Adversarial Attacks in Autonomous Vehicles
Shivam Mukherjee (~ShivamMukherjee) |
We drew out a survey detailing the kinds of Autonomous Attacks that can be used to target Autonomous Vehicles (AVs) and Connected Autonomous Vehicles (CAVs) by surveying recent literature which highlight modes of deployment of adversarial examples, including Physical World based approaches, and those that target the learning and inference models already present in the vehicle. We also present a number of defensive approaches that can be used to mitigate the adversarial techniques shown.
Our survey touches upon the concepts in the simplest possible manner, eschewing the (actually) important mathematics and presenting information in a rudimentary language, easily comprehensible to the beginner and the programmer who has recently taken interest in Machine Learning. Code has been cited sparsely to append to our survey's comprehensibility to the programmer community.
- Familiarity with Python
- Vocabulary used in Deep Learning (this is a good read)
- Not much else, really!
Suprateem Banerjee (shamelessly plagiarised from his LinkedIn profile): I am an engineering student pursuing Btech in Computer Science and Engineering, from SRM University, Chennai, India. I intend to pursue a Master's degree, possibly from United States, after the completion of my course, in 2020. I am currently working autonomously on projects which I am positive will help develop a sustainable future. I also intend to create a major difference in the industrial sector with my ideas and solutions.
Shivam Mukherjee (also equally culpable): I am an absolute neophyte, passionate about developing games. The occasional internship continues to gift me with experience in different fields. I'm planning to work for 2 years after 2020 (when my course ends - Suprateem and I are final year students currently), beyond which I'm keeping all options open. While Data Science and Machine Learning aren't my core skills, it's become all too important that I at least understand this growing fascination of the industry while it's in the mainstream.