Zobrazeno 1 - 10
of 1 478
pro vyhledávání: '"Liu, Huijun"'
Code Pre-trained Models (CodePTMs) based vulnerability detection have achieved promising results over recent years. However, these models struggle to generalize as they typically learn superficial mapping from source code to labels instead of underst
Externí odkaz:
http://arxiv.org/abs/2406.03718
Autor:
Li, Xiaopeng, Li, Shasha, Song, Shezheng, Liu, Huijun, Ji, Bin, Wang, Xi, Ma, Jun, Yu, Jie, Liu, Xiaodong, Wang, Jing, Zhang, Weimin
The general capabilities of large language models (LLMs) make them the infrastructure for various AI applications, but updating their inner knowledge requires significant resources. Recent model editing is a promising technique for efficiently updati
Externí odkaz:
http://arxiv.org/abs/2401.17809
Autor:
Liu Caixia, Wang Jinmei, Tan Yurong, Liu Chi, Qu Xiangping, Liu Huijun, Tan Meiling, Deng Changqing, Qin Xiaoqun, Xiang Yang
Publikováno v:
Acta Biochimica et Biophysica Sinica, Vol 56, Pp 753-762 (2024)
Adhesion molecules play critical roles in maintaining the structural integrity of the airway epithelium in airways under stress. Previously, we reported that catenin alpha-like 1 (CTNNAL1) is downregulated in an asthma animal model and upregulated at
Externí odkaz:
https://doaj.org/article/4bb3672d0db0420a8f367165a333d8c0
Span-based joint extraction simultaneously conducts named entity recognition (NER) and relation extraction (RE) in text span form. However, since previous span-based models rely on span-level classifications, they cannot benefit from token-level labe
Externí odkaz:
http://arxiv.org/abs/2210.12720
An exhaustive study has been conducted to investigate span-based models for the joint entity and relation extraction task. However, these models sample a large number of negative entities and negative relations during the model training, which are es
Externí odkaz:
http://arxiv.org/abs/2208.08659
Textual adversarial attacks expose the vulnerabilities of text classifiers and can be used to improve their robustness. Existing context-aware methods solely consider the gold label probability and use the greedy search when searching an attack path,
Externí odkaz:
http://arxiv.org/abs/2208.08029
Few-shot named entity recognition (NER) enables us to build a NER system for a new domain using very few labeled examples. However, existing prototypical networks for this task suffer from roughly estimated label dependency and closely distributed pr
Externí odkaz:
http://arxiv.org/abs/2208.08023
Document retrieval enables users to find their required documents accurately and quickly. To satisfy the requirement of retrieval efficiency, prevalent deep neural methods adopt a representation-based matching paradigm, which saves online matching ti
Externí odkaz:
http://arxiv.org/abs/2207.04656
For Named Entity Recognition (NER), sequence labeling-based and span-based paradigms are quite different. Previous research has demonstrated that the two paradigms have clear complementary advantages, but few models have attempted to leverage these a
Externí odkaz:
http://arxiv.org/abs/2207.03300
Publikováno v:
Gongye shui chuli, Vol 44, Iss 1, Pp 132-137 (2024)
The evaporation characteristic of desulfurization wastewater is an important indicator for determining the process parameters of three effect evaporation. Through evaporation experiments,the relationship between the evaporation rate of desulfurizat
Externí odkaz:
https://doaj.org/article/e2571799e4f94533ad876bbb0be908cc