We propose using the concept of decisive moment in order to deconstruct the obvious ideological effects found in discourse. The subject is constituted in enunciation, and its polysemic discourse clashes with the trans...We propose using the concept of decisive moment in order to deconstruct the obvious ideological effects found in discourse. The subject is constituted in enunciation, and its polysemic discourse clashes with the transparency of meaning. According to Pêcheux （1988）, the contradictions in discourse simultaneously establish regularity and instability of meanings, leading it to misunderstanding, to the event. Photography destabilizes that which is already formulated and brings out that which is new, the unexpected meaning, the decisive moment. We analyze this process in the picture by Sebastiao Salgado---＂The cradle of inequality lies in the inequality of the cradle＂. （CAPES-BEX 4394/10-0, FAPESP 09/54417-4, CNPq.）展开更多
大型活动散场期间的地铁车站客流属于可预知的非常规客流,采用常规客流的统计预测方法难以准确预测其客流需求。基于深度学习,将历史客流规律、大型活动数据与实时自动售检票系统数据相结合,提出了一种适用于大型活动散场期间地铁车站...大型活动散场期间的地铁车站客流属于可预知的非常规客流,采用常规客流的统计预测方法难以准确预测其客流需求。基于深度学习,将历史客流规律、大型活动数据与实时自动售检票系统数据相结合,提出了一种适用于大型活动散场期间地铁车站的短时客流预测模型。首先对历史客流数据进行了拆分及降噪处理,并分析了活动客流特征。之后,基于深度学习框架构建多层结构的卷积神经网络,拟合活动客流特征与客流时空分布的映射关系,并选取Adam(adaptive moment estimation)算法优化训练过程,以适用于活动散场时客流集中进站的情况。最后,以北京地铁奥林匹克公园站为例,利用实测数据验证了模型的准确性。预测结果表明:建立的Adam-CNN(convolution neural network)模型相对于常用时间序列方法自回归滑动平均和传统神经网络SGD-CNN模型具有更高的精度,能够为大型活动的组织提供更为有力的支持。展开更多
文摘We propose using the concept of decisive moment in order to deconstruct the obvious ideological effects found in discourse. The subject is constituted in enunciation, and its polysemic discourse clashes with the transparency of meaning. According to Pêcheux （1988）, the contradictions in discourse simultaneously establish regularity and instability of meanings, leading it to misunderstanding, to the event. Photography destabilizes that which is already formulated and brings out that which is new, the unexpected meaning, the decisive moment. We analyze this process in the picture by Sebastiao Salgado---＂The cradle of inequality lies in the inequality of the cradle＂. （CAPES-BEX 4394/10-0, FAPESP 09/54417-4, CNPq.）
文摘大型活动散场期间的地铁车站客流属于可预知的非常规客流,采用常规客流的统计预测方法难以准确预测其客流需求。基于深度学习,将历史客流规律、大型活动数据与实时自动售检票系统数据相结合,提出了一种适用于大型活动散场期间地铁车站的短时客流预测模型。首先对历史客流数据进行了拆分及降噪处理,并分析了活动客流特征。之后,基于深度学习框架构建多层结构的卷积神经网络,拟合活动客流特征与客流时空分布的映射关系,并选取Adam(adaptive moment estimation)算法优化训练过程,以适用于活动散场时客流集中进站的情况。最后,以北京地铁奥林匹克公园站为例,利用实测数据验证了模型的准确性。预测结果表明:建立的Adam-CNN(convolution neural network)模型相对于常用时间序列方法自回归滑动平均和传统神经网络SGD-CNN模型具有更高的精度,能够为大型活动的组织提供更为有力的支持。
