21.10.2015, 18:30
Ostatnie wieści...wciąż trwają prace nad poprawkami w wersji 3 beta...ale to nie wszystko. Dziś Dan ogłosił, że w VS zapowiada się spora zmiana związana z zaimplementowaniem całkiem nowej technologii...jest nowa nie tylko dla samego programu, ale nowa jest jako samodzielna dziedzina nauki związana ze sztuczną inteligencją - to tzw. "systemy uczące się" lub inaczej "uczenie maszynowe".
Nieco więcej o tej dziedzinie np tu
A wracając do programu...fragment opisu zamieszczonego na Wildersach
Nieco więcej o tej dziedzinie np tu
[Aby zobaczyć linki, zarejestruj się tutaj]
A wracając do programu...fragment opisu zamieszczonego na Wildersach
Cytat:VS is going to have a new feature soon, and there will probably be a few questions, so I figured that I would tell you as much about it now so that you will know what it is all about. The new feature is called “Voodoo Ai” for now, and if anyone has a better, less pretentious name, please let us know and we will change it[Aby zobaczyć linki, zarejestruj się tutaj]
. Voodoo Ai utilizes 3-4 machine learning / artificial intelligence algorithms along with other “features” of the portable executable (PE) binary (exe’s and dll’s) to assist users in making the correct decision (block or allow) when VS prompts them. Machine learning / artificial intelligence is kind of the latest “craze” in security software, and it differs from behavior blocking / heuristics in that the analysis all occurs pre-execution. VS is not the first to offer this (there is at least one other company specializing in this), and in the last 4-5 years there have been a lot of academic studies that have accessed the viability of machine learning and artificial intelligence in detecting malware, and as it turns out, while it will never be 100% perfect, it is quite accurate.
(...)
This new feature is not quite as complex or impressive as it might sound, but so far it does seem to be quite accurate. For example, when I analyzed all of the executables in several of the Windows and Program Files directories that were known to be safe, it was 100% accurate, with the exception of 6 files from an ATI display driver. I looked at these 6 files a lot closer, and let’s just put it this way… if it was not obvious that they were from the ATI display driver, I would think they were malicious as well… one example is that they were not digitally signed. Not that I am picking on ATI, but really, these should be signed, along with the other “features” that should have been handled properly.
[Aby zobaczyć linki, zarejestruj się tutaj]
"Bezpieczeństwo jest podróżą, a nie celem samym w sobie - to nie jest problem, który można rozwiązać raz na zawsze"
"Zaufanie nie stanowi kontroli, a nadzieja nie jest strategią"
"Zaufanie nie stanowi kontroli, a nadzieja nie jest strategią"