In this paper, we explore text classification with extremely weak supervision, i.e., only relying on the surface text of class names. This is a more challenging setting than the seed-driven weak supervision, which allows a few seed words per class. We opt to attack this problem from a representation learning perspective -- ideal document representations should lead to nearly the same results between clustering and the desired classification. In particular, one can classify the same corpus differently (e.g., based on topics and locations), so document representations should be adaptive to the given class names. We propose a novel framework X-Class to realize the adaptive representations. Specifically, we first estimate class representations by incrementally adding the most similar word to each class until inconsistency arises. Following a tailored mixture of class attention mechanisms, we obtain the document representation via a weighted average of contextualized word representations. With the prior of each document assigned to its nearest class, we then cluster and align the documents to classes. Finally, we pick the most confident documents from each cluster to train a text classifier. Extensive experiments demonstrate that X-Class can rival and even outperform seed-driven weakly supervised methods on 7 benchmark datasets. Our dataset and code are released at https://github.com/ZihanWangKi/XClass/ .
A recent approach for few-shot text classification is to convert textual inputs to cloze questions that contain some form of task description, process them with a pretrained language model and map the predicted words to labels. Manually defining this mapping between words and labels requires both domain expertise and an understanding of the language model's abilities. To mitigate this issue, we devise an approach that automatically finds such a mapping given small amounts of training data. For a number of tasks, the mapping found by our approach performs almost as well as hand-crafted label-to-word mappings.
Unsupervised Data Augmentation (UDA) is a semi-supervised technique that applies a consistency loss to penalize differences between a model's predictions on (a) observed (unlabeled) examples; and (b) corresponding 'noised' examples produced via data augmentation. While UDA has gained popularity for text classification, open questions linger over which design decisions are necessary and over how to extend the method to sequence labeling tasks. This method has recently gained traction for text classification. In this paper, we re-examine UDA and demonstrate its efficacy on several sequential tasks. Our main contribution is an empirical study of UDA to establish which components of the algorithm confer benefits in NLP. Notably, although prior work has emphasized the use of clever augmentation techniques including back-translation, we find that enforcing consistency between predictions assigned to observed and randomly substituted words often yields comparable (or greater) benefits compared to these complex perturbation models. Furthermore, we find that applying its consistency loss affords meaningful gains without any unlabeled data at all, i.e., in a standard supervised setting. In short: UDA need not be unsupervised, and does not require complex data augmentation to be effective.
Corporate mergers and acquisitions (M&A) account for billions of dollars of investment globally every year, and offer an interesting and challenging domain for artificial intelligence. However, in these highly sensitive domains, it is crucial to not only have a highly robust and accurate model, but be able to generate useful explanations to garner a user's trust in the automated system. Regrettably, the recent research regarding eXplainable AI (XAI) in financial text classification has received little to no attention, and many current methods for generating textual-based explanations result in highly implausible explanations, which damage a user's trust in the system. To address these issues, this paper proposes a novel methodology for producing plausible counterfactual explanations, whilst exploring the regularization benefits of adversarial training on language models in the domain of FinTech. Exhaustive quantitative experiments demonstrate that not only does this approach improve the model accuracy when compared to the current state-of-the-art and human performance, but it also generates counterfactual explanations which are significantly more plausible based on human trials.
Companies today are racing to leverage the latest digital technologies, such as artificial intelligence, blockchain, and cloud computing. However, many companies report that their strategies did not achieve the anticipated business results. This study is the first to apply state of the art NLP models on unstructured data to understand the different clusters of digital strategy patterns that companies are Adopting. We achieve this by analyzing earnings calls from Fortune Global 500 companies between 2015 and 2019. We use Transformer based architecture for text classification which show a better understanding of the conversation context. We then investigate digital strategy patterns by applying clustering analysis. Our findings suggest that Fortune 500 companies use four distinct strategies which are product led, customer experience led, service led, and efficiency led. This work provides an empirical baseline for companies and researchers to enhance our understanding of the field.
