PyTorchKR Papers

최신 논문

총 2,437편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2010.10267

Text Classification of Manifestos and COVID-19 Press Briefings using BERT and Convolutional Neural Networks

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.

arXiv:2010.09535공식 코드 · Papers with Code

Cold-start Active Learning through Self-supervised Language Modeling

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.

arXiv:2010.07988공식 코드 · Papers with Code

CXP949 at WNUT-2020 Task 2: Extracting Informative COVID-19 Tweets -- RoBERTa Ensembles and The Continued Relevance of Handcrafted Features

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.

arXiv:2010.07212

Geometry matters: Exploring language examples at the decision boundary

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.

arXiv:2010.07003공식 코드 · Papers with Code

Length-Adaptive Transformer: Train Once with Length Drop, Use Anytime with Search

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.

arXiv:2010.06993

Weight Squeezing: Reparameterization for Knowledge Transfer and Model Compression

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.

arXiv:2010.07557

Token Sequence Labeling vs. Clause Classification for English Emotion Stimulus Detection

Emotion stimulus detection is the task of finding the cause of an emotion in a textual description, similar to target or aspect detection for sentiment analysis. Previous work approached this in three ways, namely (1) as text classification into an inventory of predefined possible stimuli ("Is the stimulus category A or B?"), (2) as sequence labeling of tokens ("Which tokens describe the stimulus?"), and (3) as clause classification ("Does this clause contain the emotion stimulus?"). So far, setting (3) has been evaluated broadly on Mandarin and (2) on English, but no comparison has been performed. Therefore, we aim to answer whether clause classification or sequence labeling is better suited for emotion stimulus detection in English. To accomplish that, we propose an integrated framework which enables us to evaluate the two different approaches comparably, implement models inspired by state-of-the-art approaches in Mandarin, and test them on four English data sets from different domains. Our results show that sequence labeling is superior on three out of four datasets, in both clause-based and sequence-based evaluation. The only case in which clause classification performs better is one data set with a high density of clause annotations. Our error analysis further confirms quantitatively and qualitatively that clauses are not the appropriate stimulus unit in English.

arXiv:2010.07245공식 코드 · Papers with Code

Text Classification Using Label Names Only: A Language Model Self-Training Approach

Current text classification methods typically require a good number of human-labeled documents as training data, which can be costly and difficult to obtain in real applications. Humans can perform classification without seeing any labeled examples but only based on a small set of words describing the categories to be classified. In this paper, we explore the potential of only using the label name of each class to train classification models on unlabeled data, without using any labeled documents. We use pre-trained neural language models both as general linguistic knowledge sources for category understanding and as representation learning models for document classification. Our method (1) associates semantically related words with the label names, (2) finds category-indicative words and trains the model to predict their implied categories, and (3) generalizes the model via self-training. We show that our model achieves around 90% accuracy on four benchmark datasets including topic and sentiment classification without using any labeled documents but learning from unlabeled data supervised by at most 3 words (1 in most cases) per class as the label name.

arXiv:2010.06812공식 코드 · Papers with Code

Explain2Attack: Text Adversarial Attacks via Cross-Domain Interpretability

Training robust deep learning models for down-stream tasks is a critical challenge. Research has shown that down-stream models can be easily fooled with adversarial inputs that look like the training data, but slightly perturbed, in a way imperceptible to humans. Understanding the behavior of natural language models under these attacks is crucial to better defend these models against such attacks. In the black-box attack setting, where no access to model parameters is available, the attacker can only query the output information from the targeted model to craft a successful attack. Current black-box state-of-the-art models are costly in both computational complexity and number of queries needed to craft successful adversarial examples. For real world scenarios, the number of queries is critical, where less queries are desired to avoid suspicion towards an attacking agent. In this paper, we propose Explain2Attack, a black-box adversarial attack on text classification task. Instead of searching for important words to be perturbed by querying the target model, Explain2Attack employs an interpretable substitute model from a similar domain to learn word importance scores. We show that our framework either achieves or out-performs attack rates of the state-of-the-art models, yet with lower queries cost and higher efficiency.

