A method to determine the required number of neural-network training repetitions

Iyer, Mahesh S.; Rhinehart, R. Russell

doi:10.1109/72.750573

Cited by 93 publications

(46 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…2, as the median performance for any given river is relatively constant with respect to the number of hidden neurons. However, the performance varies considerably from one replicate to the other, exemplifying the risk of stand-alone models, as mentioned by Iyer and Rhinehart (1999). If exception is made of models with isolated extremely low performance, this variability tends to increase with the number of hidden neurons for most rivers in Fig.…”

Section: Stand-alone Modelsmentioning

confidence: 99%

Generalisation for neural networks through data sampling and training procedures, with applications to streamflow predictions

Anctil

Lauzon

2004

Hydrol. Earth Syst. Sci.

View full text Add to dashboard Cite

Since the 1990s, neural networks have been applied to many studies in hydrology and water resources. Extensive reviews on neural network modelling have identified the major issues affecting modelling performance; one of the most important is generalisation, which refers to building models that can infer the behaviour of the system under study for conditions represented not only in the data employed for training and testing but also for those conditions not present in the data sets but inherent to the system. This work compares five generalisation approaches: stop training, Bayesian regularisation, stacking, bagging and boosting. All have been tested with neural networks in various scientific domains; stop training and stacking having been applied regularly in hydrology and water resources for some years, while Bayesian regularisation, bagging and boosting have been less common. The comparison is applied to streamflow modelling with multi-layer perceptron neural networks and the Levenberg-Marquardt algorithm as training procedure. Six catchments, with diverse hydrological behaviours, are employed as test cases to draw general conclusions and guidelines on the use of the generalisation techniques for practitioners in hydrology and water resources. All generalisation approaches provide improved performance compared with standard neural networks without generalisation. Stacking, bagging and boosting, which affect the construction of training sets, provide the best improvement from standard models, compared with stop-training and Bayesian regularisation, which regulate the training algorithm. Stacking performs better than the others although the benefit in performance is slight compared with bagging and boosting; furthermore, it is not consistent from one catchment to another. For a good combination of improvement and stability in modelling performance, the joint use of stop training or Bayesian regularisation with either bagging or boosting is recommended.

show abstract

Section: Stand-alone Modelsmentioning

confidence: 99%

Generalisation for neural networks through data sampling and training procedures, with applications to streamflow predictions

Anctil

Lauzon

2004

Hydrol. Earth Syst. Sci.

View full text Add to dashboard Cite

show abstract

“…Sixty percent of the data set was selected for the training set and the remainder only for the test set (the classification evaluation) that is never used for training the networks. The network training was repeated 15 times, and then the network producing the best accuracy was selected; this implies that the retained model resides within the best 14% of the distribution of all possible models (99% confidence) [35].…”

Section: Discussionmentioning

confidence: 99%

Multilayer Perceptron Neural Networks Model for Meteosat Second Generation SEVIRI Daytime Cloud Masking

et al. 2015

View full text Add to dashboard Cite

A multilayer perceptron neural network cloud mask for Meteosat Second Generation SEVIRI (Spinning Enhanced Visible and Infrared Imager) images is introduced and evaluated. The model is trained for cloud detection on MSG SEVIRI daytime data. It consists of a multi-layer perceptron with one hidden sigmoid layer, trained with the error back-propagation algorithm. The model is fed by six bands of MSG data (0.6, 0.8, 1.6, 3.9, 6.2 and 10.8 μm) with 10 hidden nodes. The multiple-layer perceptrons lead to a cloud detection accuracy of 88.96%, when trained to map two predefined values that classify cloud and clear sky. The network was further evaluated using sixty MSG images taken at different dates. The network detected not only bright thick clouds but also thin or less bright clouds. The analysis demonstrated the feasibility of using machine learning models of cloud detection in MSG SEVIRI imagery.

show abstract

“…In the first case, training should just be restarted with random weights, while in the second case, the number of hidden nodes should be increased -otherwise training may not converge. Taking a probabilistic approach to this problem, one can estimate the number of training attempts that should be made before one establishes the inability of the training procedure to converge for a given network state by computing [8]:…”

Section: Adaptive Architecturementioning

confidence: 99%

A Method to Extend Functionality of Pointer Input Devices

Gusyatin

Urinson

Betke

2004

User-Centered Interaction Paradigms for Universal Access in the Information Society

View full text Add to dashboard Cite

Abstract. We describe a general method for extending any pointer input device with an arbitrary set of commands. The proposed interface can be trained by the user to recognize certain cursor movement patterns and interpret them as special input events. Methods for extraction and recognition of such patterns are general enough to work with low-precision pointing devices, and they can be adjusted to provide computer access for people with disabilities. The core of the system is a trainable classifier, in the current implementation an artificial neural network. The architecture of the neural network automatically adjusts according to complexity of the classification task. The system demonstrated good accuracy and responsiveness during extensive experiments. Some tests included a severely motion-impaired individual.

show abstract

A method to determine the required number of neural-network training repetitions

Cited by 93 publications

References 13 publications

Generalisation for neural networks through data sampling and training procedures, with applications to streamflow predictions

Generalisation for neural networks through data sampling and training procedures, with applications to streamflow predictions

Multilayer Perceptron Neural Networks Model for Meteosat Second Generation SEVIRI Daytime Cloud Masking

A Method to Extend Functionality of Pointer Input Devices

Contact Info

Product

Resources

About