Recent progress in text classification has been focused on high-resource languages such as English and Chinese. For low-resource languages, amongst them most African languages, the lack of well-annotated data and effective preprocessing, is hindering the progress and the transfer of successful methods. In this paper, we introduce two news datasets (KINNEWS and KIRNEWS) for multi-class classification of news articles in Kinyarwanda and Kirundi, two low-resource African languages. The two languages are mutually intelligible, but while Kinyarwanda has been studied in Natural Language Processing (NLP) to some extent, this work constitutes the first study on Kirundi. Along with the datasets, we provide statistics, guidelines for preprocessing, and monolingual and cross-lingual baseline models. Our experiments show that training embeddings on the relatively higher-resourced Kinyarwanda yields successful cross-lingual transfer to Kirundi. In addition, the design of the created datasets allows for a wider use in NLP beyond text classification in future studies, such as representation learning, cross-lingual learning with more distant languages, or as base for new annotations for tasks such as parsing, POS tagging, and NER. The datasets, stopwords, and pre-trained embeddings are publicly available at https://github.com/Andrews2017/KINNEWS-and-KIRNEWS-Corpus .
The problem of learning one task with samples from another task is central to transfer learning (TL). In this paper, we examine a fundamental question: When does combining the data samples from a source task and a target task perform better than single-task learning with the target task alone? This question is motivated by an intriguing phenomenon known as negative transfer often observed in the TL literature. Precise quantification of TL effects -- even within simple statistical models -- has remained elusive in the statistical learning literature. A critical challenge is that to compare TL to single-task learning, we would need to compare the risks between two different estimators in a very precise way. In particular, the comparative advantage of one estimator over another would depend on the specific distribution shifts between the two tasks. This paper applies recent developments in the random matrix theory literature to tackle this challenge in a high-dimensional linear regression setting with two tasks. We provide precise high-dimensional asymptotics for the bias and variance of hard parameter sharing (HPS) estimators in the proportional limit, when the sample sizes of both tasks increase proportionally with dimension at fixed ratios. The precise asymptotics are expressed as a function of the sample sizes of both tasks, the covariate shift between their feature population covariate matrices, and the model shift. We provide illustrative examples of our results in a random-effects model to determine positive and negative transfers. For example, we can identify a phase transition in the high-dimensional linear regression setting from positive transfer to negative transfer under a model shift between the source and target tasks. The finding regarding phase transition can be extended to a multiple-task learning setting where the feature covariates are shared across all tasks.
Fine-tuned pre-trained language models can suffer from severe miscalibration for both in-distribution and out-of-distribution (OOD) data due to over-parameterization. To mitigate this issue, we propose a regularized fine-tuning method. Our method introduces two types of regularization for better calibration: (1) On-manifold regularization, which generates pseudo on-manifold samples through interpolation within the data manifold. Augmented training with these pseudo samples imposes a smoothness regularization to improve in-distribution calibration. (2) Off-manifold regularization, which encourages the model to output uniform distributions for pseudo off-manifold samples to address the over-confidence issue for OOD data. Our experiments demonstrate that the proposed method outperforms existing calibration methods for text classification in terms of expectation calibration error, misclassification detection, and OOD detection on six datasets. Our code can be found at https://github.com/Lingkai-Kong/Calibrated-BERT-Fine-Tuning.
As biological gender is one of the aspects of presenting individual human, much work has been done on gender classification based on people names. The proposals for English and Chinese languages are tremendous; still, there have been few works done for Vietnamese so far. We propose a new dataset for gender prediction based on Vietnamese names. This dataset comprises over 26,000 full names annotated with genders. This dataset is available on our website for research purposes. In addition, this paper describes six machine learning algorithms (Support Vector Machine, Multinomial Naive Bayes, Bernoulli Naive Bayes, Decision Tree, Random Forrest and Logistic Regression) and a deep learning model (LSTM) with fastText word embedding for gender prediction on Vietnamese names. We create a dataset and investigate the impact of each name component on detecting gender. As a result, the best F1-score that we have achieved is up to 96% on LSTM model and we generate a web API based on our trained model.
