Nevermind Creations

Deep Player Behavior Modeling


  • Deep Player Behavior Modeling is a machine learning technique that replicates individual player decision making.

  • Atomic state-action pairs are recorded for all symbolic player inputs, so that game state, player state, target state and action history are mapped to a distribution among predicted following actions.

  • Players reported to recognize their own behavioral patterns and enjoyed the challenge of opposing a believable, continuously learning foe.

  • In online multiplayer sessions, players did not notice when disconnected fellow players were replaced with their individual substitute.

  • Above that, DPBM can also be used for large-scale automated balancing simulations to closely resemble a whole population of players.

Selected Conference Talks:

Publications

Published Games