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} .
We study the problem of weakly supervised text classification, which aims to classify text documents into a set of pre-defined categories with category surface names only and without any annotated training document provided. Most existing classifiers leverage textual information in each document. However, in many domains, documents are accompanied by various types of metadata (e.g., authors, venue, and year of a research paper). These metadata and their combinations may serve as strong category indicators in addition to textual contents. In this paper, we explore the potential of using metadata to help weakly supervised text classification. To be specific, we model the relationships between documents and metadata via a heterogeneous information network. To effectively capture higher-order structures in the network, we use motifs to describe metadata combinations. We propose a novel framework, named MotifClass, which (1) selects category-indicative motif instances, (2) retrieves and generates pseudo-labeled training samples based on category names and indicative motif instances, and (3) trains a text classifier using the pseudo training data. Extensive experiments on real-world datasets demonstrate the superior performance of MotifClass to existing weakly supervised text classification approaches. Further analysis shows the benefit of considering higher-order metadata information in our framework.
The rapid proliferation of machine learning models across domains and deployment settings has given rise to various communities (e.g. industry practitioners) which seek to benchmark models across tasks and objectives of personal value. Unfortunately, these users cannot use standard benchmark results to perform such value-driven comparisons as traditional benchmarks evaluate models on a single objective (e.g. average accuracy) and fail to facilitate a standardized training framework that controls for confounding variables (e.g. computational budget), making fair comparisons difficult. To address these challenges, we introduce the open-source Ludwig Benchmarking Toolkit (LBT), a personalized benchmarking toolkit for running end-to-end benchmark studies (from hyperparameter optimization to evaluation) across an easily extensible set of tasks, deep learning models, datasets and evaluation metrics. LBT provides a configurable interface for controlling training and customizing evaluation, a standardized training framework for eliminating confounding variables, and support for multi-objective evaluation. We demonstrate how LBT can be used to create personalized benchmark studies with a large-scale comparative analysis for text classification across 7 models and 9 datasets. We explore the trade-offs between inference latency and performance, relationships between dataset attributes and performance, and the effects of pretraining on convergence and robustness, showing how LBT can be used to satisfy various benchmarking objectives.
We introduce a new language representation model in finance called Financial Embedding Analysis of Sentiment (FinEAS). In financial markets, news and investor sentiment are significant drivers of security prices. Thus, leveraging the capabilities of modern NLP approaches for financial sentiment analysis is a crucial component in identifying patterns and trends that are useful for market participants and regulators. In recent years, methods that use transfer learning from large Transformer-based language models like BERT, have achieved state-of-the-art results in text classification tasks, including sentiment analysis using labelled datasets. Researchers have quickly adopted these approaches to financial texts, but best practices in this domain are not well-established. In this work, we propose a new model for financial sentiment analysis based on supervised fine-tuned sentence embeddings from a standard BERT model. We demonstrate our approach achieves significant improvements in comparison to vanilla BERT, LSTM, and FinBERT, a financial domain specific BERT.
Short text classification is a fundamental task in natural language processing. It is hard due to the lack of context information and labeled data in practice. In this paper, we propose a new method called SHINE, which is based on graph neural network (GNN), for short text classification. First, we model the short text dataset as a hierarchical heterogeneous graph consisting of word-level component graphs which introduce more semantic and syntactic information. Then, we dynamically learn a short document graph that facilitates effective label propagation among similar short texts. Thus, compared with existing GNN-based methods, SHINE can better exploit interactions between nodes of the same types and capture similarities between short texts. Extensive experiments on various benchmark short text datasets show that SHINE consistently outperforms state-of-the-art methods, especially with fewer labels.