arXiv:2010.05990

Chatbot Interaction with Artificial Intelligence: Human Data Augmentation with T5 and Language Transformer Ensemble for Text Classification

In this work, we present the Chatbot Interaction with Artificial Intelligence (CI-AI) framework as an approach to the training of deep learning chatbots for task classification. The intelligent system augments human-sourced data via artificial paraphrasing in order to generate a large set of training data for further classical, attention, and language transformation-based learning approaches for Natural Language Processing. Human beings are asked to paraphrase commands and questions for task identification for further execution of a machine. The commands and questions are split into training and validation sets. A total of 483 responses were recorded. Secondly, the training set is paraphrased by the T5 model in order to augment it with further data. Seven state-of-the-art transformer-based text classification algorithms (BERT, DistilBERT, RoBERTa, DistilRoBERTa, XLM, XLM-RoBERTa, and XLNet) are benchmarked for both sets after fine-tuning on the training data for two epochs. We find that all models are improved when training data is augmented by the T5 model, with an average increase of classification accuracy by 4.01%. The best result was the RoBERTa model trained on T5 augmented data which achieved 98.96% classification accuracy. Finally, we found that an ensemble of the five best-performing transformer models via Logistic Regression of output label predictions led to an accuracy of 99.59% on the dataset of human responses. A highly-performing model allows the intelligent system to interpret human commands at the social-interaction level through a chatbot-like interface (e.g. "Robot, can we have a conversation?") and allows for better accessibility to AI by non-technical users.

arXiv:2010.06572

Does my multimodal model learn cross-modal interactions? It's harder to tell than you might think!

Modeling expressive cross-modal interactions seems crucial in multimodal tasks, such as visual question answering. However, sometimes high-performing black-box algorithms turn out to be mostly exploiting unimodal signals in the data. We propose a new diagnostic tool, empirical multimodally-additive function projection (EMAP), for isolating whether or not cross-modal interactions improve performance for a given model on a given task. This function projection modifies model predictions so that cross-modal interactions are eliminated, isolating the additive, unimodal structure. For seven image+text classification tasks (on each of which we set new state-of-the-art benchmarks), we find that, in many cases, removing cross-modal interactions results in little to no performance degradation. Surprisingly, this holds even when expressive models, with capacity to consider interactions, otherwise outperform less expressive models; thus, performance improvements, even when present, often cannot be attributed to consideration of cross-modal feature interactions. We hence recommend that researchers in multimodal machine learning report the performance not only of unimodal baselines, but also the EMAP of their best-performing model.

arXiv:2010.04883

Adversarial Self-Supervised Data-Free Distillation for Text Classification

Large pre-trained transformer-based language models have achieved impressive results on a wide range of NLP tasks. In the past few years, Knowledge Distillation(KD) has become a popular paradigm to compress a computationally expensive model to a resource-efficient lightweight model. However, most KD algorithms, especially in NLP, rely on the accessibility of the original training dataset, which may be unavailable due to privacy issues. To tackle this problem, we propose a novel two-stage data-free distillation method, named Adversarial self-Supervised Data-Free Distillation (AS-DFD), which is designed for compressing large-scale transformer-based models (e.g., BERT). To avoid text generation in discrete space, we introduce a Plug & Play Embedding Guessing method to craft pseudo embeddings from the teacher's hidden knowledge. Meanwhile, with a self-supervised module to quantify the student's ability, we adapt the difficulty of pseudo embeddings in an adversarial training manner. To the best of our knowledge, our framework is the first data-free distillation framework designed for NLP tasks. We verify the effectiveness of our method on several text classification datasets.

