Scheduled sampling for one-shot learning via matching network
Scheduled sampling for one-shot learning via matching network
复制标题
通过匹配网络进行一次性学习的计划采样
DOI：
10.1016/j.patcog.2019.07.007
复制
发表时间：
2019-12
影响因子：
8
通讯作者：
Minnan Luo
中科院分区：
文献类型：
--
作者：
Lingling Zhang;Jun Liu;Minnan Luo
考虑到人类可以从一个样本中成功地学习新的对象，一次学习，其中每个视觉类只有一个标记的样本进行训练，引起了越来越多的关注。在过去的几年中，大多数研究人员通过训练匹配网络来实现一次学习，将一个小的标记支持集和一个未标记的图像映射到其标签。支持集由一幅与未标记图像具有相同标记的图像和随机抽样产生的少量具有其他标记的图像组合而成。这种随机抽样策略很容易产生大量的过易支持集，其中大多数标签与未标记图像的标签不太相关。这导致了匹配网络在不可区分标签集上进行一次性预测的局限性。针对这个问题，我们提出了一种新的度量来评估支持集的学习难度，该度量联合考虑了视觉标签的语义多样性和相似性。在此基础上，我们引入了一种定时采样策略，由易到难地训练匹配网络。在mini-Imagenet、Birds和Flowers等三个数据集上的实验结果表明，该方法比以往的方法有了显著的改进。
Considering human can learn new object successfully from just one sample, one-shot learning, where each visual class just has one labeled sample for training, has attracted more and more attention. In the past years, most researchers achieve one-shot learning by training a matching network to map a small labeled support set and an unlabeled image to its label. The support set is combined by one image with the same label as unlabeled image and few images with other labels generated by random sampling. This random sampling strategy easily generates massive over-easy support sets in which most labels are less relevant to the label of unlabeled image. It leads to the limitation of matching network for one-shot prediction over indistinguishable label sets. For this issue, we propose a novel metric to evaluate the learning difficulty of support set, where this metric jointly considers the semantic diversity and similarity of visual labels. Based on the metric, we introduce a scheduled sampling strategy to train the matching network from easy to difficult. Extensive experimental results on three datasets, including mini-Imagenet, Birds and Flowers, indicate that our method could achieve significant improvements over other previous methods.
登录
查看更多内容
影响因子：
2.5
作者：
B. Lake;Chia-ying Lee;James R. Glass;J. Tenenbaum
通讯作者：
B. Lake;Chia-ying Lee;James R. Glass;J. Tenenbaum
DOI：
--
发表时间：
2011-07
期刊：
--
影响因子：
--
作者：
C. Wah;Steve Branson;P. Welinder;P. Perona;Serge J. Belongie
通讯作者：
C. Wah;Steve Branson;P. Welinder;P. Perona;Serge J. Belongie
DOI：
10.1016/j.patcog.2018.02.028
发表时间：
2017-09
期刊：
Pattern Recognit.
影响因子：
--
作者：
N. Sarafianos;Theodore Giannakopoulos;Christophoros Nikou;I. Kakadiaris
通讯作者：
N. Sarafianos;Theodore Giannakopoulos;Christophoros Nikou;I. Kakadiaris
DOI：
10.1145/2783258.2783264
发表时间：
2015-08
期刊：
Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining
影响因子：
--
作者：
P. Xie;Yuntian Deng;E. Xing
通讯作者：
P. Xie;Yuntian Deng;E. Xing
影响因子：
7.5
作者：
J. Schmidhuber;Jieyu Zhao;M. Wiering
通讯作者：
J. Schmidhuber;Jieyu Zhao;M. Wiering