A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
Filters
Modeling Diagnostic Label Correlation for Automatic ICD Coding
[article]
2021
arXiv
pre-print
Preserved Fulltext
This paper is the first attempt at learning the label set distribution as a reranking module for medical code prediction. ...
Given the clinical notes written in electronic health records (EHRs), it is challenging to predict the diagnostic codes which is formulated as a multi-label classification task. ...
Multi-Label Classification Multi-label classification problems are of broad interest to the machine learning community. The goal is to predict a subset of labels associated with a given object. ...
arXiv:2106.12800v1
fatcat:ejzatbw6yve7jmukn67bkmudky
Revisiting Calibration for Question Answering
[article]
2022
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
We examine various conventional calibration methods including temperature scaling, feature-based classifier, neural answer reranking, and label smoothing, all of which do not bring significant gains under ...
Building on those observations, we propose a new calibration metric, MacroCE, that better captures whether the model assigns low confidence to wrong predictions and high confidence to correct predictions ...
During inference, we use the trained reranker to rerank the top five predictions. ...
arXiv:2205.12507v1
fatcat:vl5mpwasnjbwbnp2juhf63wmue
An empirical evaluation of supervised learning approaches in assigning diagnosis codes to electronic medical records
2015
Artificial Intelligence in Medicine
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2017; you can also visit the original URL.
The file type is application/pdf.
label calibration to select the optimal number of labels is an indispensable final step. ...
This is a necessary and complex task involving coders adhering to coding guidelines and coding all assignable codes. ...
Acknowledgements Many thanks to anonymous reviewers for their detailed comments and suggestions that helped improve the content and presentation of this paper. ...
doi:10.1016/j.artmed.2015.04.007
pmid:26054428
pmcid:PMC4605853
fatcat:sxqh2dxh35dntcaw2ayvu4e3se
Using calibrator to improve robustness in Machine Reading Comprehension
[article]
2022
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
The calibrator combines both manual features and representation learning features to rerank candidate results. ...
In this paper, we propose a method to improve the robustness by using a calibrator as the post-hoc reranker, which is implemented based on XGBoost model. ...
We propose to use the calibrator as a reranker to choose the best, so it is a multi-classification problem rather than binary classification. ...
arXiv:2202.11865v1
fatcat:jvqv7waq7vfoji4tv6d5gqhenu
RoR: Read-over-Read for Long Document Machine Reading Comprehension
[article]
2021
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
Other Versions
The latter further predicts the global answers from this condensed document. Eventually, a voting strategy is utilized to aggregate and rerank the regional and global answers for final prediction. ...
To address this problem, we propose RoR, a read-over-read method, which expands the reading field from chunk to document. Specifically, RoR includes a chunk reader and a document reader. ...
Acknowledge We would like to thank three anonymous reviewers for their useful feedback. We also sincerely appreciate He He and Mark Yatskar for their help in evaluating our models on QuAC dataset. ...
arXiv:2109.04780v2
fatcat:kuehwaqikrbq3fpqdffwumri5y
PAQ: 65 Million Probably-Asked Questions and What You Can Do With Them
[article]
2021
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
to conventional models which retrieve and read from text corpora. ...
This enables RePAQ to "back-off" to a more expensive state-of-the-art model, leading to a combined system which is both more accurate and 2x faster than the state-of-the-art model alone. ...
Learning the embedding function g q is complicated by the lack of labelled question pair paraphrases in ODQA datasets. ...
arXiv:2102.07033v1
fatcat:6yik3pxksff77kkb26a54ggpmy
PAQ: 65 Million Probably-Asked Questions and What You Can Do With Them
2021
Transactions of the Association for Computational Linguistics
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
This enables RePAQ to "back-off" to a more expensive state-of-the-art model, leading to a combined system which is both more accurate and 2x faster than the state-of-the-art model alone. ...
We find that PAQ preempts and caches test questions, enabling RePAQ to match the accuracy of recent retrieve-and-read models, whilst being significantly faster. ...
Acknowledgments The authors would like to extend their gratitude to the anonymous reviewers and Action Editor for their highly detailed and insightful comments and feedback. ...
doi:10.1162/tacl_a_00415
fatcat:aqhh5yis6vchlankjmf2xbyzsm
Stretching Sentence-pair NLI Models to Reason over Long Documents and Clusters
[article]
2022
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
Finally, we go further and investigate whole document clusters to identify both discrepancies and consensus among sources. ...
First, we analyze the robustness of these models to longer and out-of-domain inputs. ...
We also thank Sumit Sanghai, Annie Louis, Roee Aharoni, Yi Tay, Kai Hui, Jai Gupta, Vinh Tran, and Dara Bahri for helpful conversations and feedback. Dracula: searching for consensus. ...
arXiv:2204.07447v1
fatcat:6pa3b6a55bedzednwrnq5s7nwu
Multilingual Lexicalized Constituency Parsing with Word-Level Auxiliary Tasks
2017
Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 2, Short Papers
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2018; you can also visit the original URL.
