针对传统近重复文本图像检索方法需人工事先确定近重复文本图像之间存在的变换类型,易受到人主观性影响这一问题,提出一个面向近重复文本图像检索的三分支孪生网络,能自动学习图像之间存在的各种变换。该网络输入为三元组,包括查询图像...针对传统近重复文本图像检索方法需人工事先确定近重复文本图像之间存在的变换类型,易受到人主观性影响这一问题,提出一个面向近重复文本图像检索的三分支孪生网络,能自动学习图像之间存在的各种变换。该网络输入为三元组,包括查询图像、查询图像的近重复图像以及其非近重复图像,训练时采用三元损失使得查询图像和近重复图像之间的距离小于查询图像与非近重复图像之间的距离。提出的方法在两个数据集上的mAP(mean average precision)分别达到98.76%和96.50%,优于目前已有方法。展开更多
This article reports a study of the effects of inference training and text repetition on Chinese university students' performance of two listening information-transfer tasks that provided built-in measures of their c...This article reports a study of the effects of inference training and text repetition on Chinese university students' performance of two listening information-transfer tasks that provided built-in measures of their comprehension and opportunities for the acquisition of ten unknown target words embedded in the listening texts. One group just listened to the text once, a second group three times, while the third listened three times and received inference-training support. The results showed that text repetition had a positive effect on both comprehension and vocabulary acquisition. However, the inference-training had no additional effect. The study lends support to the findings of previous studies which have shown that repeated opportunities to process oral input have a positive effect on listening comprehension and extends these studies by showing that it also facilitates incidental vocabulary acquisition.展开更多
文摘针对传统近重复文本图像检索方法需人工事先确定近重复文本图像之间存在的变换类型,易受到人主观性影响这一问题,提出一个面向近重复文本图像检索的三分支孪生网络,能自动学习图像之间存在的各种变换。该网络输入为三元组,包括查询图像、查询图像的近重复图像以及其非近重复图像,训练时采用三元损失使得查询图像和近重复图像之间的距离小于查询图像与非近重复图像之间的距离。提出的方法在两个数据集上的mAP(mean average precision)分别达到98.76%和96.50%,优于目前已有方法。
文摘This article reports a study of the effects of inference training and text repetition on Chinese university students' performance of two listening information-transfer tasks that provided built-in measures of their comprehension and opportunities for the acquisition of ten unknown target words embedded in the listening texts. One group just listened to the text once, a second group three times, while the third listened three times and received inference-training support. The results showed that text repetition had a positive effect on both comprehension and vocabulary acquisition. However, the inference-training had no additional effect. The study lends support to the findings of previous studies which have shown that repeated opportunities to process oral input have a positive effect on listening comprehension and extends these studies by showing that it also facilitates incidental vocabulary acquisition.