of the node activations of successive layers of the network. It is simple, cheap to ..."
Abstract

Cited by 28 (4 self)
 Add to MetaCart
We present a new algorithm for eliminating excess parameters and improving network generalization after supervised training. The method, "Principal Components Pruning (PCP)", is based on principal component analysis of the node activations of successive layers of the network. It is simple, cheap to implement, and effective. It requires no network retraining, and does not involve calculating the full Hessian of the cost function. Only the weight andthenode activity correlation matrices for each layer of nodes are required. We demonstrate the efficacy of the method on a regression problem using polynomial basis functions, and on an economic time series prediction problem using a twolayer, feedforward network.
Speech Recognition Using Augmented Conditional Random Fields
"... Abstract—Acoustic modeling based on hidden Markov models (HMMs) is employed by stateoftheart stochastic speech recognition systems. Although HMMs are a natural choice to warp the time axis and model the temporal phenomena in the speech signal, their conditional independence properties limit their ..."
Abstract

Cited by 22 (0 self)
 Add to MetaCart
Abstract—Acoustic modeling based on hidden Markov models (HMMs) is employed by stateoftheart stochastic speech recognition systems. Although HMMs are a natural choice to warp the time axis and model the temporal phenomena in the speech signal, their conditional independence properties limit their ability to model spectral phenomena well. In this paper, a new acoustic modeling paradigm based on augmented conditional random fields (ACRFs) is investigated and developed. This paradigm addresses some limitations of HMMs while maintaining many of the aspects which have made them successful. In particular, the acoustic modeling problem is reformulated in a data driven, sparse, augmented space to increase discrimination. Acoustic context modeling is explicitly integrated to handle the sequential phenomena of the speech signal. We present an efficient framework for estimating these models that ensures scalability and generality. In the TIMIT
Pruning with Generalization Based Weight Saliencies: γOBD, γOBS
 Advances in Neural Information Processing Systems 8, Proceedings of the 1995 Conference
, 1995
"... The purpose of most architecture optimization schemes is to improve generalization. In this presentation we suggest to estimate the weight saliency as the associated change in generalization error if the weight is pruned. We detail the implementation of both an O(N)storage scheme extending OBD, as ..."
Abstract

Cited by 15 (2 self)
 Add to MetaCart
The purpose of most architecture optimization schemes is to improve generalization. In this presentation we suggest to estimate the weight saliency as the associated change in generalization error if the weight is pruned. We detail the implementation of both an O(N)storage scheme extending OBD, as well as an O(N 2 ) scheme extending OBS. We illustrate the viability of the approach on prediction of a chaotic time series. 1 BACKGROUND Optimization of feedforward neural networks by pruning is a wellestablished tool, used in many practical applications. By careful fine tuning of the network architecture we may improve generalization, decrease the amount of computation, and facilitate interpretation. The two most widely used schemes for pruning of feedforward nets are: Optimal Brain Damage (OBD) due to (LeCun et al., 90) and the Optimal Brain Surgeon (OBS) (Hassibi et al., 93). Both schemes are based on weight ranking according to saliency defined as the change in training error whe...
Fast Network Pruning and Feature Extraction Using the UnitOBS Algorithm
 in Advances in Neural Information Processing Systems
, 1997
"... The algorithm described in this article is based on the OBS algorithm by Hassibi, Stork and Wolff ([1] and [2]). The main disadvantage of OBS is its high complexity. OBS needs to calculate the inverse Hessian to delete only one weight (thus needing much time to prune a big net). A better algorithm s ..."
Abstract

Cited by 12 (0 self)
 Add to MetaCart
The algorithm described in this article is based on the OBS algorithm by Hassibi, Stork and Wolff ([1] and [2]). The main disadvantage of OBS is its high complexity. OBS needs to calculate the inverse Hessian to delete only one weight (thus needing much time to prune a big net). A better algorithm should use this matrix to remove more than only one weight, because calculating the inverse Hessian takes the most time in the OBS algorithm. The algorithm, called UnitOBS, described in this article is a method to overcome this disadvantage. This algorithm only needs to calculate the inverse Hessian once to remove one whole unit thus drastically reducing the time to prune big nets. A further advantage of UnitOBS is that it can be used to do a feature extraction on the input data. This can be helpful on the understanding of unknown problems. 1 Introduction This article is based on the technical report [3] about speeding up the OBS algorithm. The main target of this work was to reduce the ...
Lean Artificial Neural Networks Regularization Helps Evolution
, 1996
"... A main criterion for the accuracy of solutions of an Artificial Neural Network (ANN) for classification tasks is the architecture. In order to find problemadapted topologies of ANNs, we adopted the evolutionary approach to ANN design by employing a Genetic Algorithm (GA) to evolve ANNs which are re ..."
Abstract

