I have a dataset in which we applied the deep learning to predict an outcome. Before that we applied blockchain methodology to store the training data. This was done to increase the security of our dataset but we had to sacrifice better timings.
Now, my query is how can I present this information that I've increased the security of the dataset. I mean should I just write "From previous work, we know blockchain increase the security of data. Thus, our data is also getting secured by blockchain methodology." ? Or is there any formal procedure to prove/present that our data is getting more secured due to the use of this technology?
I could not find a useful way so far. Please guide for further advancement.
Could you elaborate a bit on your understanding of "security" here? Are you addressing security objectives (as in the CIA-model), and, if so, which one? I might have completely misunderstood your question, because I also do not understand why it has been useful to store "training data" in a distributed ledger.
Lars Fischer The training data is getting updated periodically in our system. To make those transactions(updating training data) secure from third party attack we applied blockchain (since it's current and famous for it). Also, we want examine the time-security trade-off.
you can keep monitoring your dataset by tools such as autoruns ,process monitoring , wireshark and then run Red attack or backtrack5 test. Then capture results and compare with prior condition.
for example:
Test on edge services such as Authentication
Test on SSL and PKI if you are using any.
Test on DDos
Test on DNSSEC
Actually I still think you should avoid to write generically about "blockchain increases security of data". Distributed Ledger Technologies provide a specific set of security attributes, depending variably on the way they are used. The attributes and limits of protection should be discussed. Additional tests on known vulnerabilities, as mentioned above, could be a nice addon, but are no replacement for a throughout threat analysis. "Security" is not an attribute that can be measured in general.
personally I think it is more the discipline rather than the technology to make "security better". So you need a baseline and then understand how much progress from that baseline you made. So, if you used proper RBA (role base authentication) with a combination of ZT (zero trust), ECDHE Encryption, MFA and IAM, you are not necessarily "more secure" if you used blockchain instead. Ultimately is "how" you use the technology that matters, not the technology per se.
I suggest that you mention your use of blockchain (briefly) along with a description of any quantitative measures in a methods section. Blockchain might not need to be in the results, unless you have an index of some kind to report or some quantitative results with respect to your 'time-security trade-off'. If it is not in the results, then it might not need to be in the discussion. Your report may really be about the application of deep learning to answer an interesting question.
- Badges
- Science topic