Course talk:CPSC522/Convolutional Neural Networks

Hi Yan Zhao,

Good job. I appreciate your work on this topic. However, I have several suggestions for you. You described much knowledge about Convolutional Neural Networks. However, I do not see the content about the incremental contribution of one paper over the other and your own thoughts on how successful it was and how it can be improved. Thanks.

Sincerely,

Ke Dai

Thanks for your feedback.

The incremental contribution is: 1. The 1st paper introduce the idea of CNN. 2. Although the idea of CNN is good, it is hard to implement at that time. So the 2nd paper proposed several suggestions on implementation, and the result is good.

I guess the organization of this page is not clear so I will try to improve. Thanks again for your suggestions.

Hi Yan,
Your page was easy to understand and generally well written.
I do have a few suggestions:

• You could add a concluding section at the end of the page talking about the exact incremental contribution of one paper over the other.
• You could also add your own thoughts on how successful it was and how it can be improved.

Hope this helps,
Adnan

Thanks for your feedback.

Hi Yan, A well written page in my opinion. The discussion was not too much convoluted and the content is easily understood. The only thing suggestion I have is regarding the background knowledge the page needs. Link to an external page which gives an idea about the basics of ANNs (such as what are neurons in an ANN, why are they called etc.) would greatly help the naive reader who might not have the necessary background. I hope this helps :)

Thanks for you feedback.

Good work Yan. The page is not too long and has a good flow. The major part of your page contains basics about CNN. I would suggest to dig a little bit more about the simple loop and modular debugging.

Thanks for you feedback.