As a contrastive body, it is  second in the Solar System only to Saturn's moon Iapetus.

作为一个对比天体，它在太阳系中排名第二，仅次于卫卫八。

So like I said about contrastive learning, you have this notion of a positive and a negative.

就像我之前提到对比学习，它涉及到正样本和负样本概念。

So a non contrastive method, for example, like clustering is one non contrastive method.

因此，非对比方法之一，例如聚类，便是一种非对比方法。

Compare the unstressed French first person pronoun je (in je sais and j'aime) with the contrastive form moi.

比较无重音法语第一人称代词je(in je sais and j'aime)和对比形式moi。

And again, you use contrastive learning for this.

再次，你为此采用对比学习方法。

(3) Contrastive distribution: If two sounds are regarded as different phonemes, they must be in contrastive distribution.

(3)对比分布：如果两个声音被视为不音素，则它们一定是对比分布。

So in contrastive learning, basically the video and the corresponding audio are positives.

因此，在对比学习中，基本原则是：视频及其对应音频被视为正样本。

It's just a simple application of contrastive learning.

这只是对比学习一个简单应用。

Is similarity learning or contrastive learning or these non-contrastive methods the same as understanding something?

相似性学习或对比学习或这些非对比方法就是理解某件事吗？

Irregularity was typical of Baroque art, which was much richer in highly contrastive forms than the plainer and more harmonious Renaissance art.

不规则是巴洛克艺术典型，文艺复兴时期艺术相比，巴洛克艺术在高度对比形式上更为丰富。

So this is the contrastive part.

所以这就是对比部分。

And second of all, what's the role of data augmentation in self-supervised learning and contrastive learning?

其次，数据增强在自监督学习和对比学习中扮演什么角色？

You can do this by contrastive learning.

你可以通过对比学习来实现这一点。

This might be a good place to, you already mentioned it, but what is contrastive learning and what are energy-based models?

这或许是个好机会， 你已经提到过，但究竟什么是对比学习， 什么是基于能量模型呢？

So that's been one of the reasons why non contrastive methods have become like popular and why people think that they're going to be more useful.

这就是为何非对比方法变得流行，并被认为将更加有用原因之一。

And the idea basically is that you can explain a lot of the contrastive models, GANs, for example, which are like generative adversarial networks.

这个基本思想是， 你可以用能量函数来很好地解释许多对比模型，比如生成对抗网络（GANs）， 它们就像生成对抗网络一样。

And so what has happened is basically that typically these methods that are contrastive really require access to lots of negatives, which becomes harder and harder to sort of scale when designing a learning algorithm.

因此，通常情况下， 这些对比方法确实需要访问大量负样本，而在设计学习算法时， 这变得越来越难以扩展。

One idea that I proposed in the early 90s with my colleagues at Bell Labs, Jane Bromley and a couple other people, which we now call contrastive learning, which is to have negative examples, right?

我于20世纪90年代初贝尔实验室事们，包括Jane Bromley等人， 共提出了一种理念，如今我们称之为对比学习。其核心在于引入负样本，对吧？ 也就是说，我们拥有成对已知不图像。

The way most of the current methods work, contrastive or otherwise, is by taking an image, in the case of images, is by taking an image and then computing basically two perturbations of it.

因此，它在计算机视觉， 尤其是在自监督学习中， 扮演了基础性角色。 当前大多数方法，无论是对比学习还是其他方式， 其工作原理通常是选取一张图片，在处理图像情况下， 就是选取一张图片， 然后基本上计算出它两种扰动版本。 这些扰动可以是图像两个不裁剪部分， 伴随着不光照条件、对比度或色彩变化。