search

d2l-ai / d2l-en

Interactive deep learning book with multi-framework code, math, and discussions. Adopted at 500 universities from 70 countries including Stanford, MIT, Harvard, and Cambridge.
https://D2L.ai
Other
22.53k stars 4.19k forks source link

17.1. Generative Adversarial Networks #1664

Open froggie901 opened 3 years ago

froggie901 commented 3 years ago

Second paragraph ~

But there is more to machine learning than just solving discriminative tasks. For example, given a large dataset, without any labels, we might want to learn a model that concisely captures the characteristics of this data. Given such a model, we could sample synthetic data examples that resemble the distribution of the training data. For example, given a large corpus of photographs of faces, we might want to be able to generate a new photorealistic image that looks like it might plausibly have come from the same dataset. This kind of learning is called generative modeling.

(1) What does learn a model mean here? Something different than the commonly used phrase train a model?

(2) This paragraph is little bit like word soup, consider revising for clarity. Below is one option.

There is more to machine learning than just solving discriminative tasks. If you have a large dataset without any labels you might want train a model that can concisely captures the characteristics of this data. You cold then could sample synthetic data examples that resemble the distribution of the training data... and then what?!. For example, given a large corpus of photographs of faces, we might want to be able to generate new photo realistic images which look like they could have came from original dataset. This kind of learning is called generative modeling.

ParamThakkar123 commented 8 months ago

(1) In the context of machine learning, the phrase "learn a model" is often used interchangeably with "train a model." Both phrases essentially mean the same thing: to develop a model's parameters based on data, so that it can make predictions or generate data. The choice of phrasing may vary based on the author's style, but they refer to the same process.