A Grammarless Language Generation Algorithm Based on Idiotypic Artificial Immune Networks
- 1.4k Downloads
The immune system is capable of evolving by learning from its environment over the lifetime of the host. Using the ideas of idiotypic network theory and artificial immune systems we explore the analogy between the immune system and linguistics to suggest a new approach to build a network of sentence phrases and train it using a learning algorithm. The learning algorithm is devised to help evolve the network sufficiently with stimulations by correct phrases or antigens. The network after sufficient stimulations, suppressions and decay is capable of detecting and differentiating between correct and wrong sentences. We verify with experimental data and observe promising results for such an immune network based algorithm. The system learns a language without any grammar rules similar to a small child who knows nothing about grammar yet learns to speak in his native language fluently after a few years of training.
Keywordsartificial immune system idiotypic network theory natural language generation grammarless language learning
Unable to display preview. Download preview PDF.
- 1.Araujo, L., Santamaria, J.: Evolving natural language grammars without supervision. In: IEEE Congress on Evolutionary Computation 2010, pp. 1–8 (2010)Google Scholar
- 2.Jerne, N.: The generative grammar of the immune system. Nobel lecture (December 1984)Google Scholar
- 4.Jerne, N.: Towards a network theory of the immune system. Ann. Immunol (Inst. Pasteur) 125, 373 (1974)Google Scholar
- 6.de Castro, L.N., Timmis, J.: Artificial Immune Systems: A New Computational Intelligence Approach, vol. XVIII, p. 357. Springer, Heidelberg (2002)Google Scholar
- 7.Bonna, C.A., Kohler, H.: Immune Networks. Annals of the New York Academy of Sciences (1983)Google Scholar
- 9.Plett, E., Das, S.: A new algorithm based on negative selection and idiotypic networks for generating parsimonious detector sets for industrial fault detection applications. In: Andrews, P.S., Timmis, J., Owens, N.D.L., Aickelin, U., Hart, E., Hone, A., Tyrrell, A.M. (eds.) ICARIS 2009. LNCS, vol. 5666, pp. 288–300. Springer, Heidelberg (2009)CrossRefGoogle Scholar
- 10.de Castro, L.N., Zuben, F.J.V.: Ainet: An artificial immune network for data analysis. Springer, London (2002)Google Scholar
- 12.Hajela, P., Yoo, J.S.: Immune Network Modelling in Design Optimization. In: Corne, D., Dorigo, M., Glover, F. (eds.) New Ideas in Optimization, McGraw Hill, London (1999)Google Scholar
- 13.Hart, E., Ross, P.: The Evolution and Analysis of a Potential Antibody Library for Use in Job-Shop Scheduling. In: Corne, D., Dorigo, M., Glover, F. (eds.) New Ideas in Optimization. McGraw Hill, London (1999)Google Scholar