no code implementations • 11 Oct 2024 • Xiaofeng Wu, Karl Stratos, Wei Xu
The glyphic writing system of Chinese incorporates information-rich visual features in each character, such as radicals that provide hints about meaning or pronunciation.
1 code implementation • 16 Feb 2024 • Govind Gangadhar, Karl Stratos
Standard fine-tuning is considered not as effective as specialized methods for model editing due to its comparatively poor performance.
1 code implementation • 20 Oct 2023 • Wenzheng Zhang, Sam Wiseman, Karl Stratos
Existing works on coreference resolution suggest that task-specific models are necessary to achieve state-of-the-art performance.
1 code implementation • 1 Jul 2023 • Wenzheng Zhang, Chenyan Xiong, Karl Stratos, Arnold Overwijk
In multitask retrieval, a single retriever is trained to retrieve relevant contexts for multiple tasks.
2 code implementations • ICLR 2022 • Wenzheng Zhang, Wenyue Hua, Karl Stratos
A conventional approach to entity linking is to first find mentions in a given document and then infer their underlying entities in the knowledge base.
Ranked #5 on
Entity Linking
on AIDA-CoNLL
1 code implementation • NAACL 2021 • Wenzheng Zhang, Karl Stratos
The choice of negative examples is important in noise contrastive estimation.
1 code implementation • EACL (Louhi) 2021 • Rajarshi Bhowmik, Karl Stratos, Gerard de Melo
Additionally, we modify our dual encoder model for end-to-end biomedical entity linking that performs both mention span detection and entity disambiguation and out-performs two recently proposed models.
1 code implementation • EMNLP 2021 • Sam Wiseman, Arturs Backurs, Karl Stratos
We propose to tackle data-to-text generation tasks by directly splicing together retrieved segments of text from "neighbor" source-target pairs.
1 code implementation • EMNLP (insights) 2021 • Ozan İrsoy, Adrian Benton, Karl Stratos
Mikolov et al. (2013a) observed that continuous bag-of-words (CBOW) word embeddings tend to underperform Skip-gram (SG) embeddings, and this finding has been reported in subsequent works.
1 code implementation • Findings (ACL) 2021 • Zewei Chu, Karl Stratos, Kevin Gimpel
This reliance causes dataless classifiers to be highly sensitive to the choice of label descriptions and hinders the broader application of dataless classification in practice.
Ranked #3 on
Zero-Shot Text Classification
on AG News
1 code implementation • Findings of the Association for Computational Linguistics 2020 • Mingda Chen, Zewei Chu, Karl Stratos, Kevin Gimpel
Accurate lexical entailment (LE) and natural language inference (NLI) often require large quantities of costly annotations.
1 code implementation • AKBC 2021 • Zewei Chu, Karl Stratos, Kevin Gimpel
We describe NatCat, a large-scale resource for text classification constructed from three data sources: Wikipedia, Stack Exchange, and Reddit.
1 code implementation • ACL 2020 • Shuning Jin, Sam Wiseman, Karl Stratos, Karen Livescu
While much work on deep latent variable models of text uses continuous latent variables, discrete latent variables are interesting because they are more interpretable and typically more space efficient.
1 code implementation • ICML 2020 • Karl Stratos, Sam Wiseman
We propose learning discrete structured representations from unlabeled data by maximizing the mutual information between a structured latent variable and a target variable.
2 code implementations • IJCNLP 2019 • Mingda Chen, Zewei Chu, Yang Chen, Karl Stratos, Kevin Gimpel
Rich entity representations are useful for a wide class of problems involving entities.
1 code implementation • ACL 2019 • Sam Wiseman, Karl Stratos
Retrieve-and-edit based approaches to structured prediction, where structures associated with retrieved neighbors are edited to form new structures, have recently attracted increased interest.
2 code implementations • ICLR 2019 • David McAllester, Karl Stratos
Measuring mutual information from finite data is difficult.
no code implementations • WS 2018 • Daniel Edmiston, Karl Stratos
StAffNet, the name of our architecture, shows competitive performance with the state-of-the-art on this task.
1 code implementation • NAACL 2019 • Karl Stratos
We address part-of-speech (POS) induction by maximizing the mutual information between the induced label and its context.
no code implementations • 16 Jan 2018 • Young-Bum Kim, Sungjin Lee, Karl Stratos
In practice, most spoken language understanding systems process user input in a pipelined manner; first domain is predicted, then intent and semantic slots are inferred according to the semantic frames of the predicted domain.
no code implementations • WS 2017 • Karl Stratos
Pre-trained word embeddings improve the performance of a neural model at the cost of increasing the model size.
1 code implementation • EMNLP 2017 • Karl Stratos
We introduce a novel sub-character architecture that exploits a unique compositional structure of the Korean language.
no code implementations • ACL 2017 • Young-Bum Kim, Karl Stratos, Dongchan Kim
Both cause a distribution mismatch between training and evaluation, leading to a model that overfits the flawed training data and performs poorly on the test data.
no code implementations • ACL 2017 • Young-Bum Kim, Karl Stratos, Dongchan Kim
When given domain K + 1, our model uses a weighted combination of the K domain experts{'} feedback along with its own opinion to make predictions on the new domain.
1 code implementation • WS 2017 • Karl Stratos
Standard approaches in entity identification hard-code boundary detection and type prediction into labels (e. g., John/B-PER Smith/I-PER) and then perform Viterbi.
no code implementations • COLING 2016 • Young-Bum Kim, Karl Stratos, Ruhi Sarikaya
Popular techniques for domain adaptation such as the feature augmentation method of Daum{\'e} III (2009) have mostly been considered for sparse binary-valued features, but not for dense real-valued features such as those used in neural networks.
no code implementations • COLING 2016 • Young-Bum Kim, Karl Stratos, Ruhi Sarikaya
In many applications such as personal digital assistants, there is a constant need for new domains to increase the system{'}s coverage of user queries.
1 code implementation • TACL 2016 • Karl Stratos, Michael Collins, Daniel Hsu
We tackle unsupervised part-of-speech (POS) tagging by learning hidden Markov models (HMMs) that are particularly well-suited for the problem.