The deep learning field is growing rapidly as witnessed by the exponential growth of papers submitted to journals, conferences, and pre-print servers. To cope with the sheer number of papers, several text mining tools from natural language processing (NLP) have been proposed that enable researchers to keep track of recent findings. In this context, our paper makes two main contributions: first, we collected and annotated a dataset of papers paired by title and sub-field from the field of artificial intelligence (AI), and, second, we present results on how to predict a paper's AI sub-field from a given paper title only. Importantly, for the latter, short-text classification task we compare several algorithms from conventional machine learning all the way up to recent, larger transformer architectures. Finally, for the transformer models, we also present gradient-based, attention visualizations to further explain the model's classification process. All code can be found at \url{https://github.com/1pha/paperswithtopic}
\emph{Funnelling} (Fun) is a recently proposed method for cross-lingual text classification (CLTC) based on a two-tier learning ensemble for heterogeneous transfer learning (HTL). In this ensemble method, 1st-tier classifiers, each working on a different and language-dependent feature space, return a vector of calibrated posterior probabilities (with one dimension for each class) for each document, and the final classification decision is taken by a metaclassifier that uses this vector as its input. The metaclassifier can thus exploit class-class correlations, and this (among other things) gives Fun an edge over CLTC systems in which these correlations cannot be brought to bear. In this paper we describe \emph{Generalized Funnelling} (gFun), a generalization of Fun consisting of an HTL architecture in which 1st-tier components can be arbitrary \emph{view-generating functions}, i.e., language-dependent functions that each produce a language-independent representation ("view") of the (monolingual) document. We describe an instance of gFun in which the metaclassifier receives as input a vector of calibrated posterior probabilities (as in Fun) aggregated to other embedded representations that embody other types of correlations, such as word-class correlations (as encoded by \emph{Word-Class Embeddings}), word-word correlations (as encoded by \emph{Multilingual Unsupervised or Supervised Embeddings}), and word-context correlations (as encoded by \emph{multilingual BERT}). We show that this instance of \textsc{gFun} substantially improves over Fun and over state-of-the-art baselines, by reporting experimental results obtained on two large, standard datasets for multilingual multilabel text classification. Our code that implements gFun is publicly available.
We study acquisition functions for active learning (AL) for text classification. The Expected Loss Reduction (ELR) method focuses on a Bayesian estimate of the reduction in classification error, recently updated with Mean Objective Cost of Uncertainty (MOCU). We convert the ELR framework to estimate the increase in (strictly proper) scores like log probability or negative mean square error, which we call Bayesian Estimate of Mean Proper Scores (BEMPS). We also prove convergence results borrowing techniques used with MOCU. In order to allow better experimentation with the new acquisition functions, we develop a complementary batch AL algorithm, which encourages diversity in the vector of expected changes in scores for unlabelled data. To allow high performance text classifiers, we combine ensembling and dynamic validation set construction on pretrained language models. Extensive experimental evaluation then explores how these different acquisition functions perform. The results show that the use of mean square error and log probability with BEMPS yields robust acquisition functions, which consistently outperform the others tested.
The premises of an argument give evidence or other reasons to support a conclusion. However, the amount of support required depends on the generality of a conclusion, the nature of the individual premises, and similar. An argument whose premises make its conclusion rationally worthy to be drawn is called sufficient in argument quality research. Previous work tackled sufficiency assessment as a standard text classification problem, not modeling the inherent relation of premises and conclusion. In this paper, we hypothesize that the conclusion of a sufficient argument can be generated from its premises. To study this hypothesis, we explore the potential of assessing sufficiency based on the output of large-scale pre-trained language models. Our best model variant achieves an F1-score of .885, outperforming the previous state-of-the-art and being on par with human experts. While manual evaluation reveals the quality of the generated conclusions, their impact remains low ultimately.
Over the recent years, large pretrained language models (LM) have revolutionized the field of natural language processing (NLP). However, while pretraining on general language has been shown to work very well for common language, it has been observed that niche language poses problems. In particular, climate-related texts include specific language that common LMs can not represent accurately. We argue that this shortcoming of today's LMs limits the applicability of modern NLP to the broad field of text processing of climate-related texts. As a remedy, we propose CLIMATEBERT, a transformer-based language model that is further pretrained on over 2 million paragraphs of climate-related texts, crawled from various sources such as common news, research articles, and climate reporting of companies. We find that CLIMATEBERT leads to a 48% improvement on a masked language model objective which, in turn, leads to lowering error rates by 3.57% to 35.71% for various climate-related downstream tasks like text classification, sentiment analysis, and fact-checking.
In natural language processing (NLP), state-of-the-art (SOTA) semi-supervised learning (SSL) frameworks have shown great performance on deep pre-trained language models such as BERT, and are expected to significantly reduce the demand for manual labeling. However, our empirical studies indicate that these frameworks are not suitable for lightweight models such as TextCNN, LSTM and etc. In this work, we develop a new SSL framework called FLiText, which stands for Faster and Lighter semi-supervised Text classification. FLiText introduces an inspirer network together with the consistency regularization framework, which leverages a generalized regular constraint on the lightweight models for efficient SSL. As a result, FLiText obtains new SOTA performance for lightweight models across multiple SSL benchmarks on text classification. Compared with existing SOTA SSL methods on TextCNN, FLiText improves the accuracy of lightweight model TextCNN from 51.00% to 90.49% on IMDb, 39.8% to 58.06% on Yelp-5, and from 55.3% to 65.08% on Yahoo. In addition, compared with the fully supervised method on the full dataset, FLiText just uses less than 1% of labeled data to improve the accuracy by 6.59%, 3.94%, and 3.22% on the datasets of IMDb, Yelp-5, and Yahoo respectively.
