We explore threshold vocabulary trimming in Byte-Pair Encoding subword tokenization, a tokenization postprocessing step that replaces rare subwords with their component subwords. The technique is available in popular tokenization libraries but has not been subjected to rigorous scientific scrutiny. While the removal of rare subwords is suggested as best practice in model implementations, both as a means to reduce model size and for improving model performance through robustness, our experiments indicate that, across a large space of hyperparameter settings, vocabulary trimming fails to consistently improve model performance, and is even prone to incurring heavy degradation.

Subword regularizations use multiple subword segmentations during training to improve the robustness of neural machine translation models. In previous subword regularizations, we use multiple segmentations in the training process but use only one segmentation in the inference. In this study, we propose an inference strategy to address this discrepancy. The proposed strategy approximates the marginalized likelihood by using multiple segmentations including the most plausible segmentation and several sampled segmentations. Because the proposed strategy aggregates predictions from several segmentations, we can regard it as a single model ensemble that does not require any additional cost for training. Experimental results show that the proposed strategy improves the performance of models trained with subword regularization in low-resource machine translation tasks.

We present two simple modifications for word-level perturbation: Word Replacement considering Length (WR-L) and Compositional Word Replacement (CWR).In conventional word replacement, a word in an input is replaced with a word sampled from the entire vocabulary, regardless of the length and context of the target word.WR-L considers the length of a target word by sampling words from the Poisson distribution.CWR considers the compositional candidates by restricting the source of sampling to related words that appear in subword regularization. Experimental results showed that the combination of WR-L and CWR improved the performance of text classification and machine translation.

We present a subword regularization method for WordPiece, which uses a maximum matching algorithm for tokenization. The proposed method, MaxMatch-Dropout, randomly drops words in a search using the maximum matching algorithm. It realizes finetuning with subword regularization for popular pretrained language models such as BERT-base. The experimental results demonstrate that MaxMatch-Dropout improves the performance of text classification and machine translation tasks as well as other subword regularization methods. Moreover, we provide a comparative analysis of subword regularization methods: subword regularization with SentencePiece (Unigram), BPE-Dropout, and MaxMatch-Dropout.

This study introduces a novel approach to the joint extraction of entities and relations by stacking convolutional neural networks (CNNs) on pretrained language models. We adopt table representations to model the entities and relations, casting the entity and relation extraction as a table-labeling problem. Regarding each table as an image and each cell in a table as an image pixel, we apply two-dimensional CNNs to the tables to capture local dependencies and predict the cell labels. The experimental results showed that the performance of the proposed method is comparable to those of current state-of-art systems on the CoNLL04, ACE05, and ADE datasets. Even when freezing pretrained language model parameters, the proposed method showed a stable performance, whereas the compared methods suffered from significant decreases in performance. This observation indicates that the parameters of the pretrained encoder may incorporate dependencies among the entity and relation labels during fine-tuning.

In traditional NLP, we tokenize a given sentence as a preprocessing, and thus the tokenization is unrelated to a target downstream task. To address this issue, we propose a novel method to explore a tokenization which is appropriate for the downstream task. Our proposed method, optimizing tokenization (OpTok), is trained to assign a high probability to such appropriate tokenization based on the downstream task loss. OpTok can be used for any downstream task which uses a vector representation of a sentence such as text classification. Experimental results demonstrate that OpTok improves the performance of sentiment analysis and textual entailment. In addition, we introduce OpTok into BERT, the state-of-the-art contextualized embeddings and report a positive effect.

For unsegmented languages such as Japanese and Chinese, tokenization of a sentence has a significant impact on the performance of text classification. Sentences are usually segmented with words or subwords by a morphological analyzer or byte pair encoding and then encoded with word (or subword) representations for neural networks. However, segmentation is potentially ambiguous, and it is unclear whether the segmented tokens achieve the best performance for the target task. In this paper, we propose a method to simultaneously learn tokenization and text classification to address these problems. Our model incorporates a language model for unsupervised tokenization into a text classifier and then trains both models simultaneously. To make the model robust against infrequent tokens, we sampled segmentation for each sentence stochastically during training, which resulted in improved performance of text classification. We conducted experiments on sentiment analysis as a text classification task and show that our method achieves better performance than previous methods.