The file type is application/pdf.
We model two important interfaces of constituency parsing with auxiliary tasks supervised at the word level: (i) part-of-speech (POS) and morphological tagging, (ii) functional label prediction. ...
On the SPMRL dataset, our parser obtains above state-of-the-art results on constituency parsing without requiring either predicted POS or morphological tags, and outputs labelled dependency trees. ...
Acknowledgments We thank Djamé Seddah, Héctor Martínez Alonso and Chloé Braud for helpful comments. ...
doi:10.18653/v1/e17-2053
dblp:conf/eacl/CrabbeC17a
fatcat:im3qdw3avzhgpfwxqxo72dped4
Making Heads and Tails of Models with Marginal Calibration for Sparse Tagsets
[article]
2021
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
We address the open problem of calibration for tagging models with sparse tagsets, and recommend strategies to measure and reduce calibration error (CE) in such models. ...
For interpreting the behavior of a probabilistic model, it is useful to measure a model's calibration--the extent to which it produces reliable confidence scores. ...
Acknowledgements We are grateful to anonymous reviewers as well as members of the NERT lab for their feedback on this work. ...
arXiv:2109.07494v1
fatcat:kpxu5vpdfne5loqwja6nij3l3a
2021 Index IEEE Transactions on Knowledge and Data Engineering Vol. 33
2022
IEEE Transactions on Knowledge and Data Engineering
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
Departments and other items may also be covered if they have been judged to have archival value. The Author Index contains the primary entry for each item, listed under the first author's name. ...
The primary entry includes the coauthors' names, the title of the paper or other item, and its location, specified by the publication abbreviation, year, month, and inclusive pagination. ...
., +, TKDE June 2021 2573-2587 Learning to Recommend With Multiple Cascading Behaviors. Gao, C., +, TKDE June 2021 2588-2601 Learning to Rerank Schema Matches. ...
doi:10.1109/tkde.2021.3128365
fatcat:4m5kefreyrbhpb3lhzvgqzm3qu
CODER: An efficient framework for improving retrieval through COntextual Document Embedding Reranking
[article]
2022
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
Other Versions
We investigate the effect of training through contextual reranking of document embeddings and show that our approach leads to substantial improvement in retrieval performance over pair-wise scoring of ...
It employs a list-wise loss and jointly scores a large set of retrieved candidate documents, rather than randomly sampled documents, for each query. ...
) relevance labels y ∈ R N , defined for the set of N candidate documents (where the relevance of all documents not explicitly defined is assumed to be 0), and a distribution over the predicted scores ...
arXiv:2112.08766v2
fatcat:zw4ssapcnje4lmlpmmkdfulwrq
Strong Baselines for Neural Semi-Supervised Learning under Domain Shift
2018
Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
In this paper, we re-evaluate classic general-purpose bootstrapping approaches in the context of neural networks under domain shifts vs. recent neural approaches and propose a novel multi-task tri-training ...
Novel neural models have been proposed in recent years for learning under domain shift. ...
Barbara is supported by NVIDIA corporation and thanks the Computing Center of the University of Groningen for HPC support. ...
doi:10.18653/v1/p18-1096
dblp:conf/acl/PlankR18
fatcat:mczex3hplzfpxo5yf3zlz2og5y
Strong Baselines for Neural Semi-supervised Learning under Domain Shift
[article]
2018
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
In this paper, we re-evaluate classic general-purpose bootstrapping approaches in the context of neural networks under domain shifts vs. recent neural approaches and propose a novel multi-task tri-training ...
Novel neural models have been proposed in recent years for learning under domain shift. ...
Barbara is supported by NVIDIA corporation and thanks the Computing Center of the University of Groningen for HPC support. ...
arXiv:1804.09530v1
fatcat:lplej4gewrbn3c4ogpop3hrqt4
Multi-Camera Vehicle Tracking with Powerful Visual Features and Spatial-Temporal Cue
2019
Computer Vision and Pattern Recognition
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Vehicle re-identification and multi-camera multi-object vehicle tracking are important components in the field of intelligent traffic, which is attracting more and more attention. ...
Besides, spatial-temporal cue is fully excavated to make up the deficiency of appearance feature and constrained hierarchical clustering is introduced into the pipeline to get the final cluster results ...
Then the cross entropy loss with label smooth can be computed as: L(ID) = N i=1 −q i log(p i ) (2) where p i is the ID prediction logits of class i. Clustering Loss. ...
dblp:conf/cvpr/HeLBW19
fatcat:7x37nr7whvhchjenjqo55556yu
« Previous
Showing results 1 — 15 out of 158 results