Sciweavers

Free Online Productivity Tools i2Speak i2Symbol i2OCR iTex2Img iWeb2Print iWeb2Shot i2Type iPdf2Split iPdf2Merge i2Bopomofo i2Arabic i2Style i2Image i2PDF iLatex2Rtf Sci2ools

14

IPPS
1998
IEEE

favoriteEmaildiscussreport

93views Distributed And Parallel Com...» more IPPS 1998»

Using the BSP Cost Model to Optimise Parallel Neural Network Training

13 years 8 months ago

Using the BSP Cost Model to Optimise Parallel Neural Network Training

Download ipdps.cc.gatech.edu

We derive cost formulae for three di erent parallelisation techniques for training supervised networks. These formulae are parameterised by properties of the target computer architecture. It is therefore possible to decide the best match between parallel computer and training technique. One technique, exemplar parallelism, is far superior for almost all parallel computer architectures. Formulae also take into account optimal batch learning as the overall training approach.

R. O. Rogers, David B. Skillicorn

Real-time Traffic

Cost Formulae | Distributed And Parallel Computing | Formulae | IPPS 1998 | Training Supervised Networks |

claim paper

Related Content

» Global Optimisation of Neural Network Models via Sequential Sampling

» Minimizing Distribution Cost of Distributed Neural Networks in Wireless Sensor Networks

» Small Models of Large Machines

» Design and Implementation of Parallel Counterpropagation Networks Using MPI

» Fast compiler optimisation evaluation using codefeature based performance prediction

» Parallel Mining of Outliers in Large Database

» Movement Prediction Using Bayesian Learning for Neural Networks

» A Neural Network Model for Interproblem Adaptive Online Time Allocation

» On the use of random neural networks for traffic matrix estimation in largescale IP networ...

Post Info
More Details (n/a)

Added	05 Aug 2010
Updated	05 Aug 2010
Type	Conference
Year	1998
Where	IPPS
Authors	R. O. Rogers, David B. Skillicorn

Comments (0)