This is the user sandboxofKekasmarik. A user sandbox is a subpage of the user's user page. It serves as a testing spot and page development space for the user and is not an encyclopedia article. Create or edit your own sandbox here.
Finished writing a draft article? Are you ready to request review of it by an experienced editor for possible inclusion in Wikipedia? Submit your draft for review!
Computational motivation models, simulates or replicates motivation using a computer, to achieve one of several ends:
To better understand human motivation and to formulate an algorithmic perspective on self-motivated behaviour in humans.
Motivation is the "cause of action" in natural systems [1]. In artificial agents, such as robots or non-player characters in computer games, computational motivation causes action by focusing attention and generating goals from experiences.
Reinforcement learning agents [2] learn from trial-and-error and reward feedback. Agents experiment with different actions in each situation they encounter and received reward or punishment. They progressively map reward earned to the situations and/or actions that earn them, and favour these situations and/or actions over time.
When static reward signals are crafted around a particular goal, reinforcement learning agents will learn behaviours that solve that goal. However, with dynamic reward signals that use principles such as curiosity, novelty or competence-seeking, motivated reinforcement learning agents [3][4] are created. Motivated reinforcement learning agents can focus their attention on different goals at different times. The agent designer no longer needs to know what goals the agent may encounter during its lifetime.
Traditional particle swarm optimisation algorithms take a fitness function as input and compute successive changes to the velocities of a number of particles. Two classes of motivated particle swarm optimisation have been proposed. Algorithms in the first class employ motivation to generate a dynamic objective function as a function of spatially mapped sensor data, while optimisation is in progress [5]. Algorithms in the second class optimise a fixed fitness function, but employ computational motivation to embed diversity in the particles [6].
^Heckhausen, J; Heckhausen, H (2010), Motivation and Action, Cambridge University Presss
^Sutton, R; Barto, A (2000), Reinforcement Learning:An Introduction, The MIT Press
^Merrick, K; Maher, ML (2009), Motivated Reinforcement Learning: Curious Characters for Multiuser Games, Springer Verlag
^Singh, S; Barto, A; Chentanez, N (2005), Intrinsically motivated reinforcement learning, Advances in Neural Information Processing Systems 17 (NIPS), pp. 1281-1288.
^Klyne, A; Merrick, K (2016), Intrinsically motivated particle swarm optimisation applied to task allocation for workplace hazard detection, Adaptive Behaviour vol. 24 no. 4 219-236
^Hardhienata, M; Merrick, K; Ougrinovski, V (2012), Task allocation in multi-agent systems using models of motivation and leadership, IEEE Congress on Evolutionary Computation