arXiv:2010.05419

Gradient-based Analysis of NLP Models is Manipulable

Gradient-based analysis methods, such as saliency map visualizations and adversarial input perturbations, have found widespread use in interpreting neural NLP models due to their simplicity, flexibility, and most importantly, their faithfulness. In this paper, however, we demonstrate that the gradients of a model are easily manipulable, and thus bring into question the reliability of gradient-based analyses. In particular, we merge the layers of a target model with a Facade that overwhelms the gradients without affecting the predictions. This Facade can be trained to have gradients that are misleading and irrelevant to the task, such as focusing only on the stop words in the input. On a variety of NLP tasks (text classification, NLI, and QA), we show that our method can manipulate numerous gradient-based analysis techniques: saliency maps, input reduction, and adversarial perturbations all identify unimportant or targeted tokens as being highly important. The code and a tutorial of this paper is available at http://ucinlp.github.io/facade.

arXiv:2010.05244공식 코드 · Papers with Code

Advanced Dropout: A Model-free Methodology for Bayesian Dropout Optimization

Due to lack of data, overfitting ubiquitously exists in real-world applications of deep neural networks (DNNs). We propose advanced dropout, a model-free methodology, to mitigate overfitting and improve the performance of DNNs. The advanced dropout technique applies a model-free and easily implemented distribution with parametric prior, and adaptively adjusts dropout rate. Specifically, the distribution parameters are optimized by stochastic gradient variational Bayes in order to carry out an end-to-end training. We evaluate the effectiveness of the advanced dropout against nine dropout techniques on seven computer vision datasets (five small-scale datasets and two large-scale datasets) with various base models. The advanced dropout outperforms all the referred techniques on all the datasets.We further compare the effectiveness ratios and find that advanced dropout achieves the highest one on most cases. Next, we conduct a set of analysis of dropout rate characteristics, including convergence of the adaptive dropout rate, the learned distributions of dropout masks, and a comparison with dropout rate generation without an explicit distribution. In addition, the ability of overfitting prevention is evaluated and confirmed. Finally, we extend the application of the advanced dropout to uncertainty inference, network pruning, text classification, and regression. The proposed advanced dropout is also superior to the corresponding referred methods. Codes are available at https://github.com/PRIS-CV/AdvancedDropout.

arXiv:2010.05223

End to End Binarized Neural Networks for Text Classification

Deep neural networks have demonstrated their superior performance in almost every Natural Language Processing task, however, their increasing complexity raises concerns. In particular, these networks require high expenses on computational hardware, and training budget is a concern for many. Even for a trained network, the inference phase can be too demanding for resource-constrained devices, thus limiting its applicability. The state-of-the-art transformer models are a vivid example. Simplifying the computations performed by a network is one way of relaxing the complexity requirements. In this paper, we propose an end to end binarized neural network architecture for the intent classification task. In order to fully utilize the potential of end to end binarization, both input representations (vector embeddings of tokens statistics) and the classifier are binarized. We demonstrate the efficiency of such architecture on the intent classification of short texts over three datasets and for text classification with a larger dataset. The proposed architecture achieves comparable to the state-of-the-art results on standard intent classification datasets while utilizing ~ 20-40% lesser memory and training time. Furthermore, the individual components of the architecture, such as binarized vector embeddings of documents or binarized classifiers, can be used separately with not necessarily fully binary architectures.

arXiv:2010.05194

Detecting Foodborne Illness Complaints in Multiple Languages Using English Annotations Only

Health departments have been deploying text classification systems for the early detection of foodborne illness complaints in social media documents such as Yelp restaurant reviews. Current systems have been successfully applied for documents in English and, as a result, a promising direction is to increase coverage and recall by considering documents in additional languages, such as Spanish or Chinese. Training previous systems for more languages, however, would be expensive, as it would require the manual annotation of many documents for each new target language. To address this challenge, we consider cross-lingual learning and train multilingual classifiers using only the annotations for English-language reviews. Recent zero-shot approaches based on pre-trained multi-lingual BERT (mBERT) have been shown to effectively align languages for aspects such as sentiment. Interestingly, we show that those approaches are less effective for capturing the nuances of foodborne illness, our public health application of interest. To improve performance without extra annotations, we create artificial training documents in the target language through machine translation and train mBERT jointly for the source (English) and target language. Furthermore, we show that translating labeled documents to multiple languages leads to additional performance improvements for some target languages. We demonstrate the benefits of our approach through extensive experiments with Yelp restaurant reviews in seven languages. Our classifiers identify foodborne illness complaints in multilingual reviews from the Yelp Challenge dataset, which highlights the potential of our general approach for deployment in health departments.

