NEW YORK - One of the most well-known artificial intelligence (AI) methods is called particle swarm optimization (PSO), which is inspired by the movements of fish in the sea when they are searching for food. In this method, each fish is responsible for moving in a certain direction to find food, and per each fish’s movement, it should then tell the other fish where to swim. In other words, if one fish gets closer to the food, then the other fish should also move toward that ‘winner’ fish. This means that there is cooperation between fish. Fig. 1 shows how this group movement of fish works:

Fig. 1) Group movement of fish in the sea

The movement, velocity, and direction of fish’s movements are defined by movement equations in physics. These equations are as follows:

The first equation is responsible for defining the movement value, and the second equation is calculating the velocity per each movement of a fish. The purple highlighted part of the velocity equation is what helps each fish to select the best movements in competition with itself. The green highlighted part defines the movement of all fish toward the best direction - in terms of the food searching process - in competition with all the fish. At each step, there is a fitness function or cost function that defines how much closer fish are getting to the food. The closer they get to the target, the better (cost) value the function gives. This kind of intelligent method, which is based on cooperation between searching agents, is called swarm intelligence.

PSO is one of the most advanced methods of training a neural network to predict diseases,