Few-shot learning for short text classification

Yan, Lianshan; Zheng, Yuhui; Cao, Jie

doi:10.1007/s11042-018-5772-4

Cited by 81 publications

(41 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Another category of metric-learning based approaches such as siamese networks [14], matching networks [15], prototypical networks [16] and relation networks [17] aim to learn a set of projection functions such that when represented in this embedding, images are easy to recognize using simple linear classifiers. Only a few works focus on the few-shot learning on NLP tasks, for example, a text classification framework based on siamese CNN network and few-shot learning is proposed in [18]. However, taking the widely used experimental setting in few-shot learning for instance, miniImagenet, the benchmark dataset is split into 64, 16, and 20 classes for training, validation and testing, respectively.…”

Section: Related Work a Few-shot Transfer Learningmentioning

confidence: 99%

Few-Shot Transfer Learning for Text Classification With Lightweight Word Embedding Based Models

et al. 2019

View full text Add to dashboard Cite

Many deep learning architectures have been employed to model the semantic compositionality for text sequences, requiring a huge amount of supervised data for parameters training, making it unfeasible in situations where numerous annotated samples are not available or even do not exist. Different from data-hungry deep models, lightweight word embedding-based models could represent text sequences in a plug-and-play way due to their parameter-free property. In this paper, a modified hierarchical pooling strategy over pre-trained word embeddings is proposed for text classification in a few-shot transfer learning way. The model leverages and transfers knowledge obtained from some source domains to recognize and classify the unseen text sequences with just a handful of support examples in the target problem domain. The extensive experiments on five datasets including both English and Chinese text demonstrate that the simple word embedding-based models (SWEMs) with parameter-free pooling operations are able to abstract and represent the semantic text. The proposed modified hierarchical pooling method exhibits significant classification performance in the few-shot transfer learning tasks compared with other alternative methods. INDEX TERMS Few-shot learning, transfer learning, text classification, word embedding based models, pooling strategy.

show abstract

Section: Related Work a Few-shot Transfer Learningmentioning

confidence: 99%

Few-Shot Transfer Learning for Text Classification With Lightweight Word Embedding Based Models

et al. 2019

View full text Add to dashboard Cite

show abstract

“…Several works (Nishida et al, 2011;Romero et al, 2013; Ramírez de la Rosa et al, 2013;Yang et al, 2013;Zhang & Zhong, 2016;Wang et al, 2016a;Dai et al, 2017;Li et al, 2018;Ravi & Kozareva, 2018;Yan et al, 2018) focused on designing new classification techniques specifically for short-texts. The reviewed techniques are organised under three categories, according to how they tackle the classification task: word/character sequence-based techniques (including techniques based on data compression and similarity), domain knowledge-based techniques (including ontologies and diverse corpora) and neural networks-based techniques (including techniques leveraging on word embeddings and deep learning).…”

Section: New Classification Techniquesmentioning

confidence: 99%

“…Traditional BOW models might have difficulties for capturing the semantic meaning of short-texts. To overcome this problem, several works (Wang et al, 2016a;Dai et al, 2017;Ravi & Kozareva, 2018;Yan et al, 2018) have leveraged on word embedding models and deep learning techniques for short-text classification. Word embeddings aim at quantifying the semantic similarity of linguistic items based on the distributional properties of words in large textual samples.…”

Section: Neural Network-based Techniquesmentioning

confidence: 99%

“…Given the sparsity and brevity of short-texts, each text might not provide enough information for labelling it or discriminating it in the feature space. As a result, thousands of labelled examples might not be sufficient for classifiers to obtain satisfying performance (Nakov et al, 2016) In this context, Yan et al (2018) presented a sentiment classification technique relying on a limited number of labelled instances and metric learning based on deep neural features and Siamese CNN. The framework comprises two training steps; first, the pre-training of a CNN model, and second, the training of a Siamese CNN based on the pre-trained CNN model and fine-tuned with the small number of labelled instances.…”

Section: Neural Network-based Techniquesmentioning

confidence: 99%

See 1 more Smart Citation

Short-text learning in social media: a review

Tommasel

Godoy

2019

The Knowledge Engineering Review

View full text Add to dashboard Cite

Social networks occupy a ubiquitous and pervasive place in the life of their users. The substantial amount of content generated and shared by social networking users offers new research opportunities across a wide variety of disciplines, including media and communication studies, linguistics, sociology, psychology, information and computer sciences, or education. This situation, in combination with the continuous growth of social media data, creates an imperative need for content organisation. Thus, large-scale text learning tasks in social environments arise as one of the most relevant problems in machine learning and data mining. Interestingly, social media data pose several challenges due to its sparse, high-dimensional and large-volume characteristics. This survey reviews the field of social media data learning, focusing on classification and clustering techniques, as they are two of the most frequent learning tasks. It reviews not only new techniques that have been developed to tackle the new challenges posed by short-texts, but also how traditional techniques can be adapted to overcome such challenges. Then, open issues and research opportunities for social media data learning are discussed.

show abstract

“…), such as patents, news, and papers. e most typical object of text without structure is short text [8][9][10]. Text with an obvious structure, due to its clear structure, often appears in formal written expressions and has a wide range of applications.…”

Section: Introductionmentioning

confidence: 99%

Fusing Logical Relationship Information of Text in Neural Network for Text Classification

Wang

Zeng

2020

Mathematical Problems in Engineering

View full text Add to dashboard Cite

With the development of computer science and information science, text classification technology has been greatly developed and its application scenarios have been widened. In traditional process of text classification, the existing method will lose much logical relationship information of text. The logical relationship information of a text refers to the relationship information among different logical parts of the text, such as title, abstract, and body. When human beings are reading, they will take title as an important part to remind the central idea of the article, abstract as a brief summary of the content of the article, and body as a detailed description of the article. In most of the text classification studies, researchers concern more about the relationship among words (word frequency, semantics, etc.) and neglect the logical relationship information of text. It will lose information about the relationship among different parts (title, body, etc.) and have an influence on the performance of text classification. Therefore, we propose a text classification algorithm—fusing the logical relationship information of text in neural network (FLRIOTINN), which complements the logical relationship information into text classification algorithms. Experiments show that the effect of FLRIOTINN is better than the conventional backpropagation neural networks which does not consider the logical relationship information of text.

show abstract

Few-shot learning for short text classification

Cited by 81 publications

References 34 publications

Few-Shot Transfer Learning for Text Classification With Lightweight Word Embedding Based Models

Few-Shot Transfer Learning for Text Classification With Lightweight Word Embedding Based Models

Short-text learning in social media: a review

Fusing Logical Relationship Information of Text in Neural Network for Text Classification

Contact Info

Product

Resources

About