The Gutenberg Literary English Corpus (GLEC) provides a rich source of textual data for research in digital humanities, computational linguistics or neurocognitive poetics. However, so far only a small subcorpus, the Gutenberg English Poetry Corpus, has been submitted to quantitative text analyses providing predictions for scientific studies of literature. Here we show that in the entire GLEC quasi error-free text classification and authorship recognition is possible with a method using the same set of five style and five content features, computed via style and sentiment analysis, in both tasks. Our results identify two standard and two novel features (i.e., type-token ratio, frequency, sonority score, surprise) as most diagnostic in these tasks. By providing a simple tool applicable to both short poems and long novels generating quantitative predictions about features that co-determe the cognitive and affective processing of specific text categories or authors, our data pave the way for many future computational and empirical studies of literature or experiments in reading psychology.
Media plays an important role in shaping public opinion. Biased media can influence people in undesirable directions and hence should be unmasked as such. We observe that featurebased and neural text classification approaches which rely only on the distribution of low-level lexical information fail to detect media bias. This weakness becomes most noticeable for articles on new events, where words appear in new contexts and hence their "bias predictiveness" is unclear. In this paper, we therefore study how second-order information about biased statements in an article helps to improve detection effectiveness. In particular, we utilize the probability distributions of the frequency, positions, and sequential order of lexical and informational sentence-level bias in a Gaussian Mixture Model. On an existing media bias dataset, we find that the frequency and positions of biased statements strongly impact article-level bias, whereas their exact sequential order is secondary. Using a standard model for sentence-level bias detection, we provide empirical evidence that article-level bias detectors that use second-order information clearly outperform those without.
Automatic short answer scoring is one of the text classification problems to assess students' answers during exams automatically. Several challenges can arise in making an automatic short answer scoring system, one of which is the quantity and quality of the data. The data labeling process is not easy because it requires a human annotator who is an expert in their field. Further, the data imbalance process is also a challenge because the number of labels for correct answers is always much less than the wrong answers. In this paper, we propose the use of a stacking model based on neural network and XGBoost for classification process with sentence embedding feature. We also propose to use data upsampling method to handle imbalance classes and hyperparameters optimization algorithm to find a robust model automatically. We use Ukara 1.0 Challenge dataset and our best model obtained an F1-score of 0.821 exceeding the previous work at the same dataset.
We build a sentence-level political discourse classifier using existing human expert annotated corpora of political manifestos from the Manifestos Project (Volkens et al., 2020a) and applying them to a corpus ofCOVID-19Press Briefings (Chatsiou, 2020). We use manually annotated political manifestos as training data to train a local topic ConvolutionalNeural Network (CNN) classifier; then apply it to the COVID-19PressBriefings Corpus to automatically classify sentences in the test corpus.We report on a series of experiments with CNN trained on top of pre-trained embeddings for sentence-level classification tasks. We show thatCNN combined with transformers like BERT outperforms CNN combined with other embeddings (Word2Vec, Glove, ELMo) and that it is possible to use a pre-trained classifier to conduct automatic classification on different political texts without additional training.
Active learning strives to reduce annotation costs by choosing the most critical examples to label. Typically, the active learning strategy is contingent on the classification model. For instance, uncertainty sampling depends on poorly calibrated model confidence scores. In the cold-start setting, active learning is impractical because of model instability and data scarcity. Fortunately, modern NLP provides an additional source of information: pre-trained language models. The pre-training loss can find examples that surprise the model and should be labeled for efficient fine-tuning. Therefore, we treat the language modeling loss as a proxy for classification uncertainty. With BERT, we develop a simple strategy based on the masked language modeling loss that minimizes labeling costs for text classification. Compared to other baselines, our approach reaches higher accuracy within less sampling iterations and computation time.
In this paper, we provide an overview of the WNUT-2020 shared task on the identification of informative COVID-19 English Tweets. We describe how we construct a corpus of 10K Tweets and organize the development and evaluation phases for this task. In addition, we also present a brief summary of results obtained from the final system evaluation submissions of 55 teams, finding that (i) many systems obtain very high performance, up to 0.91 F1 score, (ii) the majority of the submissions achieve substantially higher results than the baseline fastText (Joulin et al., 2017), and (iii) fine-tuning pre-trained language models on relevant language data followed by supervised training performs well in this task.
