Mat Leonard

My biggest hesitation with the Nanodegree before enrolling was the price tag. I was worried I wouldn’t get anything close to the value of the sticker price, particularly because I wasn’t primarily motivated by career advancement.

In the end, I think my worry was justified. The Nanodegree is overpriced for what it provides. The instructional videos are high quality, but many of the exercises are lifted directly from Kaggle or from research papers, and there still aren’t enough of them. The Nanodegree also offers a lot of personalized services that I consider expensive but low-value, like career coaching and 1on1 mentoring. As a self-motivated learner, these kinds of services only draw resources away from the product that I really care about: quality instructional content and challenging exercises.

Beyond the steep price, I also felt that the subscription model was detrimental to my learning. The Nanodegree was self-paced, meaning you didn’t have to stick to a set schedule, but you had to pay the same amount every month. I felt like I had to rush toward the end of the course to avoid an extra monthly payment, and this made me brush past topics I was interested in instead of pausing to learn more. As someone with a lot of curiosity but very little free time, this was a painful choice to have to make.

The pricing model for Udacity courses seems like it changes a lot, so my concerns over pricing may change in the future. Many people also get their bosses to pay for courses like these, so maybe price isn’t even a concern for you. But as someone working on the Nanodegree on my own time to satisfy my own curiosity, the price and the subscription model were strong negatives.

Conclusion: Who should take the Deep Learning Nanodegree?
The Deep Learning Nanodegree might be a good fit for you if you’re eager to get some practical experience building neural networks, you’re not too concerned about math, and you have a way to pay for an expensive course.

If, on the other hand, you’re interested primarily in theory, you want to go deep into the details, or you already have a handle on the basics of neural networks and deep learning, the Nanodegree probably won’t deliver the value you’re looking for. I’d instead try Nielsen’s book for a basic introduction, or Goodfellow, Bengio, and Courville’s book for advanced instruction. In both cases, I would strongly recommend convening a reading group around the book, ideally including mathematicians.

In the end, I think the course was a good use of my time. I got unstuck on some personal projects, I got ideas for new ones, and I was introduced to a lot of interesting ideas and good research papers. But I don’t expect these conditions to hold for every learner, and I want you to make the best choice you can with the information available to you.
- Review by Jean Cochrane on Deep Learning Nanodegree