Hackaday.com Features Neural Network Simulation of a Pet Squid

The website Hackaday.com presents a neural network simulation of a pet squid, available on GitHub under the name Dosidicus. This simulation allows the squid to move autonomously by making decisions based on its current state, such as hunger or sleepiness. The squid uses a vision cone to detect food, simulating a realistic food-seeking behavior. The neural network can make decisions and form associations, with weights analyzed, adjusted, and trained by a Hebbian learning algorithm. Short-term and long-term memory experiences influence decision-making, and the squid can create new neurons in response to its environment (neurogenesis).