Strategy AI

 How Smart Machines Think by Sean Gerrish feels relevant to many advanced algorithms, simulations, #AI, and decision-making applications. For example, chapters on #reinforcementlearning or #neuralnetworks are relatively well-known. However, the section on Real-Time AI and #Starcraft is significant. It is hard to find a better platform if one needs a starting point for building an AI system capable of theater-wide decision-making. It describes one of the most advanced and challenging AI architectures or, more precisely, simulation platforms. Starcraft is a real-time complex simulation/strategy game with a deep decision space. An opportunity for machine learning or optimization is second to none on this platform. Some of the critical characteristics of the platform include a vast sensing capability, the complex status of each actor, a wide range of actions, spatial analytics, tracking of the enemy and friendly units, reasoning time frames for strategy, short-term tactics, or reactive control.

#machinelearning #algorithms #building #strategy #strategygames