|
| This Article | ||
| ||
| Share | ||
| Bibliographic References | ||
| Add to: | ||
 Digg  Furl  Spurl  Blink  Simpy  Google  Del.icio.us  Y!MyWeb | ||
| Search | ||
| ||
| ASCII Text | x | ||
| Katia S. Guimar˜es, Jeane C. B. Melo, George D. C. Cavalcanti, "Combining Few Neural Networks for Effective Secondary Structure Prediction," 2012 IEEE 12th International Conference on Bioinformatics & Bioengineering (BIBE), pp. 415, Third IEEE Symposium on BioInformatics and BioEngineering (BIBE'03), 2003. | |||
| BibTex | x | ||
| @article{ 10.1109/BIBE.2003.1188981, author = {Katia S. Guimar˜es and Jeane C. B. Melo and George D. C. Cavalcanti}, title = {Combining Few Neural Networks for Effective Secondary Structure Prediction}, journal ={2012 IEEE 12th International Conference on Bioinformatics & Bioengineering (BIBE)}, volume = {0}, year = {2003}, isbn = {0-7695-1907-5}, pages = {415}, doi = {http://doi.ieeecomputersociety.org/10.1109/BIBE.2003.1188981}, publisher = {IEEE Computer Society}, address = {Los Alamitos, CA, USA}, } | |||
| RefWorks Procite/RefMan/Endnote | x | ||
| TY - CONF JO - 2012 IEEE 12th International Conference on Bioinformatics & Bioengineering (BIBE) TI - Combining Few Neural Networks for Effective Secondary Structure Prediction SN - 0-7695-1907-5 SP EP A1 - Katia S. Guimar˜es, A1 - Jeane C. B. Melo, A1 - George D. C. Cavalcanti, PY - 2003 KW - null VL - 0 JA - 2012 IEEE 12th International Conference on Bioinformatics & Bioengineering (BIBE) ER - | |||
The prediction of secondary structure is treated with a simple and efficient method. Combining only three neural networks, an average Q3 accuracy prediction by residues of 75.93% is achieved. This value is better than the best result reported on the same test and training database, CB396, using the same validation method. For a second database, RS126, an average Q3 accuracy of 74.13% is attained, which is better than each individual method, being defeated only by CONSENSUS, a rather intrincate engine, which is a combination of several methods.
The networks are trained with RPROP, an efficient variation of the back-propagation algorithm. Five combination rules are applied independently afterwards. Each one increases the accuracy of prediction by at least 1%, due to the fact that each network used converges to a different local minimum. The Product rule derives the best results.