Work

Dual Decomposition with Many Overlapping Components

Year: 2018

Type: article

Abstract: Dual decomposition has been recently proposed as a way of combining complementary models, with a boost in predictive power. However, in cases where lightweight decompositions are not readily available (e.g., due to the presence of rich features or logical constraints), the original subgradient algorithm is inefficient. We sidestep that difficulty by adopting an augmented Lagrangian method that accelerates model consensus by regularizing towards the averaged votes. We show how first-order logical constraints can be handled efficiently, even though the corresponding subproblems are no longer combinatorial, and report experiments in dependency parsing, with state-of-the-art results.

Source: Figshare

DOI: 10.1184/r1/6473336

Authors: Martins, Andre F.T., A. Smith, Noah, Aguiar, Pedro M.Q., Figueiredo, Mario A. T.

Institutions: Carnegie Mellon University, Instituto de Telecomunicações, Universidade Técnica de Lisboa

Language: English

Topic: Natural Language Processing Techniques

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: green

Funders: National Science Foundation, Carnegie Mellon Portugal, European Commission, Fundação para a Ciência e a Tecnologia

Awards: CAREER: Flexible Learning for Natural Language Processing, G6671297155, Beyond Features: Similarity-Based Pattern Analysis and Recognition