Behavioural modelling

Functional building block

Behavioural modelling is used to derive the actions that an intelligent agent should take given a certain set of environmental variables to reach a defined goal. Algorithms that perform behavioural modelling learn using reinforcement learning.

In a very wide sense, behavioural modelling could be regarded as a type of classification because the intelligent agent typically makes choices between sets of possible actions, but the comparison is laboured because:

  • behavioural modelling aims to learn sequences of appropriate choices to reach a goal or goals, which is not the case for classification functions.
     
  • the range of possible actions for an intelligent agent (the output) may vary very considerably according to the state of the environment (the input), which is not the case for classification functions.
alias
used by
ALG_Actor-critic ALG_Deep Q-network ALG_Monte-Carlo tree search ALG_Neural actor-critic ALG_Policy gradient estimation ALG_Q-learning ALG_SARSA ALG_Temporal difference learning SPT_Evolutionary selection