精神:结构透指导前调整为层次的文本分类
He Zhu1, Jinxiang Xia1, Ruomei Liu1
1State Key Laboratory of Software Development Environment, School of Computer Science and Engineering, Beihang University, No. 37 Xue Yuan Road, Hai Dian District, Beijing 100191, China.
层次性文本分类 (HTC) 得到了SPIRIT的改进,这是一种使用结构的语言模型指导的新方法. 这种方法提高了所有层面的及时关注,以提高绩效.
科学领域:
- 自然语言处理自然语言处理.
- 机器学习 机器学习
- 人工智能的人工智能
背景情况:
- 层次性文本分类 (HTC) 由于复杂的标签依赖性而带来挑战.
- 对于HTC的快速调整方法,在跨语言模型层保持模型注意力方面存在局限性.
- 现有的快速调整很难有效地利用标签层次结构的复杂结构.
研究的目的:
- 为了解决目前在层次性文本分类中提示调整的局限性.
- 引入一种新的方法,增强语言模型对等级结构的关注.
- 通过更好地整合标签层次信息来提高层次文本分类的性能.
主要方法:
- 建议采用结构导向前调整 (SPIRIT),利用结构最小化.
- 基本的标签层次结构被提取并解码为语言模型的前.
- 采用深度智能重定格化策略来优化并通过所有LM层传播前.
主要成果:
- SPIRIT 在四个广泛使用的层次性文本分类数据集上展示了最先进的性能.
- 该方法有效地提示语言模型的所有中间层,克服以前的注意力衰减问题.
- 增强的优化和前传播导致显著的性能增长.
结论:
- 通过有效地整合标签层次信息,SPIRIT提供了对层次文本分类的卓越方法.
- 拟议的方法克服了HTC先前的提示调技术的局限性.
- SPIRIT 设定了在层次性文本分类任务中的性能新基准.
更多相关视频
12:27Large-scale Reconstructions and Independent, Unbiased Clustering Based on Morphological Metrics to Classify Neurons in Selective Populations
Published on: February 15, 2017
08:56Automatic Image Processing to Determine the Community Size Structure of Riverine Macroinvertebrates
Published on: January 13, 2023
相关概念视频
Survival Tree
Building a Survival Tree
Constructing a...
Classification of Systems-I
Homogeneity dictates that if an input x(t) is multiplied by a constant c, the output y(t) is multiplied by the same constant. Mathematically, this is expressed as:
Structural Classification of Joints
A fibrous joint is where the adjacent bones are united by fibrous connective...
Classification of Systems-II
Force Classification
Contact and non-contact forces are two of the most widely used categories of forces. As the name suggests, contact forces require physical contact between two objects to act upon each other. Examples of contact forces include frictional,...
Aggregates Classification
Petrographic classification groups aggregates based on common mineralogical characteristics. Some of the common mineral groups found in aggregates are...
