RST-style document-level discourse parsing remains a difficult task and efficient deep learning models on this task have rarely been presented. In this paper, we propose an attention-based hierarchical neural network model for discourse parsing. We also incorporate tensor-based transformation function to model complicated feature interactions. Experimental results show that our approach obtains comparable performance to the contemporary state-of-the-art systems with little manual feature engineering.

No abstract

Despite intense recent research, the neural correlates of conscious visual perception remain elusive. The most established paradigm for studying brain mechanisms underlying conscious perception is to keep the physical sensory inputs constant and identify brain activities that correlate with the changing content of conscious awareness. However, such a contrast based on conscious content alone would not only reveal brain activities directly contributing to conscious perception, but also include brain activities that precede or follow it. To address this issue, we devised a paradigm whereby we collected, trial-by-trial, measures of objective performance, subjective awareness, and the confidence level of subjective awareness. Using magnetoencephalography recordings in healthy human volunteers, we dissociated brain activities underlying these different cognitive phenomena. Our results provide strong evidence that widely distributed slow cortical potentials (SCPs) correlate with subjective awareness, even after the effects of objective performance and confidence were both removed. The SCP correlate of conscious perception manifests strongly in its waveform, phase, and power. In contrast, objective performance and confidence were both contributed by relatively transient brain activity. These results shed new light on the brain mechanisms of conscious, unconscious, and metacognitive processing.