Showing posts from April, 2024

How to beat you

  Scope of planning When you query a search engine for pictures on a specific topic (for example, an airplane with stirrup engines) and select one of the ones found, you are training a Neural Network. Millions of users, requesting a search engine and making a choice, train it even more extensively. But your queries and choices can be recorded and used to train future versions of Neural Networks (and do it faster). Therefore, the emergence of Large Distributed Fake Systems is inevitable, which will introduce incorrect training data into Neural Networks (or Artificial Intelligence). And the scale of systems generating queries and making choices must be enormous to have an advantage or maintain a leading position in the development of Neural Networks. New reality

Naval unmanned attack boats

  Opinion Naval unmanned attack boats have demonstrated their effectiveness. The issue of protecting ships from them is very important. This is probably important for existing ships. One option is narrow pontoons along each side. With a streamlined shape, they will slow down the ship less, but will significantly reduce roll and prevent an unmanned boat from hitting the side of the ship. This is some kind of trimaran. Moreover, it is possible to provide for the release of pontoons to achieve maximum speed. The bow and stern of the ship will remain uncovered. But these are narrow sectors that can be protected with rapid-fire systems. Ships systems