Macrovector / Freepik
It would be expected that the encryption would slow down the computing. However, a new study has shown that applying encryption tools to ‘deceive’ an algorithm can, after all, make it faster.
A encryption It consists of shuffling the messages to seem random in the eyes of the malicious. But with the correct key, os hackers manage to unlock this randomness.
For this reason, it would be expected that the encryption would slow down the computing. But after all, it seems not.
A study recently in arXiv revealed that adding a pinch of encryption to key algorithms used in artificial intelligence (AI) models can make them more efficient, Thanks to a mathematical trick.
In the new study, researchers have found that similar use of encryption can also improve the efficiency of certain algorithms by exploring this “pseudo-aleatoity.”
As explained, the key to this trick is that, like the encrypted messages, the new matrices created can be unlocked With a password, which reveals a pattern for the seemingly random numbers.
This pattern serves as a shortcut: new matrices can be multiplied at once, instead of multiplying each number for each other number, as it would normally be needed.
“It’s the difference between a line and a square of the same size,” explained the study co-author Or Zamirfrom the University of Tel Aviv, in Israel, to.
These traps matrices can have multiple applications.
Automatic learning algorithms often use random matrices when they are being trained as part of a tool called classifier, which is used to distinguish between different data points, such as cats or dogs in images. This involves the analysis of very large data sets, but the new matrices can help make them smaller.
The pair also found that this trick can help in other areas, such as the data compressionwhich uses random matrices to preserve important characteristics of a data set, discarding irrelevant data, or even in the Search of similar itemssuch as those used by music recommendation services.