arXiv:2010.05763

Layer-wise Guided Training for BERT: Learning Incrementally Refined Document Representations

Although BERT is widely used by the NLP community, little is known about its inner workings. Several attempts have been made to shed light on certain aspects of BERT, often with contradicting conclusions. A much raised concern focuses on BERT's over-parameterization and under-utilization issues. To this end, we propose o novel approach to fine-tune BERT in a structured manner. Specifically, we focus on Large Scale Multilabel Text Classification (LMTC) where documents are assigned with one or more labels from a large predefined set of hierarchically organized labels. Our approach guides specific BERT layers to predict labels from specific hierarchy levels. Experimenting with two LMTC datasets we show that this structured fine-tuning approach not only yields better classification results but also leads to better parameter utilization.

arXiv:2010.04609

Causal Feature Selection with Dimension Reduction for Interpretable Text Classification

Text features that are correlated with class labels, but do not directly cause them, are sometimesuseful for prediction, but they may not be insightful. As an alternative to traditional correlation-basedfeature selection, causal inference could reveal more principled, meaningful relationships betweentext features and labels. To help researchers gain insight into text data, e.g. for social scienceapplications, in this paper we investigate a class of matching-based causal inference methods fortext feature selection. Features used in document classification are often high dimensional, howeverexisting causal feature selection methods use Propensity Score Matching (PSM) which is known to beless effective in high-dimensional spaces. We propose a new causal feature selection framework thatcombines dimension reduction with causal inference to improve text feature selection. Experiments onboth synthetic and real-world data demonstrate the promise of our methods in improving classificationand enhancing interpretability.

arXiv:2010.04582공식 코드 · Papers with Code

Denoising Multi-Source Weak Supervision for Neural Text Classification

We study the problem of learning neural text classifiers without using any labeled data, but only easy-to-provide rules as multiple weak supervision sources. This problem is challenging because rule-induced weak labels are often noisy and incomplete. To address these two challenges, we design a label denoiser, which estimates the source reliability using a conditional soft attention mechanism and then reduces label noise by aggregating rule-annotated weak labels. The denoised pseudo labels then supervise a neural classifier to predicts soft labels for unmatched samples, which address the rule coverage issue. We evaluate our model on five benchmarks for sentiment, topic, and relation classifications. The results show that our model outperforms state-of-the-art weakly-supervised and semi-supervised methods consistently, and achieves comparable performance with fully-supervised methods even without any labeled data. Our code can be found at https://github.com/weakrules/Denoise-multi-weak-sources.

arXiv:2010.03648

A Mathematical Exploration of Why Language Models Help Solve Downstream Tasks

Autoregressive language models, pretrained using large text corpora to do well on next word prediction, have been successful at solving many downstream tasks, even with zero-shot usage. However, there is little theoretical understanding of this success. This paper initiates a mathematical study of this phenomenon for the downstream task of text classification by considering the following questions: (1) What is the intuitive connection between the pretraining task of next word prediction and text classification? (2) How can we mathematically formalize this connection and quantify the benefit of language modeling? For (1), we hypothesize, and verify empirically, that classification tasks of interest can be reformulated as sentence completion tasks, thus making language modeling a meaningful pretraining task. With a mathematical formalization of this hypothesis, we make progress towards (2) and show that language models that are $\epsilon$-optimal in cross-entropy (log-perplexity) learn features that can linearly solve such classification tasks with $\mathcal{O}(\sqrt{\epsilon})$ error, thus demonstrating that doing well on language modeling can be beneficial for downstream tasks. We experimentally verify various assumptions and theoretical findings, and also use insights from the analysis to design a new objective function that performs well on some classification tasks.