Chevron Left
返回到 Dimensionality Reduction using an Autoencoder in Python

学生对 Coursera Project Network 提供的 Dimensionality Reduction using an Autoencoder in Python 的评价和反馈

16 个评分
5 条评论


In this 1-hour long project, you will learn how to generate your own high-dimensional dummy dataset. You will then learn how to preprocess it effectively before training a baseline PCA model. You will learn the theory behind the autoencoder, and how to train one in scikit-learn. You will also learn how to extract the encoder portion of it to reduce dimensionality of your input data. In the course of this project, you will also be exposed to some basic clustering strength metrics. Note: This course works best for learners who are based in the North America region. We’re currently working on providing the same experience in other regions....

1 - Dimensionality Reduction using an Autoencoder in Python 的 5 个评论(共 5 个)

创建者 Ulvi I

May 04, 2020

Very practical and useful introductory course. Looking for the next courses :)

创建者 Mayank S

May 05, 2020

Nice Course, Well Explained, Thanks :)

创建者 chandrasekhar u

May 06, 2020

Quite a new experience

创建者 Joerg A

May 19, 2020

The time is too short, especially if you want to not just type in the desktop, but also take notes. My huge problem was, that whenever I wanted to type in a different window, the video would stop. In the end I was kicked out about 5minutes when I would normally have finished.

The part about autoencoder, like which attributes (when doing print(autoencoder) are important could have been deeper.

I also learned some nice python and data science tricks. Hence the 4 stars and not 3. Also I guess the player constraints should not be accounted to the teacher.

创建者 Sujeet B

May 07, 2020

1. The cloud time given was not enough. I thought we get ample time to do experiments and verify each step (and not just copy things over to the desktop). Disappointed to find, I was not allowed more "cloud time". Not sure, if that had something to do with the course fee paid (time limit based on fee amount?).

2. In the final set of questions: #6 was unrelated to the content covered in the project (Task 5). "Treating outliers as singletons was necessary to get a valid value for our Silhouette Scores". I am not sure when did we talk about 'outliers' and 'singletons' in Task 5. In the answers, Task #6 was referred. Was there a Task 6 dealt with, in the video? I couldn't find it in this project.

3. Did you miss "Does not" in Question #4 statement: "Which of these snippets express a ReLU function"?

All in all, it was a nice course. I do have an understanding of auto-encoders, compared with PCA now. I look forward to more such courses....