Commonsense reasoning is a long-standing challenge for deep learning. For example, it is difficult to use neural networks to tackle the Winograd Schema dataset~\cite{levesque2011winograd}. In this paper, we present a simple method for commonsense reasoning with neural networks, using unsupervised learning. Key to our method is the use of language models, trained on a massive amount of unlabled data, to score multiple choice questions posed by commonsense reasoning tests. On both Pronoun Disambiguation and Winograd Schema challenges, our models outperform previous state-of-the-art methods by a large margin, without using expensive annotated knowledge bases or hand-engineered features. We train an array of large RNN language models that operate at word or character level on LM-1-Billion, CommonCrawl, SQuAD, Gutenberg Books, and a customized corpus for this task and show that diversity of training data plays an important role in test performance. Further analysis also shows that our system successfully discovers important features of the context that decide the correct answer, indicating a good grasp of commonsense knowledge.
Trieu joined Google AI Residency in 2017 after obtaining a Bachelor degree in Computer Science at University of Science, Vietnam. He had prior experience in developing statistical software package in Educational Data Mining and building Machine Vision for mobile devices. At Google, Trieu focuses on Natural Language Processing related problems, collaborates with teams working on Question Answering and published at ICML 2018. His most recent work achieves state-of-the-art results on the challenging Commonsense Reasoning test.