Mostra el registre d'ítem simple
Unsupervised spectral learning of FSTs
dc.contributor.author | Bailly, Raphaël |
dc.contributor.author | Carreras Pérez, Xavier |
dc.contributor.author | Quattoni, Ariadna Julieta |
dc.date.accessioned | 2014-08-28T07:59:56Z |
dc.date.available | 2014-08-28T07:59:56Z |
dc.date.created | 2013 |
dc.date.issued | 2013 |
dc.identifier.citation | Bailly, R.; Carreras, X.; Quattoni, A.J. Unsupervised spectral learning of FSTs. A: Neural Information Processing Systems Conference. "Advances in Neural Information Processing Systems 26 (NIPS 2013)". Lake Tahoe, Nevada: 2013, p. 1-13. |
dc.identifier.uri | http://hdl.handle.net/2117/23685 |
dc.description.abstract | Finite-State Transducers (FST) are a standard tool for modeling paired input output sequences and are used in numerous applications, ranging from computational biology to natural language processing. Recently Balle et al. [4] presented a spectral algorithm for learning FST from samples of aligned input-output sequences. In this paper we address the more realistic, yet challenging setting where the alignments are unknown to the learning algorithm. We frame FST learning as finding a low rank Hankel matrix satisfying constraints derived from observable statistics. Under this formulation, we provide identifiability results for FST distributions. Then, following previous work on rank minimization, we propose a regularized convex relaxation of this objective which is based on minimizing a nuclear norm penalty subject to linear constraints and can be solved efficiently. |
dc.format.extent | 13 p. |
dc.language.iso | eng |
dc.subject | Àrees temàtiques de la UPC::Informàtica::Intel·ligència artificial::Llenguatge natural |
dc.subject.lcsh | Learning algorithms |
dc.subject.other | Bioinformatics |
dc.subject.other | Matrix algebra |
dc.subject.other | Natural language processing systems |
dc.subject.other | Relaxation processes |
dc.subject.other | Computational biology |
dc.subject.other | Finite state transducers |
dc.subject.other | Linear constraints |
dc.subject.other | Low-rank Hankel matrixes |
dc.subject.other | Natural language processing |
dc.subject.other | Rank minimizations |
dc.subject.other | Spectral algorithm |
dc.subject.other | Spectral learning |
dc.title | Unsupervised spectral learning of FSTs |
dc.type | Conference report |
dc.subject.lemac | Algorismes d'aprenentage |
dc.contributor.group | Universitat Politècnica de Catalunya. GPLN - Grup de Processament del Llenguatge Natural |
dc.contributor.group | Universitat Politècnica de Catalunya. LARCA - Laboratori d'Algorísmia Relacional, Complexitat i Aprenentatge |
dc.relation.publisherversion | http://papers.nips.cc/paper/4862-unsupervised-spectral-learning-of-finite-state-transducers |
dc.rights.access | Open Access |
local.identifier.drac | 14581218 |
dc.description.version | Postprint (published version) |
local.citation.author | Bailly, R.; Carreras, X.; Quattoni, A.J. |
local.citation.contributor | Neural Information Processing Systems Conference |
local.citation.pubplace | Lake Tahoe, Nevada |
local.citation.publicationName | Advances in Neural Information Processing Systems 26 (NIPS 2013) |
local.citation.startingPage | 1 |
local.citation.endingPage | 13 |