Lesson 1 survey results

Here are the quantitative results from lesson 1’s survey:

Representative answers to: Any areas where Jeremy did particularly well? Any areas of improvement for him? Any areas where the course materials and resources are working particularly well for you? Any suggestions for improvement for the materials and resources?

  • I think it’d be helpful to have the background about how Jeremy is trying to teach on the first night - I was a little surprised by the chaos of the first week’s homework. It took a while just to figure out what was required, which was a little frustrating when I knew I had limited time. But it was very effective pedagogy, so I can’t complain much :slight_smile:
  • Explanations are very clear! It’s great that we have people connecting through skype, but it can be a bit distracting when Jeremy needs to interrupt the middle of an explanation to fix a skype issue (i.e. when someone needs to get called back in). If we can have another person managing that, it would be great! Material and resources are fantastic so far. After just the first week and assignment, I’ve learned so much. Keep up the awesome job! (NB: The skype participants now have their own class 24 hours before, so this no longer happens)
  • I would love a single place where I can find the assignment and the resources for that class. E.g. I there is a page for Lesson 1 on the wiki but it was missing the assignment. The assignment is on the forum page titled “Lesson 1 Discussion”. (Thanks for the suggestion. I had hoped that class participants would create that themselves, but that didn’t happen, so I’ve pre-populated the page this time. Please do help us to improve the page however!)
  • Very clear. Very kind. He makes you feel that you can use what we are going to learn to do good things.
  • Jeremy has put in a lot of effort in creating the AMIs which saved us a lot of time in setting up the environment. A good amount of time was spent on talking about jupyter, tmux, and other topics which were not related to Deep Learning. Since its only a 7 day course, it would be better if the focus was more on Deep Learning. ( My view is that you can’t apply deep learning effectively if you don’t know how to use these tools. The goal of this course if to learn to apply deep learning. :slight_smile: )
  • The course is really great. I am really happy a high bar is being set and we are moving fast.

Representative answers to: Anything that you could do differently over the following lessons to get more out of the class and assignments?

  • I need to do more homework before the weekend. I did everything on Saturday and Sunday and simply ran out of time. I got pretty far though so it was ok. I should also KISS for the homework - Jeremy’s code is simpler than mine. Better to spend more time on the analysis than the infrastructure. Btw for someone with a job and normal family life it’s tough to devote more than 8 hours to homework. 15 hours sounds amazing but totally unrealistic.
  • I need to spend more time on it. I also should group up with people near me so we can collaborate on mutual interests.
  • I could better document my progress, what I have tried, what didn’t work, etc.
  • I did too little to address the assignment this week. Next week i will try to commit more time and do more than the minimum required.
  • Will spread the time that I will study over the week instead of cramming it in the end.
  • Spend more time with the material before class.
  • I could be taking more initiative to meet with group members!
  • I might be taking a break from my full time job to have more time for this course.
  • I will probably read the Keras docs more carefully. It is difficult as a beginner to know if your results are due to expected behavior of model architectures versus unintuitive behavior of library functions resulting in incorrect data being written out.
  • I should have set up AWS/GPU sooner so that I had more time to work on the HW.
  • participate more
  • Spend ever more time in understanding the course materials and helping others - this can solidify our own understanding of the subject.
  • Coming from an R background, I need to be better trained in python/numpy to be quicker on the steps that involve data manipulation/processing.
  • I should commit more time.
  • Ask more questions on the forum.
  • Try to interact more on Slack and learn more from them.