"Show me the cup": Reference with Continuous Representations. Boleda, G., Padó, S., & Baroni, M. arXiv:1606.08777 [cs], June, 2016. arXiv: 1606.08777Paper abstract bibtex One of the most basic functions of language is to refer to objects in a shared scene. Modeling reference with continuous representations is challenging because it requires individuation, i.e., tracking and distinguishing an arbitrary number of referents. We introduce a neural network model that, given a definite description and a set of objects represented by natural images, points to the intended object if the expression has a unique referent, or indicates a failure, if it does not. The model, directly trained on reference acts, is competitive with a pipeline manually engineered to perform the same task, both when referents are purely visual, and when they are characterized by a combination of visual and linguistic properties.
@article{boleda_show_2016,
title = {"{Show} me the cup": {Reference} with {Continuous} {Representations}},
shorttitle = {"{Show} me the cup"},
url = {http://arxiv.org/abs/1606.08777},
abstract = {One of the most basic functions of language is to refer to objects in a shared scene. Modeling reference with continuous representations is challenging because it requires individuation, i.e., tracking and distinguishing an arbitrary number of referents. We introduce a neural network model that, given a definite description and a set of objects represented by natural images, points to the intended object if the expression has a unique referent, or indicates a failure, if it does not. The model, directly trained on reference acts, is competitive with a pipeline manually engineered to perform the same task, both when referents are purely visual, and when they are characterized by a combination of visual and linguistic properties.},
urldate = {2017-03-30},
journal = {arXiv:1606.08777 [cs]},
author = {Boleda, Gemma and Padó, Sebastian and Baroni, Marco},
month = jun,
year = {2016},
note = {arXiv: 1606.08777},
keywords = {Computer Science - Artificial Intelligence, Computer Science - Computation and Language, Computer Science - Learning},
}
Downloads: 0
{"_id":"SmwrEiBrx7banywqE","bibbaseid":"boleda-pad-baroni-showmethecupreferencewithcontinuousrepresentations-2016","downloads":0,"creationDate":"2017-04-01T11:49:38.717Z","title":"\"Show me the cup\": Reference with Continuous Representations","author_short":["Boleda, G.","Padó, S.","Baroni, M."],"year":2016,"bibtype":"article","biburl":"https://api.zotero.org/groups/1152176/items?key=bi2Q7duoPuqjf6lgym4TgM83&format=bibtex&limit=100","bibdata":{"bibtype":"article","type":"article","title":"\"Show me the cup\": Reference with Continuous Representations","shorttitle":"\"Show me the cup\"","url":"http://arxiv.org/abs/1606.08777","abstract":"One of the most basic functions of language is to refer to objects in a shared scene. Modeling reference with continuous representations is challenging because it requires individuation, i.e., tracking and distinguishing an arbitrary number of referents. We introduce a neural network model that, given a definite description and a set of objects represented by natural images, points to the intended object if the expression has a unique referent, or indicates a failure, if it does not. The model, directly trained on reference acts, is competitive with a pipeline manually engineered to perform the same task, both when referents are purely visual, and when they are characterized by a combination of visual and linguistic properties.","urldate":"2017-03-30","journal":"arXiv:1606.08777 [cs]","author":[{"propositions":[],"lastnames":["Boleda"],"firstnames":["Gemma"],"suffixes":[]},{"propositions":[],"lastnames":["Padó"],"firstnames":["Sebastian"],"suffixes":[]},{"propositions":[],"lastnames":["Baroni"],"firstnames":["Marco"],"suffixes":[]}],"month":"June","year":"2016","note":"arXiv: 1606.08777","keywords":"Computer Science - Artificial Intelligence, Computer Science - Computation and Language, Computer Science - Learning","bibtex":"@article{boleda_show_2016,\n\ttitle = {\"{Show} me the cup\": {Reference} with {Continuous} {Representations}},\n\tshorttitle = {\"{Show} me the cup\"},\n\turl = {http://arxiv.org/abs/1606.08777},\n\tabstract = {One of the most basic functions of language is to refer to objects in a shared scene. Modeling reference with continuous representations is challenging because it requires individuation, i.e., tracking and distinguishing an arbitrary number of referents. We introduce a neural network model that, given a definite description and a set of objects represented by natural images, points to the intended object if the expression has a unique referent, or indicates a failure, if it does not. The model, directly trained on reference acts, is competitive with a pipeline manually engineered to perform the same task, both when referents are purely visual, and when they are characterized by a combination of visual and linguistic properties.},\n\turldate = {2017-03-30},\n\tjournal = {arXiv:1606.08777 [cs]},\n\tauthor = {Boleda, Gemma and Padó, Sebastian and Baroni, Marco},\n\tmonth = jun,\n\tyear = {2016},\n\tnote = {arXiv: 1606.08777},\n\tkeywords = {Computer Science - Artificial Intelligence, Computer Science - Computation and Language, Computer Science - Learning},\n}\n","author_short":["Boleda, G.","Padó, S.","Baroni, M."],"key":"boleda_show_2016","id":"boleda_show_2016","bibbaseid":"boleda-pad-baroni-showmethecupreferencewithcontinuousrepresentations-2016","role":"author","urls":{"Paper":"http://arxiv.org/abs/1606.08777"},"keyword":["Computer Science - Artificial Intelligence","Computer Science - Computation and Language","Computer Science - Learning"],"metadata":{"authorlinks":{}},"downloads":0},"search_terms":["show","cup","reference","continuous","representations","boleda","padó","baroni"],"keywords":["computer science - artificial intelligence","computer science - computation and language","computer science - learning"],"authorIDs":[],"dataSources":["W7n7ZGD7akQhMdBQh"]}