-
Comparatively 相比之下,相对地
-
they are a strong candidate for unsupervised learning. 他们是无监督学习的有力候选人。
-
Additionally 此外
-
Learning reusable feature representations from large unlabeled datasets has been an area of active research.
从大型未标记数据集中学习可重用特征表示一直是一个活跃的研究领域。(开门见山,introduction第一句) -
We propose that one way to build good image representations is by training Generative Adversarial Networks (GANs)
我们提出,构建良好图像表示的一种方法是训练生成性对抗性网络(GANs) -
We show that the generators have interesting vector arithmetic properties allowing for easy manipulation of many semantic qualities of generated samples.
我们证明了生成器具有有趣的向量算术特性,允许对生成样本的许多语义性质进行简单的操作。 -
Core to our approach is adopting and modifying three recently demonstrated changes to CNN archi-tectures.
我们的方法的核心是采用和修改最近对CNN架构所做的三个修改。 -
Previous work has demonstrated that supervised training of CNNs on large image datasets results in very powerful learned features
以往的研究表明,在大图像数据集上对cnn进行有监督的训练可以产生非常强大的学习特征 -
ground truth 真实信息
-
Inspired by GAN, in this paper we propose GraphGAN,a novel framework that unifies generative and discriminative thinking for graph representation learning. Specifically,
在× × 的启发下,本文提出了一种新的×× × × 学习框架× × × ,它将× × 。具体说******** -
We attribute the superiority of GraphGAN to its unified adversarial learning framework as well as the design of the proximity-aware graph softmax that
naturally captures structural information from graphs.
我们将× × 的优越性归因于其× × × ,以及× × × × × 。
English notes forEnglish notes for GAN
最新推荐文章于 2022-07-03 23:58:29 发布