Cross-domain Text Classification with Multiple Domains and Disparate Label Sets

Bhatt, Himanshu; Sinha, Manjira; Roy, Shourya

doi:10.18653/v1/p16-1155

Cited by 12 publications

(5 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…should be classified as the topic 'Apple' although it does not contain the keyword 'Apple' and the keyword 'Tim Cook' is not contained in the training samples. In other words, the reliable classifier should learn decision rules that generalize across domains (Fei and Liu 2015;Bhatt, Semwal, and Roy 2015;Bhatt, Sinha, and Roy 2016). This problematic phenomenon frequently happens in realworld datasets.…”

Section: Class : Applementioning

confidence: 99%

MASKER: Masked Keyword Regularization for Reliable Text Classification

Moon

Lee

et al. 2021

AAAI

View full text Add to dashboard Cite

Pre-trained language models have achieved state-of-the-art accuracies on various text classification tasks, e.g., sentiment analysis, natural language inference, and semantic textual similarity. However, the reliability of the fine-tuned text classifiers is an often underlooked performance criterion. For instance, one may desire a model that can detect out-of-distribution (OOD) samples (drawn far from training distribution) or be robust against domain shifts. We claim that one central obstacle to the reliability is the over-reliance of the model on a limited number of keywords, instead of looking at the whole context. In particular, we find that (a) OOD samples often contain in-distribution keywords, while (b) cross-domain samples may not always contain keywords; over-relying on the keywords can be problematic for both cases. In light of this observation, we propose a simple yet effective fine-tuning method, coined masked keyword regularization (MASKER), that facilitates context-based prediction. MASKER regularizes the model to reconstruct the keywords from the rest of the words and make low-confidence predictions without enough context. When applied to various pre-trained language models (e.g., BERT, RoBERTa, and ALBERT), we demonstrate that MASKER improves OOD detection and cross-domain generalization without degrading classification accuracy. Code is available at https://github.com/alinlab/MASKER.

show abstract

Section: Class : Applementioning

confidence: 99%

MASKER: Masked Keyword Regularization for Reliable Text Classification

Moon

Lee

et al. 2021

AAAI

View full text Add to dashboard Cite

show abstract

“…Even if Word2Vec models are more flexible as regards the training domain, Doc2vec models are more advantageous when document size is large (Lau & Baldwin, 2016). However, it is supposed that both neural networks outperform traditional TF-IDF representation (Bhatt et al, 2016).…”

Section: Literature Reviewmentioning

confidence: 99%

Efficiency of SVM classifier with Word2Vec and Doc2Vec models

Truşcǎ

2019

Proceedings of the International Conference on Applied Statistics

View full text Add to dashboard Cite

Support Vector Machine model is one of the most intensive used text data classifiers ever since the moment of its development. However, its performance depends not only on its features but also on data preprocessing and model tuning. The main purpose of this paper is to compare the efficiency of more Support Vector Machine models using both TF-IDF approach and Word2Vec and Doc2Vec neural networks for text data representation. Besides the data vectorization process, I try to enhance the models’ efficiency by identifying which kind of kernel fits better the data or if it is just better to opt for the linear case. My results prove that for the “Reuters 21578” dataset, nonlinear Support Vector Machine is more efficient when the conversion of text data into numerical attributes is realized using Word2Vec models instead of TF-IDF and Doc2Vec representations. When it is considered that data meet linear separability requirements, TF-IDF representation outperforms all other options. Surprisingly, Doc2Vec models have the lowest performance and only in terms of computational cost they provide satisfactory results. This paper proves that while Word2Vec models are truly efficient for text data representation, Doc2Vec neural networks are unable to exceed even TF-IDF index representation. This evidence contradicts the common idea according to which Doc2Vec models should provide a better insight into the training data domain than Word2Vec models and certainly than the TF-IDF index.

show abstract

“…In the text field, researchers have proposed various methods to tackle this problem based on non-deep neural networks [4,[28][29][30][31][32][33][34][35][36]. Blitzer et al presented structural correspondence learning (SCL) for transfer learning [4].…”

Section: Cross-domain Transfer Learningmentioning

confidence: 99%

“…Bhatt et al presented a cross-domain classification method. It can learn an accurate model for the new unlabeled target domain given labeled data from multiple source domains where all domains have (possibly) different label sets [34]. Qu et al proposed a transfer learning-based approach to named entity recognition in novel domains with label mismatch over a source domain [35].…”

Section: Cross-domain Transfer Learningmentioning

confidence: 99%

A Convolution-LSTM-Based Deep Neural Network for Cross-Domain MOOC Forum Post Classification

et al. 2017

View full text Add to dashboard Cite

Learners in a massive open online course often express feelings, exchange ideas and seek help by posting questions in discussion forums. Due to the very high learner-to-instructor ratios, it is unrealistic to expect instructors to adequately track the forums, find all of the issues that need resolution and understand their urgency and sentiment. In this paper, considering the biases among different courses, we propose a transfer learning framework based on a convolutional neural network and a long short-term memory model, called ConvL, to automatically identify whether a post expresses confusion, determine the urgency and classify the polarity of the sentiment. First, we learn the feature representation for each word by considering the local contextual feature via the convolution operation. Second, we learn the post representation from the features extracted through the convolution operation via the LSTM model, which considers the long-term temporal semantic relationships of features. Third, we investigate the possibility of transferring parameters from a model trained on one course to another course and the subsequent fine-tuning. Experiments on three real-world MOOC courses confirm the effectiveness of our framework. This work suggests that our model can potentially significantly increase the effectiveness of monitoring MOOC forums in real time.

show abstract

Cross-domain Text Classification with Multiple Domains and Disparate Label Sets

Cited by 12 publications

References 24 publications

MASKER: Masked Keyword Regularization for Reliable Text Classification

MASKER: Masked Keyword Regularization for Reliable Text Classification

Efficiency of SVM classifier with Word2Vec and Doc2Vec models

A Convolution-LSTM-Based Deep Neural Network for Cross-Domain MOOC Forum Post Classification

Contact Info

Product

Resources

About