SwitchNet: A modular neural network for adaptive relation extraction

Zhu, Hongyin; Tiwari, Prayag; Zhang, Yazhou; Gupta, Das; Alharbi, Meshal; Nguyen, Thi Thanh Huyen; Dehdashti, Shahram

doi:10.1016/j.compeleceng.2022.108445

Cited by 5 publications

(4 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…e model has the potential to be extended to other cascaded tasks, such as information extraction and downstream applications, in the future [38].…”

Section: Discussionmentioning

confidence: 99%

Reranking passages with coarse-to-ne neural retriever enhanced by list-context information

2024

JAIaR

View full text Add to dashboard Cite

Passage reranking is a critical task in various applications, particularly when dealing with large volumes of documents. Existing neural architectures have limitations in retrieving the most relevant passage for a given question because the semantics of the segmented passages are often incomplete, and they typically match the question to each passage individually, rarely considering contextual information from other passages that could provide comparative and reference information. This paper presents a list-context attention mechanism to augment the passage representation by incorporating the list-context information from other candidates. The proposed coarse-to-ne (C2F) neural retriever addresses the out-of-memory limitation of the passage attention mechanism by dividing the list-context modeling process into two sub-processes with a cache policy learning algorithm, enabling the e cient encoding of context information from a large number of candidate answers. This method can be generally used to encode context information from any number of candidate answers in one pass. Di erent from most multi-stage information retrieval architectures, this model integrates the coarse and ne rankers into the joint optimization process, allowing for feedback between the two layers to update the model simultaneously. Experiments demonstrate the e ectiveness of the proposed approach.

show abstract

“…e model has the potential to be extended to other cascaded tasks, such as information extraction and downstream applications, in the future [38].…”

Section: Discussionmentioning

confidence: 99%

Reranking passages with coarse-to-ne neural retriever enhanced by list-context information

2024

JAIaR

View full text Add to dashboard Cite

show abstract

“…In this subsection, we will introduce the PM modules in the framework, each named after some Beijing locations where I have lived. To realize the controllability of human creativity, we design different PM information flows (Zhu et al 2022). As shown in Figure 1, the modules in it can be combined in a custom way.…”

Section: Functional Module Architecturementioning

confidence: 99%

“…Human controllability can be divided into two dimensions: the controllability of human creativity and the controllability of model outputs responsibility. Through the PM information flow (Zhu et al 2022) set by users, the controllability of human creativity can be realized, the value of human imagination can be maximized, and cultural innovation can be carried out. By generating URI-extension for output resources, responsibility for output data can be controlled, enabling "data ownership" by the way.…”

Section: Introductionmentioning

confidence: 99%

A Simple Survey of Pre-trained Language Models

Zhu¹

2022

Preprint

View full text Add to dashboard Cite

Pre-trained Language Models (PTLM) have remarkable and successful performance in solving lots of NLP tasks nowadays. And previous researchers have created many SOTA models and these models are included in many long surveys(Qiu et al., 2020). So, we would like to conduct a simple and short survey on this topic to help researchers understand the sketch of PTLM more quickly and comprehensively. In this short survey, we would provide a simple but comprehensive review of techniques, benchmarks, and methodologies in PTLM. And we would also introduce the applications evaluation of PTLM in this simple survey.

show abstract

“…This algorithm enables the model to dynamically adapt to new patterns in the data. The downside is that there is a sort of information loss [Zhu et al (2022)] in the discrete inference process.…”

Section: Training Algorithmmentioning

confidence: 99%

Design of Motion Damage Estimation Model Based on Improved Recursive Neural Network Algorithm

Zhu

Zhang

2022

Mobile Information Systems

View full text Add to dashboard Cite

In recent years, the China’s rapid economic development and living standards of people have been improved, and the usage of clothing, food, housing, and transportation has increased. However, due to the prevalence of sports injuries and their high frequency of incidence, sports injury incidents do occur from time to time. Sports injury will not only affect the normal training of athletes and sports enthusiasts but also adversely affect their health. An estimation model of sports injury is designed based on an improved recursive neural network algorithm to reduce the incidence of sports injury in the process of physical exercise so that athletes and sports enthusiasts can reduce sports injury, which is conducive to physical health and sports health.

show abstract

SwitchNet: A modular neural network for adaptive relation extraction

Cited by 5 publications

References 6 publications

Reranking passages with coarse-to-ne neural retriever enhanced by list-context information

Reranking passages with coarse-to-ne neural retriever enhanced by list-context information

A Simple Survey of Pre-trained Language Models

Design of Motion Damage Estimation Model Based on Improved Recursive Neural Network Algorithm

Contact Info

Product

Resources

About