This paper presents our submission to Task 2 of the Workshop on Noisy User-generated Text. We explore improving the performance of a pre-trained transformer-based language model fine-tuned for text classification through an ensemble implementation that makes use of corpus level information and a handcrafted feature. We test the effectiveness of including the aforementioned features in accommodating the challenges of a noisy data set centred on a specific subject outside the remit of the pre-training data. We show that inclusion of additional features can improve classification results and achieve a score within 2 points of the top performing team.
We propose several new models for semi-supervised nonnegative matrix factorization (SSNMF) and provide motivation for SSNMF models as maximum likelihood estimators given specific distributions of uncertainty. We present multiplicative updates training methods for each new model, and demonstrate the application of these models to classification, although they are flexible to other supervised learning tasks. We illustrate the promise of these models and training methods on both synthetic and real data, and achieve high classification accuracy on the 20 Newsgroups dataset.
A growing body of recent evidence has highlighted the limitations of natural language processing (NLP) datasets and classifiers. These include the presence of annotation artifacts in datasets, classifiers relying on shallow features like a single word (e.g., if a movie review has the word "romantic", the review tends to be positive), or unnecessary words (e.g., learning a proper noun to classify a movie as positive or negative). The presence of such artifacts has subsequently led to the development of challenging datasets to force the model to generalize better. While a variety of heuristic strategies, such as counterfactual examples and contrast sets, have been proposed, the theoretical justification about what makes these examples difficult for the classifier is often lacking or unclear. In this paper, using tools from information geometry, we propose a theoretical way to quantify the difficulty of an example in NLP. Using our approach, we explore difficult examples for several deep learning architectures. We discover that both BERT, CNN and fasttext are susceptible to word substitutions in high difficulty examples. These classifiers tend to perform poorly on the FIM test set. (generated by sampling and perturbing difficult examples, with accuracy dropping below 50%). We replicate our experiments on 5 NLP datasets (YelpReviewPolarity, AGNEWS, SogouNews, YelpReviewFull and Yahoo Answers). On YelpReviewPolarity we observe a correlation coefficient of -0.4 between resilience to perturbations and the difficulty score. Similarly we observe a correlation of 0.35 between the difficulty score and the empirical success probability of random substitutions. Our approach is simple, architecture agnostic and can be used to study the fragilities of text classification models. All the code used will be made publicly available, including a tool to explore the difficult examples for other datasets.
Despite transformers' impressive accuracy, their computational cost is often prohibitive to use with limited computational resources. Most previous approaches to improve inference efficiency require a separate model for each possible computational budget. In this paper, we extend PoWER-BERT (Goyal et al., 2020) and propose Length-Adaptive Transformer that can be used for various inference scenarios after one-shot training. We train a transformer with LengthDrop, a structural variant of dropout, which stochastically determines a sequence length at each layer. We then conduct a multi-objective evolutionary search to find a length configuration that maximizes the accuracy and minimizes the efficiency metric under any given computational budget. Additionally, we significantly extend the applicability of PoWER-BERT beyond sequence-level classification into token-level classification with Drop-and-Restore process that drops word-vectors temporarily in intermediate layers and restores at the last layer if necessary. We empirically verify the utility of the proposed approach by demonstrating the superior accuracy-efficiency trade-off under various setups, including span-based question answering and text classification. Code is available at https://github.com/clovaai/length-adaptive-transformer.
In this work, we present a novel approach for simultaneous knowledge transfer and model compression called Weight Squeezing. With this method, we perform knowledge transfer from a teacher model by learning the mapping from its weights to smaller student model weights. We applied Weight Squeezing to a pre-trained text classification model based on BERT-Medium model and compared our method to various other knowledge transfer and model compression methods on GLUE multitask benchmark. We observed that our approach produces better results while being significantly faster than other methods for training student models. We also proposed a variant of Weight Squeezing called Gated Weight Squeezing, for which we combined fine-tuning of BERT-Medium model and learning mapping from BERT-Base weights. We showed that fine-tuning with Gated Weight Squeezing outperforms plain fine-tuning of BERT-Medium model as well as other concurrent SoTA approaches while much being easier to implement.