Cited by 6 (4 self)
 Add to MetaCart
A main criterion for the accuracy of solutions of an Artificial Neural Network (ANN) for classification tasks is the architecture. In order to find problemadapted topologies of ANNs, we adopted the evolutionary approach to ANN design by employing a Genetic Algorithm (GA) to evolve ANNs which are represented using a direct encoding method. As ANNs of low complexity show better generalization capabilities than more complex networks, we incorporated a regularization term into the fitness function which together with the problem representation is determining the preferred regions of the search space the GA will focus on. Especially, we investigated two different regularization terms proposed in literature and experimented with various degrees of impact on the fitness function. The parallel netGEN system which has been implemented by the authors is generating problemadapted FeedForward ANNs being trained by ErrorBackPropagation. Empirical results on a real world problem taken from ...
Generalization in Neural Networks
 Projektarbejde ved Elektronisk Institut, DTU
, 1993
"... 1 Abstract This report is concerned with methods for optimizing the generalization ability of neural networks. The framework is developed to deal with regression type problems, where the networks are trained on a limited amount of noisy data. In this context the problem can be formulated as findin ..."
Abstract

Cited by 5 (0 self)
 Add to MetaCart
1 Abstract This report is concerned with methods for optimizing the generalization ability of neural networks. The framework is developed to deal with regression type problems, where the networks are trained on a limited amount of noisy data. In this context the problem can be formulated as finding the optimal trade off between data fit and model complexity. Two paradigms for reducing model complexity are discussed: pruning and weight decay. It is shown by numerical experiments that application of weight decay is essential for obtaining good generalization performance. This is explained by the way in which weight decay confines the space of possible networks to a space of `reasonable' networks. Two methods for making statistical estimates of the generalization performance without use of validation sets are presented: the Generalization method and the Bayesian method. The advantage of not needing validation sets is that all available data can be utilized in the training phase. This f...
Evolution of Neural Network Training Set through Addition of Virtual Samples
 in Proc. 1996 IEEE Int. Conf. Evolutionary Computation, ICEC’96
, 1996
"... Using an oversized neural network or too small a training sample set results in overfitting. In order to improve generalization capability, either the network should be reduced or additional training samples have to be collected. Obtaining additional training samples, however, can be often very expe ..."
Abstract

Cited by 4 (1 self)
 Add to MetaCart
Using an oversized neural network or too small a training sample set results in overfitting. In order to improve generalization capability, either the network should be reduced or additional training samples have to be collected. Obtaining additional training samples, however, can be often very expensive or impossible. Here we propose an evolutionary approach where new virtual samples are added to the training sample set as a population of MLPs evolve over generations. At each generation, these newly added virtual samples are used to retrain the MLPs. This approach is in contrast to previous evolutionary neural network approaches where connection weights, network architectures, learning rules, or their mixtures evolve. A preliminary result obtained from a robot arm kinematics problem is promising. The generalization error was reduced more than 50%. The approach can be applied in various practical situations where additional training samples are expensive or impossible. I. Introduction ...
Learning and adaptation in cognitive radios using neural networks
 Proc. 5th IEEE CCNC, pp.998 2008
"... Abstract — The estimation of the communication performance achievable with respect to environmental factors and configuration parameters plays a key role in the optimization process performed by a Cognitive Radio according to the original definition by Mitola [1]. In this paper we propose the use of ..."
Abstract

Cited by 4 (0 self)
 Add to MetaCart
Abstract — The estimation of the communication performance achievable with respect to environmental factors and configuration parameters plays a key role in the optimization process performed by a Cognitive Radio according to the original definition by Mitola [1]. In this paper we propose the use of Multilayered Feedforward Neural Networks as an effective technique for realtime characterization of the communication performance which is based on measurements carried out by the device and therefore offers some interesting learning capabilities. Learn associate environment and configuration with experienced performance Observe take measurements from environment Act configure system experience performance for new configuration Orient infer environment from measurements predict performance for possible configurations Decide identify configuration which provides best performance I.