Nowadays, data analysis has become a problem as the amount of data is constantly increasing. In order to overcome this problem in textual data, many models and methods are used in natural language processing. The topic modeling field is one of these methods. Topic modeling allows determining the semantic structure of a text document. Latent Dirichlet Allocation (LDA) is the most common method among topic modeling methods. In this article, the proposed n-stage LDA method, which can enable the LDA method to be used more effectively, is explained in detail. The positive effect of the method has been demonstrated by the applied English and Turkish studies. Since the method focuses on reducing the word count in the dictionary, it can be used language-independently. You can access the open-source code of the method and the example: https://github.com/anil1055/n-stage_LDA
Recent advances in prompt-based learning have shown strong results on few-shot text classification by using cloze-style templates. Similar attempts have been made on named entity recognition (NER) which manually design templates to predict entity types for every text span in a sentence. However, such methods may suffer from error propagation induced by entity span detection, high cost due to enumeration of all possible text spans, and omission of inter-dependencies among token labels in a sentence. Here we present a simple demonstration-based learning method for NER, which lets the input be prefaced by task demonstrations for in-context learning. We perform a systematic study on demonstration strategy regarding what to include (entity examples, with or without surrounding context), how to select the examples, and what templates to use. Results on in-domain learning and domain adaptation show that the model's performance in low-resource settings can be largely improved with a suitable demonstration strategy (e.g., a 4-17% improvement on 25 train instances). We also find that good demonstration can save many labeled examples and consistency in demonstration contributes to better performance.
Distilling state-of-the-art transformer models into lightweight student models is an effective way to reduce computation cost at inference time. The student models are typically compact transformers with fewer parameters, while expensive operations such as self-attention persist. Therefore, the improved inference speed may still be unsatisfactory for real-time or high-volume use cases. In this paper, we aim to further push the limit of inference speed by distilling teacher models into bigger, sparser student models -- bigger in that they scale up to billions of parameters; sparser in that most of the model parameters are n-gram embeddings. Our experiments on six single-sentence text classification tasks show that these student models retain 97% of the RoBERTa-Large teacher performance on average, and meanwhile achieve up to 600x speed-up on both GPUs and CPUs at inference time. Further investigation reveals that our pipeline is also helpful for sentence-pair classification tasks, and in domain generalization settings.
The goal of meta-learning is to learn to adapt to a new task with only a few labeled examples. To tackle this problem in NLP, we propose $\textit{in-context tuning}$, which recasts adaptation and prediction as a simple sequence prediction problem: to form the input sequence, we concatenate the task instruction, the labeled examples, and the target input to predict; to meta-train the model to learn from in-context examples, we fine-tune a pre-trained language model (LM) to predict the target label from the input sequences on a collection of tasks. We benchmark our method on two collections of text classification tasks: LAMA and BinaryClfs. Compared to first-order MAML which adapts the model with gradient descent, our method better leverages the inductive bias of LMs to perform pattern matching, and outperforms MAML by an absolute $6\%$ AUC ROC score on BinaryClfs, with increasing advantage w.r.t. model size. Compared to non-fine-tuned in-context learning (i.e. prompting a raw LM), in-context tuning directly learns to learn from in-context examples. On BinaryClfs, in-context tuning improves the average AUC-ROC score by an absolute $10\%$, and reduces the variance with respect to example ordering by 6x and example choices by 2x.
Identifying vulnerabilities in the source code is essential to protect the software systems from cyber security attacks. It, however, is also a challenging step that requires specialized expertise in security and code representation. To this end, we aim to develop a general, practical, and programming language-independent model capable of running on various source codes and libraries without difficulty. Therefore, we consider vulnerability detection as an inductive text classification problem and propose ReGVD, a simple yet effective graph neural network-based model for the problem. In particular, ReGVD views each raw source code as a flat sequence of tokens to build a graph, wherein node features are initialized by only the token embedding layer of a pre-trained programming language (PL) model. ReGVD then leverages residual connection among GNN layers and examines a mixture of graph-level sum and max poolings to return a graph embedding for the source code. ReGVD outperforms the existing state-of-the-art models and obtains the highest accuracy on the real-world benchmark dataset from CodeXGLUE for vulnerability detection. Our code is available at: \url{https://github.com/daiquocnguyen/GNN-ReGVD}.
