An AI Message Decoder Based on Bacterial Growth Patterns:
From a box of Cracker Jack to The Da Vinci Code, everybody enjoys deciphering secret messages. But biomedical engineers at Duke University have taken the decoder ring to place it's never been before — the patterns created by bacterial colonies.
Depending on the initial conditions used, such as nutrient levels and space constraints, bacteria tend to grow in specific ways. The researchers created a virtual bacterial colony and then controlled growth conditions and the numbers and sizes of simulated bacterial dots to create an entire alphabet based on how the colonies would look after they fill a virtual Petri dish. They call this encoding scheme emorfi.
The encoding is not one-to-one, as the final simulated pattern corresponding to each letter is not exactly the same every time. However, the researchers discovered that a machine learning program could learn to distinguish between them to recognize the letter intended.
"A friend may see many images of me over the course of time, but none of them will be exactly the same," explained Lingchong You, professor of biomedical engineering at Duke. "But if the images are all consistently reinforcing what I generally look like, the friend will be able to recognize me even if they're shown a picture of me they've never seen before."
[...] Give the cypher a try yourself. You can type in anything from your name to the Gettysburg Address, or even the Christmas classic, "Be sure to drink your Ovaltine."
Journal Reference:
Jia Lu, Ryan Tsoi, Nan Luo, et al. New encryption method uses simulated bacterial growth based on specific initial conditions to form patterns corresponding to letters [open], Patterns, 2022. DOI: 10.1016/j.patter.2022.100590
(Score: 0) by Anonymous Coward on Sunday October 02 2022, @03:50PM (3 children)
And where's the missing part about how everybody else can't do this and read your message...?
(Score: 2) by hopdevil on Sunday October 02 2022, @04:37PM (2 children)
In theory, you would need to know the correct initial conditions
(Score: 1, Insightful) by Anonymous Coward on Sunday October 02 2022, @04:59PM (1 child)
Oh, a 1-time pad. How interesting.
(Score: 3, Insightful) by driverless on Monday October 03 2022, @02:55AM
Oh, a press release hyping up some mathematical wankery from a we-publish-anything journal. How interesting.
(Score: 2) by looorg on Sunday October 02 2022, @04:50PM (1 child)
Does the encryption autodestruct if I drop some penicillin in the dish?
(Score: 2) by maxwell demon on Sunday October 02 2022, @05:00PM
No, that would be manual destruction by you. it would be autodestruction if there was e.g. some Penicillium chrysogenum on the dish, and that would grow and release the penicillin over time.
The Tao of math: The numbers you can count are not the real numbers.
(Score: 2, Insightful) by Anonymous Coward on Sunday October 02 2022, @05:28PM
The game of LIFE, 2022 style.
(Score: 2) by legont on Monday October 03 2022, @12:53AM (3 children)
In the end, their simulation is based on one or more random number generators. It can be proved that any function of such generators - such as their model - produce a weaker encryption than the generators themselves.
"Wealth is the relentless enemy of understanding" - John Kenneth Galbraith.
(Score: 3, Insightful) by hopdevil on Monday October 03 2022, @02:19AM
I'd add that putting "ML" into the mix hints at some potential issues. Proper encryption is based in mathematically hard problems. Here "ML" is used as a secret algorithm that surely no one can figure out:
AI is really just a compression algorithm, not a hard mathematical problem. Nice try Duke, but leave encryption to the experts
(Score: 2) by maxwell demon on Monday October 03 2022, @03:54PM (1 child)
I think you mean "weaker or equal". I'm pretty sure the identity function doesn't weaken anything. Nor does the function that simply flips every bit from a random bit generator. Or the function that flips every other bit.
The Tao of math: The numbers you can count are not the real numbers.
(Score: 2) by legont on Tuesday October 04 2022, @02:02AM
Yes, off course you are right, in theory. In practice, folks tend to introduce functions that weaken it significantly. Some of those functions look really innocent. So, if a function is needed for whatever reason, it has to be proved that it does not weaken the algorithm. It's harder than most people, including mathematicians, think it is.
"Wealth is the relentless enemy of understanding" - John Kenneth Galbraith.
(Score: 0) by Anonymous Coward on Monday October 03 2022, @09:10AM
So just how different is this from any of the myriad, well-tested and analysed ways that have been invented for doing cryptography? Have they gotten an actual cryptography expert to look at the technique to evaluate its security? Has anyone who knows the first thing about cryptanalysis tried their hand at cracking it?