Extreme Multilabel Text Classification (XMTC) is a text classification problem in which, (i) the output space is extremely large, (ii) each data point may have multiple positive labels, and (iii) the data follows a strongly imbalanced distribution. With applications in recommendation systems and automatic tagging of web-scale documents, the research on XMTC has been focused on improving prediction accuracy and dealing with imbalanced data. However, the robustness of deep learning based XMTC models against adversarial examples has been largely underexplored. In this paper, we investigate the behaviour of XMTC models under adversarial attacks. To this end, first, we define adversarial attacks in multilabel text classification problems. We categorize attacking multilabel text classifiers as (a) positive-targeted, where the target positive label should fall out of top-k predicted labels, and (b) negative-targeted, where the target negative label should be among the top-k predicted labels. Then, by experiments on APLC-XLNet and AttentionXML, we show that XMTC models are highly vulnerable to positive-targeted attacks but more robust to negative-targeted ones. Furthermore, our experiments show that the success rate of positive-targeted adversarial attacks has an imbalanced distribution. More precisely, tail classes are highly vulnerable to adversarial attacks for which an attacker can generate adversarial samples with high similarity to the actual data-points. To overcome this problem, we explore the effect of rebalanced loss functions in XMTC where not only do they increase accuracy on tail classes, but they also improve the robustness of these classes against adversarial attacks. The code for our experiments is available at https://github.com/xmc-aalto/adv-xmtc
Forms are a widespread type of template-based document used in a great variety of fields including, among others, administration, medicine, finance, or insurance. The automatic extraction of the information included in these documents is greatly demanded due to the increasing volume of forms that are generated in a daily basis. However, this is not a straightforward task when working with scanned forms because of the great diversity of templates with different location of form entities, and the quality of the scanned documents. In this context, there is a feature that is shared by all forms: they contain a collection of interlinked entities built as key-value (or label-value) pairs, together with other entities such as headers or images. In this work, we have tacked the problem of entity linking in forms by combining image processing techniques and a text classification model based on the BERT architecture. This approach achieves state-of-the-art results with a F1-score of 0.80 on the FUNSD dataset, a 5% improvement regarding the best previous method. The code of this project is available at https://github.com/mavillot/FUNSD-Entity-Linking.
Text classification is one of the fundamental tasks in natural language processing to label an open-ended text and is useful for various applications such as sentiment analysis. In this paper, we discuss various classification approaches for Khmer text, ranging from a classical TF-IDF algorithm with support vector machine classifier to modern word embedding-based neural network classifiers including linear layer model, recurrent neural network and convolutional neural network. A Khmer word embedding model is trained on a 30-million-Khmer-word corpus to construct word vector representations that are used to train three different neural network classifiers. We evaluate the performance of different approaches on a news article dataset for both multi-class and multi-label text classification tasks. The result suggests that neural network classifiers using a word embedding model consistently outperform the traditional classifier using TF-IDF. The recurrent neural network classifier provides a slightly better result compared to the convolutional network and the linear layer network.
In this paper, we present a method of building strong, explainable classifiers in the form of Boolean search rules. We developed an interactive environment called CASE (Computer Assisted Semantic Exploration) which exploits word co-occurrence to guide human annotators in selection of relevant search terms. The system seamlessly facilitates iterative evaluation and improvement of the classification rules. The process enables the human annotators to leverage the benefits of statistical information while incorporating their expert intuition into the creation of such rules. We evaluate classifiers created with our CASE system on 4 datasets, and compare the results to machine learning methods, including SKOPE rules, Random forest, Support Vector Machine, and fastText classifiers. The results drive the discussion on trade-offs between superior compactness, simplicity, and intuitiveness of the Boolean search rules versus the better performance of state-of-the-art machine learning models for text classification.
The ubiquity of offensive and hateful content on online fora necessitates the need for automatic solutions that detect such content competently across target groups. In this paper we show that text classification models trained on large publicly available datasets despite having a high overall performance, may significantly under-perform on several protected groups. On the \citet{vidgen2020learning} dataset, we find the accuracy to be 37\% lower on an under annotated Black Women target group and 12\% lower on Immigrants, where hate speech involves a distinct style. To address this, we propose to perform token-level hate sense disambiguation, and utilize tokens' hate sense representations for detection, modeling more general signals. On two publicly available datasets, we observe that the variance in model accuracy across target groups drops by at least 30\%, improving the average target group performance by 4\% and worst case performance by 13\%.
