Relation extraction (RE) plays an important role in extracting knowledge from unstructured text but requires a large amount of labeled corpus. To reduce the expensive annotation efforts, semisupervised learning aims to leverage both labeled and unlab
eled data. In this paper, we review and compare three typical methods in semi-supervised RE with deep learning or meta-learning: self-ensembling, which forces consistent under perturbations but may confront insufficient supervision; self-training, which iteratively generates pseudo labels and retrain itself with the enlarged labeled set; dual learning, which leverages a primal task and a dual task to give mutual feedback. Mean-teacher (Tarvainen and Valpola, 2017), LST (Li et al., 2019), and DualRE (Lin et al., 2019) are elaborated as the representatives to alleviate the weakness of these three methods, respectively.
Trading in Over-The-Counter (OTC) markets is facilitated by broker-dealers, in comparison to public exchanges, e.g., the New York Stock Exchange (NYSE). Dealers play an important role in stabilizing prices and providing liquidity in OTC markets. We a
pply machine learning methods to model and predict the trading behavior of OTC dealers for US corporate bonds. We create sequences of daily historical transaction reports for each dealer over a vocabulary of US corporate bonds. Using this history of dealer activity, we predict the future trading decisions of the dealer. We consider a range of neural network-based prediction models. We propose an extension, the Pointwise-Product ReZero (PPRZ) Transformer model, and demonstrate the improved performance of our model. We show that individual history provides the best predictive model for the most active dealers. For less active dealers, a collective model provides improved performance. Further, clustering dealers based on their similarity can improve performance. Finally, prediction accuracy varies based on the activity level of both the bond and the dealer.
Recent studies have demonstrated a perceivable improvement on the performance of neural machine translation by applying cross-lingual language model pretraining (Lample and Conneau, 2019), especially the Translation Language Modeling (TLM). To allevi
ate the need for expensive parallel corpora by TLM, in this work, we incorporate the translation information from dictionaries into the pretraining process and propose a novel Bilingual Dictionary-based Language Model (BDLM). We evaluate our BDLM in Chinese, English, and Romanian. For Chinese-English, we obtained a 55.0 BLEU on WMT-News19 (Tiedemann, 2012) and a 24.3 BLEU on WMT20 news-commentary, outperforming the Vanilla Transformer (Vaswani et al., 2017) by more than 8.4 BLEU and 2.3 BLEU, respectively. According to our results, the BDLM also has advantages on convergence speed and predicting rare words. The increase in BLEU for WMT16 Romanian-English also shows its effectiveness in low-resources language translation.
