Generalising the nodes of the error propagation network

Robinson, Stewart; Niranjan,; Fallside,

doi:10.1109/ijcnn.1989.118343

Cited by 25 publications

(30 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For instance, the viability of two RBF layers in cascade (replacing the linear associator between hidden and output layers of an RBFN with another Euclidean/ Gaussian layer) has been shown (Robinson et al, 1988;Dorffner, 1992). Also, the combination of MLP and RBF units in one Downloaded by [University of California, San Diego] at 12:31 29 June 2016 hidden layer can lead to improved results (Weymaere, 1992).…”

Section: Previous Attempts On Unificationmentioning

confidence: 96%

“…Important examples of such cross-fertilizations are the use of gradient descent for RBFNs (Robinson et al, 1988;Weymaere, 1992;Dorffner, 1992) and the use of initialization + delta for MLPs (Smyth 1992;Weymaere, 1992). The former is a viable way of fine-tuning the centers and/or widths of RBFs, and the latter has proved to improve speed (Smyth, 1992) or even performance (Dorffner and Porenta, 1994) of MLPs (explained in more detail later).…”

Section: Previous Attempts On Unificationmentioning

confidence: 99%

See 1 more Smart Citation

UNIFIED FRAMEWORK FOR MLPs AND RBFNs: INTRODUCING CONIC SECTION FUNCTION NETWORKS

Dorffner

1994

Cybernetics and Systems

View full text Add to dashboard Cite

Section: Previous Attempts On Unificationmentioning

confidence: 96%

Section: Previous Attempts On Unificationmentioning

confidence: 99%

UNIFIED FRAMEWORK FOR MLPs AND RBFNs: INTRODUCING CONIC SECTION FUNCTION NETWORKS

Dorffner

1994

Cybernetics and Systems

View full text Add to dashboard Cite

“…Momentum is the addition of a smoothing term to the weight updating equation in order to reinforce weight changes that occur in a consistent direct.ion through weight space. The other is lea.rning ra.te a,dapta.tion of t.he style utilized in [5]. If t,he t~otal squared error for all pat.terns in the training set decreased from one epoch to the next, then the learning rate is increased by a fixed factor (here 0.1%).…”

Section: Neural Network Modelsmentioning

confidence: 99%

“…On the other hand, the third type uses generalized error propagation [5], a generalization of back propagation, t.0 adjust the parameters of the Gaussian hidden units.…”

Section: Neural Network Modelsmentioning

confidence: 99%

See 1 more Smart Citation

Neural networks for classification of ARMA models: an experimental study

McKee¹,

Moura²

[1991 Proceedings] IEEE Conference on Neural Networks for Ocean Engineering

View full text Add to dashboard Cite

This paper presents a set of extensive experiments with alternative neural network learning algorithms. These neural network configurations were tested on the problem of discriminating signals generated by autoregressive moving-average ( ARMA) linear systems driven by white noise. These ARMA signals model a wide variety of signals arising in the ocean environment. We tested the various network models for their classification accuracy and speed of learning. The models investigated were back propagation, quickprop, Gaussian node networks, radial basis functions, the modified Kanerva method, and networks without hidden units. For comparison, nearest-neighbor classifiers were also tested. Classification performance and learning time results are presented.

show abstract

Research on multiscale features and integrated forecasting of the Belt and Road exchange rate index

Yixi

Liu

Li³

et al. 2020

Concurrency and Computation

View full text Add to dashboard Cite

Massive progress and fruitful results have been achieved since the implementation of the Belt and Road Initiative (B&R). However, the risk of exchange rate between countries along the Belt and Road cannot be ignored. In this paper, we propose a B&R exchange rate index to evaluate the currency risk of countries along the B&R, along with a research on the multiscale features of the B&R exchange rate index and the effective RMB exchange rate index using the EMD algorithm. Then, we propose an integrated forecasting approach. We adopt the EMD algorithm and the Grey Relational Analysis to decompose and reconstruct the B&R exchange rate index and the effective RMB exchange rate index into three subsections with different volatility characteristics. The prediction outcome of the integrated model for the original series equals a sum of the Auto-regressive Integrated Moving Average model for trend term and the Long Short-Term Memory model for market fluctuation term as well as noise term. The results of our prediction demonstrate a much better performance of our proposed integrated forecasting model than pure methods. Lastly, several suggestions are put forward to help relevant enterprises effectively avoid and manage the exchange risks. K E Y W O R D S belt and road exchange rate index, EMD algorithm, integrated forecasting model, multiscale features 1 INTRODUCTION According to BELT AND ROAD PORTAL 1 , as of April 2019, China has signed 173 cooperation documents with 125 countries along the Belt and Road (B&R), expanding the scope from the Asia-Europe region originally to Africa, Latin America, South Pacific, Western Europe and other countries gradually. Besides, from 2013 to 2018, The total trade volume between China and regions along the route has exceeded 6 trillion U.S. dollars, according to the National Development and Reform Commission. In terms of the Ministry of commerce, 15.64 billion dollars has been invested by Chinese enterprises in non-financial direct investments in 56 countries along the B&R, up 8.9% year on year, taking up 13% of the total amount of investment over the same period; 7721 foreign contractual projects along the B&R have been newly signed, with a total value of 125.78 billion dollars; the total turnover was up to 89.33 billion dollars with a growth of 4.4% over the same period in 2017, which accounted for 52.8% of the total amount in 2018. However, the exchange rate fluctuations of currencies around the "Belt and Road" area are usually higher than that of mainstream currencies such as the U.S. dollar and the Euro. One reason is that the exchange rate systems of these countries are mainly soft pegged and floating, especially in Europe and Africa, which makes these countries' currencies more sensitive to the impact of uncertain market factors. Another reason is that the economic development in countries and regions along the B&R varies greatly, making the country risk relatively high. 1 https://eng.yidaiyilu.gov.cn/

show abstract

Generalising the nodes of the error propagation network

Cited by 25 publications

References 0 publications

UNIFIED FRAMEWORK FOR MLPs AND RBFNs: INTRODUCING CONIC SECTION FUNCTION NETWORKS

UNIFIED FRAMEWORK FOR MLPs AND RBFNs: INTRODUCING CONIC SECTION FUNCTION NETWORKS

Neural networks for classification of ARMA models: an experimental study

Research on multiscale features and integrated forecasting of the Belt and Road exchange rate index

Contact Info

Product

Resources

About