Modern Web systems such as social media and e-commerce contain rich contents expressed in images and text. Leveraging information from multi-modalities can improve the performance of machine learning tasks such as classification and recommendation. In this paper, we propose the Cross-Modality Attention Contrastive Language-Image Pre-training (CMA-CLIP), a new framework which unifies two types of cross-modality attentions, sequence-wise attention and modality-wise attention, to effectively fuse information from image and text pairs. The sequence-wise attention enables the framework to capture the fine-grained relationship between image patches and text tokens, while the modality-wise attention weighs each modality by its relevance to the downstream tasks. In addition, by adding task specific modality-wise attentions and multilayer perceptrons, our proposed framework is capable of performing multi-task classification with multi-modalities. We conduct experiments on a Major Retail Website Product Attribute (MRWPA) dataset and two public datasets, Food101 and Fashion-Gen. The results show that CMA-CLIP outperforms the pre-trained and fine-tuned CLIP by an average of 11.9% in recall at the same level of precision on the MRWPA dataset for multi-task classification. It also surpasses the state-of-the-art method on Fashion-Gen Dataset by 5.5% in accuracy and achieves competitive performance on Food101 Dataset. Through detailed ablation studies, we further demonstrate the effectiveness of both cross-modality attention modules and our method's robustness against noise in image and text inputs, which is a common challenge in practice.
Text classification is a fundamental language task in Natural Language Processing. A variety of sequential models is capable making good predictions yet there is lack of connection between language semantics and prediction results. This paper proposes a novel influence score (I-score), a greedy search algorithm called Backward Dropping Algorithm (BDA), and a novel feature engineering technique called the "dagger technique". First, the paper proposes a novel influence score (I-score) to detect and search for the important language semantics in text document that are useful for making good prediction in text classification tasks. Next, a greedy search algorithm called the Backward Dropping Algorithm is proposed to handle long-term dependencies in the dataset. Moreover, the paper proposes a novel engineering technique called the "dagger technique" that fully preserve the relationship between explanatory variable and response variable. The proposed techniques can be further generalized into any feed-forward Artificial Neural Networks (ANNs) and Convolutional Neural Networks (CNNs), and any neural network. A real-world application on the Internet Movie Database (IMDB) is used and the proposed methods are applied to improve prediction performance with an 81% error reduction comparing with other popular peers if I-score and "dagger technique" are not implemented.
Dataless text classification, i.e., a new paradigm of weakly supervised learning, refers to the task of learning with unlabeled documents and a few predefined representative words of categories, known as seed words. The recent generative dataless methods construct document-specific category priors by using seed word occurrences only, however, such category priors often contain very limited and even noisy supervised signals. To remedy this problem, in this paper we propose a novel formulation of category prior. First, for each document, we consider its label membership degree by not only counting seed word occurrences, but also using a novel prototype scheme, which captures pseudo-nearest neighboring categories. Second, for each label, we consider its frequency prior knowledge of the corpus, which is also a discriminative knowledge for classification. By incorporating the proposed category prior into the previous generative dataless method, we suggest a novel generative dataless method, namely Weakly Supervised Prototype Topic Model (WSPTM). The experimental results on real-world datasets demonstrate that WSPTM outperforms the existing baseline methods.
Mining user-generated data often suffers from the lack of enough labeled data, short document lengths, and the informal user language. In this paper, we propose a novel active learning model to overcome these obstacles in the tasks tailored for query phrases--e.g., detecting positive reports of natural disasters. Our model has three novelties: 1) It is the first approach to employ multi-view active learning in this domain. 2) It uses the Parzen-Rosenblatt window method to integrate the representativeness measure into multi-view active learning. 3) It employs a query-by-committee strategy, based on the agreement between predictors, to address the usually noisy language of the documents in this domain. We evaluate our model in four publicly available Twitter datasets with distinctly different applications. We also compare our model with a wide range of baselines including those with multiple classifiers. The experiments testify that our model is highly consistent and outperforms existing models.
