ID 原文 译文
56228 实验结果表明,在δ–准确率损失的衡量标准下,相比于传统的机器学习框架,联邦学习系统在DeepConfuse攻击下更加脆弱. The empirical results showed that the federatedlearning system is even more vulnerable under the DeepConfuse attack in terms of δ-accuracy loss.
56229 在实时地震监测中,地震P波(primary wave)的初动拾取任务具有至关重要的作用,其有助于地震应急响应的及时实施. Identifying the arrival times of seismic P-phases plays a significant role in real-time seismic monitoring,which provides critical guidance for emergency response activities.
56230 虽然此前在该领域已开展了大量的研究,但是如何从地震分布密集并且充满噪声的监测波形中有效地识别出P波仍然是一个具有挑战性的任务. While considerable research has been conductedon this topic, efficiently capturing the arrival times of seismic P-phases hidden within intensively distributed andnoisy seismic waves, such as those generated by the aftershocks of destructive earthquakes, remains a real challengesince most common existing methods in seismology rely on laborious expert supervision.
56231 例如对于大地震的余震监测,实践中使用的普遍方法仍依赖于专家辅助标注.本文针对地震实时监测任务,基于集成学习策略,提出一个全新的技术框架——EL-Picker,实现从连续地震波形中自主拾取P波的初动到时.具体而言,EL-Picker包含3个模块,即触发器、分类器和精化器.其中,分类器模块借鉴集成学习策略,实现对多个个体学习器的整合,提升整体模型性能.基于汶川Ms8.0地震的余震数据集进行的大量实验,我们发现EL-Picker不仅较好地实现P波初动拾取效果,并且多诊断出120%被人工遗漏的地震P波.同时,实验结果也启发我们探索如何针对不同的地震站台选取个性化的个体学习器构建分类器模块.此外,我们进一步地讨论了被人工遗漏的地震波形的规律特点,用于指导人工地震标注.这些发现清晰地验证了EL-Picker框架的鲁棒性、时效性、灵活性以及稳定性. To this end, in thispaper, we present a machine learning-enhanced framework based on ensemble learning strategy, EL-Picker, forthe automatic identification of seismic P-phase arrivals on continuous and massive waveforms. More specifically,EL-Picker consists of three modules, namely, Trigger, Classifier, and Refiner, and an ensemble learning strategyis exploited to integrate several machine learning classifiers. An evaluation of the aftershocks following the Ms 8. 0Wenchuan earthquake demonstrates that EL-Picker can not only achieve the best identification performance butalso identify 120% more seismic P-phase arrivals as complementary data. Meanwhile, experimental results alsoreveal both the applicability of different machine learning models for waveforms collected from different seismicstations and the regularities of seismic P-phase arrivals that might be neglected during the manual inspection. These findings clearly validate the effectiveness, efficiency, flexibility, and stability of EL-Picker.
56232 环状RNA (circluar RNA, circRNA)在基因表达、剪切和转录的过程中扮演着重要角色.越来越多的证据表明, circRNA与疾病的产生与发展存在着重要的联系.本文提出了一种基于多数据融合的非负矩阵分解算法(EDNMF)预测circRNA–疾病关联关系.该方法首先对circRNA–疾病关联关系进行预处理,解决了circRNA–疾病关联关系过少对算法产生的负面影响的问题.然后, EDNMF算法将circRNA表达谱和癌症相似性数据转化为约束条件,基于预处理后的circRNA–疾病关联关系采用改进的非负矩阵分解算法得到最终的打分值,从而预测circRNA–疾病关联关系.五折和十折交叉验证结果表明, EDNMF算法相比其他算法能更有效地预测circRNA–疾病关联关系. Circular RNA (circRNA) plays a significant role in gene expression, splicing, and transcription. Moreand more evidence indicates that circRNA is related to the pathogenesis and development of diseases. In thispaper, a non-negative matrix factorization algorithm based on circRNA expression profiles data and diseasesimilarity data (EDNMF) is proposed to predict circRNA-disease associations. The EDNMF algorithm firstlypreprocesses the circRNA-disease associations to solve the impact of too little the number of known circRNA?disease associations. Then, the EDNMF algorithm converts circRNA expression profile and cancer similaritydata into constraints. Finally, we can obtain the final scores for circRNA-disease associations by improved NMFalgorithm based on pre-processed circRNA-disease associations.
56233 此外,采用EDNMF算法预测新的circRNA–结肠直肠癌关联关系打分排名前10的结果中,大部分结果已经得到了佐证,表明了该算法可以有效地预测未知的circRNA–疾病关联关系. The performance results of 5-fold and 10-foldcross-validation indicate that the EDNMF algorithm achieves satisfactory performance comparing with otheralgorithms. Besides, the case study shows that EDNMF can mine new circRNA-disease associations very well,which can provide a reference for studying circRNA-disease associations.
56234 脊髓电刺激作为一种有效的意识促醒手段已经在临床上得到了较为广泛的应用,但是其内在机制仍不完全明确. Although spinal cord stimulation (SCS), as an effective method for consciousness modulation, hasbeen applied in MCS, the underlying mechanism is still not clear.
56235 本文将正常人静息态脑电作为对照组,利用样本熵对微意识状态患者的脊髓电刺激前后的脑电信号进行计算,并分析了基于互样本熵构造的脑网络在刺激前后的变化. In this study, the resting-state EEG signalsrecorded from healthy volunteers were employed as a control group for comparison. The sample entropy, as wellas the brain networks characteristics derived from the cross-sample entropy, were used to investigate the neuralresponses between the states of the pre-SCS and post-SCS.
56236 结果表明,脊髓电刺激提高了微意识状态患者额叶和中央区内的脑电信号复杂度; The results showed that the complexity of EEGsignals increased in the frontal and central regions after the SCS.
56237 还提高了患者在额叶内、中央区内以及额叶与其他脑区间的高频段(α～γ, 8～45 Hz)的耦合模式复杂度,表明通道间信息交互作用的显著增强; The SCS also enhanced the coupling mode inintra-region of pairwise channels within the frontal region and central region, as well as the inter-region channelsbetween the frontal region with other regions in high frequency bands (α 8∼13 Hz, β 13∼30 Hz, γ 30∼45 Hz).