search

awesome-davian / awesome-reviews-kaist

Computer vision paper reviews written by KAIST AI students
Creative Commons Attribution Share Alike 4.0 International
43 stars 118 forks source link

[2022 spring] ICCV 2021 REFICS (20226054) #490

Open wonjeongchoi opened 2 years ago

wonjeongchoi commented 2 years ago

Thank you for your detailed and well-ordered review. The subject of your work is unfamiliar to me but it is quite good to understand and get the gist of paper.

I have some questions & suggestion about your review below.

  1. In motivation section, you mentioned that "Trojans in IC and PCB can make the devices vulnerable to adversarial attacks". What kinds of adversarial attacks can be done?? To my knowledge, the adversarial attack is to perturb the output of model by injecting some noises. So I wonder what kinds of attacks can be applied to hidden modifications(Trojans) in harware.
  2. I suggest you to provide more detailed explanation of method. What is the meaning of 'libraries' in method section that is used to extract about 10,000 standard cells? And there are many technical terms in that sections which are quite hard to understand. I recommend you to provide some additional explanations or revise with understandable terminology.

Thank you!

MaheZ20Kaist commented 2 years ago

It is a good review, I found it to be very detailed. However, there are a few items missing:

  1. References in related work section
  2. A separate section dedicated to references after conclusion.
  3. The technical terms in the methods section requires explanation.
koo616 commented 2 years ago

Thank you for your interesting and detailed review. I didn't know much about hardware assurance, but I read your review and thought it was a very interesting topic. Your review was written in an easy-to-understand manner, even though it was an unfamiliar field.

However, it would have been better if the methodology of the paper was explained in detail using images in the method section. In addition, it would be better if you could interpret the results of Table 4 in more detail.