This work presents a framework to classify and evaluate distinct research abstract texts which are focused on the description of processes and their applications. In this context, this paper proposes natural language processing algorithms to classify, segment and evaluate the results of scientific work. Initially, the proposed framework categorize the abstract texts into according to the problems intended to be solved by employing a text classification approach. Then, the abstract text is segmented into problem description, methodology and results. Finally, the methodology of the abstract is ranked based on the sentiment analysis of its results. The proposed framework allows us to quickly rank the best methods to solve specific problems. To validate the proposed framework, oil production anomaly abstracts were experimented and achieved promising results.
Transformer based language models have led to impressive results across all domains in Natural Language Processing. Pretraining these models on language modeling tasks and finetuning them on downstream tasks such as Text Classification, Question Answering and Neural Machine Translation has consistently shown exemplary results. In this work, we propose a Multitask Finetuning methodology which combines the Bilingual Machine Translation task with an auxiliary Causal Language Modeling task to improve performance on the former task on Indian Languages. We conduct an empirical study on three language pairs, Marathi-Hindi, Marathi-English and Hindi-English, where we compare the multitask finetuning approach to the standard finetuning approach, for which we use the mBART50 model. Our study indicates that the multitask finetuning method could be a better technique than standard finetuning, and could improve Bilingual Machine Translation across language pairs.
As the fourth largest language family in the world, the Dravidian languages have become a research hotspot in natural language processing (NLP). Although the Dravidian languages contain a large number of languages, there are relatively few public available resources. Besides, text classification task, as a basic task of natural language processing, how to combine it to multiple languages in the Dravidian languages, is still a major difficulty in Dravidian Natural Language Processing. Hence, to address these problems, we proposed a multilingual text classification framework for the Dravidian languages. On the one hand, the framework used the LaBSE pre-trained model as the base model. Aiming at the problem of text information bias in multi-task learning, we propose to use the MLM strategy to select language-specific words, and used adversarial training to perturb them. On the other hand, in view of the problem that the model cannot well recognize and utilize the correlation among languages, we further proposed a language-specific representation module to enrich semantic information for the model. The experimental results demonstrated that the framework we proposed has a significant performance in multilingual text classification tasks with each strategy achieving certain improvements.
In this paper, we propose that the dot product pairwise matching attention layer, which is widely used in Transformer-based models, is redundant for the model performance. Attention, in its original formulation, has to be seen rather as a human-level tool to explore and/or visualize relevancy scores in sequential data. However, the way how it is constructed leads to significant computational complexity. Instead, we present SimpleTRON: Simple Transformer with O(N) Complexity, a simple and fast alternative without any approximation that, unlike other approximation models, does not have any architecture-related overhead and therefore can be seen as a purely linear Transformer-like model. This architecture, to the best of our knowledge, outperforms existing sub-quadratic attention approximation models on several tasks from the Long-Range Arena benchmark. Moreover, we show, that SimpleTRON can benefit from weight transfer from pretrained large language models, as its parameters can be fully transferable.
This paper presents the novel way combining the BERT embedding method and the graph convolutional neural network. This combination is employed to solve the text classification problem. Initially, we apply the BERT embedding method to the texts (in the BBC news dataset and the IMDB movie reviews dataset) in order to transform all the texts to numerical vector. Then, the graph convolutional neural network will be applied to these numerical vectors to classify these texts into their ap-propriate classes/labels. Experiments show that the performance of the graph convolutional neural network model is better than the perfor-mances of the combination of the BERT embedding method with clas-sical machine learning models.
