在生成对抗网络中的部分和外观的双线模型
概括
本研究介绍了一种无监督的控制生成对抗网络 (GAN) 的方法,可以在没有手动输入的情况下精确地进行本地图像编辑. 这种方法有效地发现了跨各种GAN架构的像素级控制的空间和外观因素.
科学领域:
- 计算机视觉 计算机视觉
- 人工智能的人工智能
- 机器学习 机器学习
背景情况:
- 生成对立网络 (GAN) 通过利用它们的潜伏空间中的语义信息来进行先进的视觉编辑和合成.
- 当前的GAN编辑方法往往缺乏架构灵活性,与本地化控制作斗争,或需要监督数据,如细分面具.
- 在GAN中需要无监督的,架构无关的方法来进行细粒度控制,这是该领域的一个重大挑战.
研究的目的:
- 开发一种无监督的,建筑不可知的方法来发现 GAN 中的空间和外观因素.
- 为了实现上下文意识,像素级本地图像编辑,而不需要手动注释或特定的GAN架构.
- 与现有最先进的技术相比,证明拟议方法的效率和准确性.
主要方法:
- 提出了一种无监督的方法,从GAN特征图中共同发现空间部分和外观因素.
- 使用半非负张量因子化,应用于特征图表,以提取这些语义因子.
- 证明了该方法能够生成与发现的外观因素相对应的突出地图,而没有明确的标签.
主要成果:
- 开发的方法成功地以完全不受监督的方式解开空间和外观因素.
- 实现了具有上下文意识的本地图像编辑,具有精确的像素级控制,适用于各种GAN架构和数据集.
- 与以前的方法相比,在训练时间方面表现出更高的效率,并显著提高了局部控制的准确性.
结论:
- 提出的架构不可知论方法为GAN中的无监督,局部控制提供了高效和有效的解决方案.
- 发现的外观因素作为隐含的突出地图,在没有监督的情况下定位概念.
- 这项工作提升了GAN的功能,用于实现精细的直观控制,实现现实的图像编辑和合成.
相关概念视频
Nonconscious Mimicry
4.5K
Nonconscious mimicry occurs when individuals alter their mannerisms to match the behaviors and expressions of those nearby, without intention.
4.5K
The Representativeness Heuristic
15.8K
The representative heuristic describes a biased way of thinking, in which you unintentionally stereotype someone or something. For example, you may assume that your professors spend their free time reading books and engaging in intellectual conversation, because the idea of them spending their time playing volleyball or visiting an amusement park does not fit in with your stereotypes of professors.
15.8K
Compartment Models: Two-Compartment Model
5.4K
The two-compartment model divides the body into central and peripheral compartments to account for varying blood perfusion rates among organs and tissues, affecting drug distribution. The central compartment includes blood and highly perfused tissues with rapid drug distribution, while the peripheral compartment contains tissues with slower drug distribution. After a single IV bolus dose, the drug concentration is high in plasma and low in tissues. The drug distribution between compartments...
5.4K
Neural Circuits
1.1K
Neural circuits and neuronal pools are two of the main structures found in the nervous system. Neural circuits are networks of neurons that work together to carry out a specific task or process. They consist of interconnected neurons and glial cells, which provide structural and metabolic support.
Neuronal pools are collections of nerve cells with similar functions and interact through chemical and electrical signals. These pools include both interneurons (the central neural circuit nodes that...
Neuronal pools are collections of nerve cells with similar functions and interact through chemical and electrical signals. These pools include both interneurons (the central neural circuit nodes that...
1.1K
Generation of Straight or Branched Actin Filaments
2.9K
The straight or branched structure formation of actin filaments is controlled by nucleating proteins such as the formins and Arp2/3 complex. Formin-mediated assembly results in straight filaments, whereas Arp2/3 protein complex-mediated assembly results in branched actin filaments.
Arp2/3 Complex
Arp2/3 complex is a seven-subunit complex consisting of two proteins similar to actin- Arp2 and Arp3, and five other subunits that help keep Arp2 and Arp3 inactive. When required, the complex is...
Arp2/3 Complex
Arp2/3 complex is a seven-subunit complex consisting of two proteins similar to actin- Arp2 and Arp3, and five other subunits that help keep Arp2 and Arp3 inactive. When required, the complex is...
2.9K
Vector Algebra: Graphical Method
12.0K
Vectors can be multiplied by scalars, added to other vectors, or subtracted from other vectors. The vector sum of two (or more) vectors is called the resultant vector or, for short, the resultant.
We use the laws of geometry to construct resultant vectors, followed by trigonometry to find vector magnitudes and directions. For a geometric construction of the sum of two vectors in a plane, we follow the parallelogram rule. Suppose two vectors are at arbitrary positions. Translate either one of...
We use the laws of geometry to construct resultant vectors, followed by trigonometry to find vector magnitudes and directions. For a geometric construction of the sum of two vectors in a plane, we follow the parallelogram rule. Suppose two vectors are at arbitrary positions. Translate either one of...
12.0K


