Approximation-Aware Dependency Parsing by Belief                     Propagation

Gormley, Matthew R.; Dredze, Mark; Eisner, Jason

doi:10.1162/tacl_a_00153

Cited by 27 publications

(30 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The message-passing mechanism has been studied in computer software [ 35 ] and NLP [ 36 ]. Recently, Gilmer et al (2017) apply neural message-passing algorithms allowing long-range interactions between nodes in the graph [ 37 ].…”

Section: Related Workmentioning

confidence: 99%

A message-passing multi-task architecture for the implicit event and polarity detection

Xiang

Zhang

2021

PLoS ONE

View full text Add to dashboard Cite

Implicit sentiment analysis is a challenging task because the sentiment of a text is expressed in a connotative manner. To tackle this problem, we propose to use textual events as a knowledge source to enrich network representations. To consider task interactions, we present a novel lightweight joint learning paradigm that can pass task-related messages between tasks during training iterations. This is distinct from previous methods that involve multi-task learning by simple parameter sharing. Besides, a human-annotated corpus with implicit sentiment labels and event labels is scarce, which hinders practical applications of deep neural models. Therefore, we further investigate a back-translation approach to expand training instances. Experiment results on a public benchmark demonstrate the effectiveness of both the proposed multi-task architecture and data augmentation strategy.

show abstract

Section: Related Workmentioning

confidence: 99%

A message-passing multi-task architecture for the implicit event and polarity detection

Xiang

Zhang

2021

PLoS ONE

View full text Add to dashboard Cite

show abstract

“…, L k are the neighbors of X excluding (2) Here v 1 • v 2 denotes the Hadamard (componentwise) product, and if k = 0 an all-ones vector is returned. ley, Dredze, & Eisner, 2015). If a clause is not a polytree, then this is detected (by analyzing the literal influence graph for loops) and reported to the user.…”

Section: Differentiable Inference For One-clause Theoriesmentioning

confidence: 99%

TensorLog: A Probabilistic Database Implemented Using Deep-Learning Infrastructure

Cohen¹,

Yang²,

Mazaitis³

2020

jair

View full text Add to dashboard Cite

We present an implementation of a probabilistic first-order logic called TensorLog, in which classes of logical queries are compiled into differentiable functions in a neural-network infrastructure such as Tensorflow or Theano. This leads to a close integration of probabilistic logical reasoning with deep-learning infrastructure: in particular, it enables high-performance deep learning frameworks to be used for tuning the parameters of a probabilistic logic. The integration with these frameworks enables use of GPU-based parallel processors for inference and learning, making TensorLog the first highly parallellizable probabilistic logic. Experimental results show that TensorLog scales to problems involving hundreds of thousands of knowledge-base triples and tens of thousands of examples.

show abstract

“…Apart from this difference, training is carried out using the same procedure. Gormley et al (2015)'s approximation-aware training is conceptually related, but focuses on variational decoding procedures. Hoang et al (2017) also propose continuous relaxations of decoders, but are focused on developing better inference procedures.…”

Section: Differentiable Relaxed Decodersmentioning

confidence: 99%

Differentiable Scheduled Sampling for Credit Assignment

Goyal¹,

Dyer²,

Berg-Kirkpatrick³

2017

Proceedings of the 55th Annual Meeting of the Association For Computational Linguistics (Volume 2: Short Papers)

View full text Add to dashboard Cite

We demonstrate that a continuous relaxation of the argmax operation can be used to create a differentiable approximation to greedy decoding for sequence-tosequence (seq2seq) models. By incorporating this approximation into the scheduled sampling training procedure (Bengio et al., 2015)-a well-known technique for correcting exposure bias-we introduce a new training objective that is continuous and differentiable everywhere and that can provide informative gradients near points where previous decoding decisions change their value. In addition, by using a related approximation, we demonstrate a similar approach to sampled-based training. Finally, we show that our approach outperforms cross-entropy training and scheduled sampling procedures in two sequence prediction tasks: named entity recognition and machine translation.

show abstract

Approximation-Aware Dependency Parsing by Belief Propagation

Cited by 27 publications

References 23 publications

A message-passing multi-task architecture for the implicit event and polarity detection

A message-passing multi-task architecture for the implicit event and polarity detection

TensorLog: A Probabilistic Database Implemented Using Deep-Learning Infrastructure

Differentiable Scheduled Sampling for Credit Assignment

Contact Info

Product

Resources

About