With the enrichment of literature resources, researchers are facing the growing problem of information explosion and knowledge overload. To help scholars retrieve literature and acquire knowledge successfully, clarifying the semantic structure of the content in academic literature has become the essential research question. In the research on identifying the structure function of chapters in academic articles, only a few studies used the deep learning model and explored the optimization for feature input. This limits the application, optimization potential of deep learning models for the research task. This paper took articles of the ACL conference as the corpus. We employ the traditional machine learning models and deep learning models to construct the classifiers based on various feature input. Experimental results show that (1) Compared with the chapter content, the chapter title is more conducive to identifying the structure function of academic articles. (2) Relative position is a valuable feature for building traditional models. (3) Inspired by (2), this paper further introduces contextual information into the deep learning models and achieved significant results. Meanwhile, our models show good migration ability in the open test containing 200 sampled non-training samples. We also annotated the ACL main conference papers in recent five years based on the best practice performing models and performed a time series analysis of the overall corpus. This work explores and summarizes the practical features and models for this task through multiple comparative experiments and provides a reference for related text classification tasks. Finally, we indicate the limitations and shortcomings of the current model and the direction of further optimization.
In this paper, we present our work participating in the BioCreative VII Track 3 - automatic extraction of medication names in tweets, where we implemented a multi-task learning model that is jointly trained on text classification and sequence labelling. Our best system run achieved a strict F1 of 80.4, ranking first and more than 10 points higher than the average score of all participants. Our analyses show that the ensemble technique, multi-task learning, and data augmentation are all beneficial for medication detection in tweets.
Zero-shot cross-lingual transfer is an important feature in modern NLP models and architectures to support low-resource languages. In this work, We study zero-shot cross-lingual transfer from English to French and German under Multi-Label Text Classification, where we train a classifier using English training set, and we test using French and German test sets. We extend EURLEX57K dataset, the English dataset for topic classification of legal documents, with French and German official translation. We investigate the effect of using some training techniques, namely Gradual Unfreezing and Language Model finetuning, on the quality of zero-shot cross-lingual transfer. We find that Language model finetuning of multi-lingual pre-trained model (M-DistilBERT, M-BERT) leads to 32.0-34.94%, 76.15-87.54% relative improvement on French and German test sets correspondingly. Also, Gradual unfreezing of pre-trained model's layers during training results in relative improvement of 38-45% for French and 58-70% for German. Compared to training a model in Joint Training scheme using English, French and German training sets, zero-shot BERT-based classification model reaches 86% of the performance achieved by jointly-trained BERT-based classification model.
Many existing neural architecture search (NAS) solutions rely on downstream training for architecture evaluation, which takes enormous computations. Considering that these computations bring a large carbon footprint, this paper aims to explore a green (namely environmental-friendly) NAS solution that evaluates architectures without training. Intuitively, gradients, induced by the architecture itself, directly decide the convergence and generalization results. It motivates us to propose the gradient kernel hypothesis: Gradients can be used as a coarse-grained proxy of downstream training to evaluate random-initialized networks. To support the hypothesis, we conduct a theoretical analysis and find a practical gradient kernel that has good correlations with training loss and validation performance. According to this hypothesis, we propose a new kernel based architecture search approach KNAS. Experiments show that KNAS achieves competitive results with orders of magnitude faster than "train-then-test" paradigms on image classification tasks. Furthermore, the extremely low search cost enables its wide applications. The searched network also outperforms strong baseline RoBERTA-large on two text classification tasks. Codes are available at \url{https://github.com/Jingjing-NLP/KNAS} .
Automatic monitoring of adverse drug events (ADEs) or reactions (ADRs) is currently receiving significant attention from the biomedical community. In recent years, user-generated data on social media has become a valuable resource for this task. Neural models have achieved impressive performance on automatic text classification for ADR detection. Yet, training and evaluation of these methods are carried out on user-generated texts about a targeted drug. In this paper, we assess the robustness of state-of-the-art neural architectures across different drug groups. We investigate several strategies to use pseudo-labeled data in addition to a manually annotated train set. Out-of-dataset experiments diagnose the bottleneck of supervised models in terms of breakdown performance, while additional pseudo-labeled data improves overall results regardless of the text selection strategy.
In light of the success of transferring language models into NLP tasks, we ask whether the full BERT model is always the best and does it exist a simple but effective method to find the winning ticket in state-of-the-art deep neural networks without complex calculations. We construct a series of BERT-based models with different size and compare their predictions on 8 binary classification tasks. The results show there truly exist smaller sub-networks performing better than the full model. Then we present a further study and propose a simple method to shrink BERT appropriately before fine-tuning. Some extended experiments indicate that our method could save time and storage overhead extraordinarily with little even no accuracy loss.