arXiv subjects

Patrick Xia

Publications and source records attributed to Patrick Xia.

27 records · Page 2Linked to original sources

arXiv2020

Which *BERT? A Survey Organizing Contextualized Encoders

Pretrained contextualized text encoders are now a staple of the NLP community. We present a survey on language representation learning with the aim of consolidating a series of shared lessons learned across a variety of recent efforts. While significant advancements continue at a rapid pace, we find that enough has now been discovered, in different directions, that we can begin to organize advances according to common themes. Through this organization, we highlight important considerations when interpreting recent contributions and choosing which model to use.

cs.CL↗

arXiv2020

Multi-Sentence Argument Linking

We present a novel document-level model for finding argument spans that fill an event's roles, connecting related ideas in sentence-level semantic role labeling and coreference resolution. Because existing datasets for cross-sentence linking are small, development of our neural model is supported through the creation of a new resource, Roles Across Multiple Sentences (RAMS), which contains 9,124 annotated events across 139 types. We demonstrate strong performance of our model on RAMS and other event-related datasets.

cs.CL↗

arXiv2020

UniMorph 2.0: Universal Morphology

The Universal Morphology UniMorph project is a collaborative effort to improve how NLP handles complex morphology across the world's languages. The project releases annotated morphological data using a universal tagset, the UniMorph schema. Each inflected form is associated with a lemma, which typically carries its underlying lexical meaning, and a bundle of morphological features from our schema. Additional supporting data and tools are also released on a per-language basis when available. UniMorph is based at the Center for Language and Speech Processing (CLSP) at Johns Hopkins University in Baltimore, Maryland and is sponsored by the DARPA LORELEI program. This paper details advances made to the collection, annotation, and dissemination of project resources since the initial UniMorph release described at LREC 2016. lexical resources} }

cs.CL↗

arXiv2019

Probing What Different NLP Tasks Teach Machines about Function Word Comprehension

We introduce a set of nine challenge tasks that test for the understanding of function words. These tasks are created by structurally mutating sentences from existing datasets to target the comprehension of specific types of function words (e.g., prepositions, wh-words). Using these probing tasks, we explore the effects of various pretraining objectives for sentence encoders (e.g., language modeling, CCG supertagging and natural language inference (NLI)) on the learned representations. Our results show that pretraining on language modeling performs the best on average across our probing tasks, supporting its widespread use for pretraining state-of-the-art NLP models, and CCG supertagging and NLI pretraining perform comparably. Overall, no pretraining objective dominates across the board, and our function word probing tasks highlight several intuitive differences between pretraining objectives, e.g., that NLI helps the comprehension of negation.

cs.CL↗

arXiv2019

Can You Tell Me How to Get Past Sesame Street? Sentence-Level Pretraining Beyond Language Modeling

Natural language understanding has recently seen a surge of progress with the use of sentence encoders like ELMo (Peters et al., 2018a) and BERT (Devlin et al., 2019) which are pretrained on variants of language modeling. We conduct the first large-scale systematic study of candidate pretraining tasks, comparing 19 different tasks both as alternatives and complements to language modeling. Our primary results support the use language modeling, especially when combined with pretraining on additional labeled-data tasks. However, our results are mixed across pretraining tasks and show some concerning trends: In ELMo's pretrain-then-freeze paradigm, random baselines are worryingly strong and results vary strikingly across target tasks. In addition, fine-tuning BERT on an intermediate task often negatively impacts downstream transfer. In a more positive trend, we see modest gains from multitask training, suggesting the development of more sophisticated multitask and transfer learning techniques as an avenue for further research.

cs.CL↗

arXiv2019

What do you learn from context? Probing for sentence structure in contextualized word representations

Contextualized representation models such as ELMo (Peters et al., 2018a) and BERT (Devlin et al., 2018) have recently achieved state-of-the-art results on a diverse array of downstream NLP tasks. Building on recent token-level probing work, we introduce a novel edge probing task design and construct a broad suite of sub-sentence tasks derived from the traditional structured NLP pipeline. We probe word-level contextual representations from four recent models and investigate how they encode sentence structure across a range of syntactic, semantic, local, and long-range phenomena. We find that existing models trained on language modeling and translation produce strong representations for syntactic phenomena, but only offer comparably small improvements on semantic tasks over a non-contextual baseline.

cs.CL↗

arXiv2017

CoNLL-SIGMORPHON 2017 Shared Task: Universal Morphological Reinflection in 52 Languages

The CoNLL-SIGMORPHON 2017 shared task on supervised morphological generation required systems to be trained and tested in each of 52 typologically diverse languages. In sub-task 1, submitted systems were asked to predict a specific inflected form of a given lemma. In sub-task 2, systems were given a lemma and some of its specific inflected forms, and asked to complete the inflectional paradigm by predicting all of the remaining inflected forms. Both sub-tasks included high, medium, and low-resource conditions. Sub-task 1 received 24 system submissions, while sub-task 2 received 3 system submissions. Following the success of neural sequence-to-sequence models in the SIGMORPHON 2016 shared task, all but one of the submissions included a neural component. The results show that high performance can be achieved with small training datasets, so long as models have appropriate inductive bias or make use of additional unlabeled data or synthetic data. However, different biasing and data augmentation resulted in disjoint sets of inflected forms being predicted correctly, suggesting that there is room for future improvement.

cs.CL↗

arXiv2015

Annotating Character Relationships in Literary Texts

We present a dataset of manually annotated relationships between characters in literary texts, in order to support the training and evaluation of automatic methods for relation type prediction in this domain (Makazhanov et al., 2014; Kokkinakis, 2013) and the broader computational analysis of literary character (Elson et al., 2010; Bamman et al., 2014; Vala et al., 2015; Flekova and Gurevych, 2015). In this work, we solicit annotations from workers on Amazon Mechanical Turk for 109 texts ranging from Homer's _Iliad_ to Joyce's _Ulysses_ on four dimensions of interest: for a given pair of characters, we collect judgments as to the coarse-grained category (professional, social, familial), fine-grained category (friend, lover, parent, rival, employer), and affinity (positive, negative, neutral) that describes their primary relationship in a text. We do not assume that this relationship is static; we also collect judgments as to whether it changes at any point in the course of the text.

cs.CL↗

arXiv2013

Polar Codes: Speed of polarization and polynomial gap to capacity

We prove that, for all binary-input symmetric memoryless channels, polar codes enable reliable communication at rates within $ε> 0$ of the Shannon capacity with a block length, construction complexity, and decoding complexity all bounded by a {\em polynomial} in $1/ε$. Polar coding gives the {\em first known explicit construction} with rigorous proofs of all these properties; previous constructions were not known to achieve capacity with less than $\exp(1/ε)$ decoding complexity except for erasure channels. We establish the capacity-achieving property of polar codes via a direct analysis of the underlying martingale of conditional entropies, without relying on the martingale convergence theorem. This step gives rough polarization (noise levels $\approx ε$ for the "good" channels), which can then be adequately amplified by tracking the decay of the channel Bhattacharyya parameters. Our effective bounds imply that polar codes can have block length (and encoding/decoding complexity) bounded by a polynomial in $1/ε$. The generator matrix of such polar codes can be constructed in polynomial time by algorithmically computing an adequate approximation of the polarization process.

cs.IT↗