Simultaneous translation, which translates sentences before they are finished, is useful in many scenarios but is notoriously difficult due to word-order differences. While the conventional seq-to-seq framework is only suitable for full-sentence translation, we propose a novel prefix-to-prefix framework for simultaneous translation that implicitly learns to anticipate in a single translation model. Within this framework, we present a very simple yet surprisingly effective "wait-k" policy trained to generate the target sentence concurrently with the source sentence, but always k words behind. Experiments show our strategy achieves low latency and reasonable quality (compared to full-sentence translation) on 4 directions: zh↔en and de↔en. * M.M. and L.H. contributed equally; L.H. conceived the main ideas (prefix-to-prefix and wait-k) and directed the project, while M.M. led the implementations on RNN and Transformer. See example videos, media reports, code, and data at https://simultrans-demo.github.io/. President Bush met with Putin in MoscowBùshí Bush zǒngtǒng President zài at Mòsīkē Moscow yǔ with Pǔjīng Putin huìwù meet prediction read write Source side → Target side → 2 Preliminaries: Full-Sentence NMT We first briefly review standard (full-sentence) neural translation to set up the notations.Regardless of the particular design of different seq-to-seq models, the encoder always takes

Balancing accuracy and latency is a great challenge for simultaneous translation. To achieve high accuracy, the model usually needs to wait for more streaming text before translation, which results in increased latency. However, keeping low latency would probably hurt accuracy. Therefore, it is essential to segment the ASR output into appropriate units for translation. Inspired by human interpreters, we propose a novel adaptive segmentation policy for simultaneous translation. The policy learns to segment the source text by considering possible translations produced by the translation model, maintaining consistency between the segmentation and translation. Experimental results on Chinese-English and German-English translation show that our method achieves a better accuracy-latency trade-off over recently proposed state-of-the-art methods. * Corresponding author. before adequate source content is delivered, the translation quality degrades. However, waiting for too much source text increases latency.The policies of recent work generally falls into two classes:

Dysregulated long noncoding RNAs (lncRNAs) remains to be explored in tumorigenesis. LncRNA HOXC13 antisense RNA (HOXC13-AS) has been found as an oncogene in many cancers; however, the role of HOXC13-AS in breast cancer still elusive. In this study, the HOXC13-AS levels and its role in cell proliferation was first measured by real-time quantitative polymerase chain reaction, Cell Counting Kit-8 assay, and colony formation assay. It showed that HOXC13-AS was increased in breast cancer tissues compared with the adjacent normal tissues and upregulated HOXC13-AS promoted the growth of breast cancer cells. Then, we found that the miR-497-5p levels were downregulated in cancer tissues compared with the adjacent tissues and miR-497-5p suppressed breast cancer cell proliferation. Further study showed that HOXC13-AS could function as a "sponge" for miR-497-5p then suppress miR-497-5p expression. Moreover, we next identified that Phosphatase and Tensin homolog (PTEN) is the target of miR-497-5p. Overexpression of miR-497-5p by chemical mimics decreased the expression of PTEN, while downregulation of miR-497-5p by HOXC13-AS rescued the expression of PTEN. Finally, we showed that HOXC13-AS promoted the proliferation of breast cancer cells and tumor growth through miR-497-5p/PTEN axis in vitro and in vivo. Hence, we conclude that HOXC13-AS, which is significantly upregulated in breast cancers, promoted cell proliferation through the suppressed miR-497-5p and further upregulated PTEN. K E Y W O R D Sbreast cancer, ceRNA, HOXC13-AS, miR-497-5p, PTEN

BackgroundOptogenetic silencing techniques have expanded the causal understanding of the functions of diverse neuronal cell types in both the healthy and diseased brain. A widely used inhibitory optogenetic actuator is eNpHR3.0, an improved version of the light-driven chloride pump halorhodopsin derived from Natronomonas pharaonis. A major drawback of eNpHR3.0 is related to its pronounced inactivation on a time-scale of seconds, which renders it unsuited for applications that require long-lasting silencing.ResultsUsing transgenic mice and Xenopus laevis oocytes expressing an eNpHR3.0-EYFP fusion protein, we here report optimized photo-stimulation techniques that profoundly increase the stability of eNpHR3.0-mediated currents during long-term photo-stimulation. We demonstrate that optimized photo-stimulation enables prolonged hyperpolarization and suppression of action potential discharge on a time-scale of minutes.ConclusionsCollectively, our findings extend the utility of eNpHR3.0 to the long-lasting inhibition of excitable cells, thus facilitating the optogenetic dissection of neural circuits.

No abstract

No abstract

Chinese spelling correction (CSC) is an important yet challenging task. Existing state-ofthe-art methods either only use a pre-trained language model or incorporate phonological information as external knowledge. In this paper, we propose a novel end-to-end CSC model that integrates phonetic features into language model by leveraging the powerful pre-training and fine-tuning method. Instead of conventionally masking words with a special token in training language model, we replace words with phonetic features and their sound-alike words. We further propose an adaptive weighted objective to jointly train error detection and correction in a unified framework. Experimental results show that our model achieves significant improvements on SIGHAN datasets and outperforms the previous state-of-the-art methods.

NKCC1 is the primary transporter mediating chloride uptake in immature principal neurons, but its role in the development of in vivo network dynamics and cognitive abilities remains unknown. Here, we address the function of NKCC1 in developing mice using electrophysiological, optical, and behavioral approaches. We report that NKCC1 deletion from telencephalic glutamatergic neurons decreases in vitro excitatory actions of γ-aminobutyric acid (GABA) and impairs neuronal synchrony in neonatal hippocampal brain slices. In vivo, it has a minor impact on correlated spontaneous activity in the hippocampus and does not affect network activity in the intact visual cortex. Moreover, long-term effects of the developmental NKCC1 deletion on synaptic maturation, network dynamics, and behavioral performance are subtle. Our data reveal a neural network function of NKCC1 in hippocampal glutamatergic neurons in vivo, but challenge the hypothesis that NKCC1 is essential for major aspects of hippocampal development.