Automated essay scoring (AES) is gaining increasing attention in the education sector as it significantly reduces the burden of manual scoring and allows ad hoc feedback for learners. Natural language processing based on machine learning has been shown to be particularly suitable for text classification and AES. While many machine-learning approaches for AES still rely on a bag-of-words (BOW) approach, we consider a transformer-based approach in this paper, compare its performance to a logistic regression model based on the BOW approach and discuss their differences. The analysis is based on 2,088 email responses to a problem-solving task, that were manually labeled in terms of politeness. Both transformer models considered in that analysis outperformed without any hyper-parameter tuning the regression-based model. We argue that for AES tasks such as politeness classification, the transformer-based approach has significant advantages, while a BOW approach suffers from not taking word order into account and reducing the words to their stem. Further, we show how such models can help increase the accuracy of human raters, and we provide a detailed instruction on how to implement transformer-based models for one's own purpose.
Semi-supervised learning (SSL) has demonstrated its potential to improve the model accuracy for a variety of learning tasks when the high-quality supervised data is severely limited. Although it is often established that the average accuracy for the entire population of data is improved, it is unclear how SSL fares with different sub-populations. Understanding the above question has substantial fairness implications when different sub-populations are defined by the demographic groups that we aim to treat fairly. In this paper, we reveal the disparate impacts of deploying SSL: the sub-population who has a higher baseline accuracy without using SSL (the "rich" one) tends to benefit more from SSL; while the sub-population who suffers from a low baseline accuracy (the "poor" one) might even observe a performance drop after adding the SSL module. We theoretically and empirically establish the above observation for a broad family of SSL algorithms, which either explicitly or implicitly use an auxiliary "pseudo-label". Experiments on a set of image and text classification tasks confirm our claims. We introduce a new metric, Benefit Ratio, and promote the evaluation of the fairness of SSL (Equalized Benefit Ratio). We further discuss how the disparate impact can be mitigated. We hope our paper will alarm the potential pitfall of using SSL and encourage a multifaceted evaluation of future SSL algorithms.
Fine-tuning large pre-trained language models on downstream tasks has become the de-facto learning paradigm in NLP. However, conventional approaches fine-tune all the parameters of the pre-trained model, which becomes prohibitive as the model size and the number of tasks grow. Recent work has proposed a variety of parameter-efficient transfer learning methods that only fine-tune a small number of (extra) parameters to attain strong performance. While effective, the critical ingredients for success and the connections among the various methods are poorly understood. In this paper, we break down the design of state-of-the-art parameter-efficient transfer learning methods and present a unified framework that establishes connections between them. Specifically, we re-frame them as modifications to specific hidden states in pre-trained models, and define a set of design dimensions along which different methods vary, such as the function to compute the modification and the position to apply the modification. Through comprehensive empirical studies across machine translation, text summarization, language understanding, and text classification benchmarks, we utilize the unified view to identify important design choices in previous methods. Furthermore, our unified framework enables the transfer of design elements across different approaches, and as a result we are able to instantiate new parameter-efficient fine-tuning methods that tune less parameters than previous methods while being more effective, achieving comparable results to fine-tuning all parameters on all four tasks.
This paper describes the system submitted to Dravidian-Codemix-HASOC2021: Hate Speech and Offensive Language Identification in Dravidian Languages (Tamil-English and Malayalam-English). This task aims to identify offensive content in code-mixed comments/posts in Dravidian Languages collected from social media. Our approach utilizes pooling the last layers of pretrained transformer multilingual BERT for this task which helped us achieve rank nine on the leaderboard with a weighted average score of 0.61 for the Tamil-English dataset in subtask B. After the task deadline, we sampled the dataset uniformly and used the MuRIL pretrained model, which helped us achieve a weighted average score of 0.67, the top score in the leaderboard. Furthermore, our approach to utilizing the pretrained models helps reuse our models for the same task with a different dataset. Our code and models are available in https://github.com/seanbenhur/tanglish-offensive-language-identification