The process of online reinforcement learning also creates a stream of experiences that an agent can store to re-learn from them. In this work, we introduce a concept of artificial perception affecting the dynamics of experience memory replay, which induces a secondary goal-directed drive that complements the main goal defined by the reinforcement function. The different perception dynamics are capable of inducing different “personality” types able to govern the agent behavior, possibly enabling it to exhibit an improved performance over an environment with specific characteristics. Experimental results show that different personalities show different performance levels when facing environment variations, therefore, showcasing the influence of artificial perception in agent’s adaptation.
AhnH.PicardR. W. (2005). Affective-cognitive learning and decision making: A motivational reward framework for affective agents. In TaoJ.TanT.PicardR. W. (Eds.), International Conference on Affective Computing and Intelligent Interaction (pp. 866–873). New York, NY: Springer.
2.
BriggsK. C. (1976). Myers-Briggs type indicator. Palo Alto, CA: Consulting Psychologists Press.
3.
BrockmanG.CheungV.PetterssonL.SchneiderJ.SchulmanJ.TangJ.ZarembaW. (2016). OpenAI gym. Retrieved from https://gym.openai.com/
4.
GadanhoS. C.HallamJ. (April, 1998). Exploring the role of emotions in autonomous robot learning (DAI Research Paper). Edinburgh, UK: The University of Edinburgh.
5.
GolemanD. (2006). Emotional intelligence. New York, NY: Bantam Books.
6.
JirenhedD. A.HesslowG.ZiemkeT. (2001). Exploring internal simulation of perception in mobile robots. In Proceedings of the 4th European Workshop on Advanced Mobile Robots (pp. 107–113). Retrieved from https://portal.research.lu.se/portal/files/5662932/4407145.pdf
7.
JohnO. P.SrivastavaS. (1999). The big five trait taxonomy: History, measurement, and theoretical perspectives. In PervinL. A.JohnO. P. (Eds.), Handbook of personality: Theory and research (pp. 102–138). New York, NY: Guilford Press.
MetcalfK.TheobaldB. J.ApostoloffN. (2018). Learning sharing behaviors with arbitrary numbers of agents. In Proceedings of the 17th International Conference on Autonomous Agents and Multiagent Systems (pp. 1232–1240). International Foundation for Autonomous Agents and Multiagent Systems.
11.
MnihV.KavukcuogluK.SilverD.GravesA.AntonoglouI.WierstraD.RiedmillerM. (2013). Playing atari with deep reinforcement learning. Arxiv Preprint Arxiv:1312.5602. Retrieved from https://arxiv.org/abs/1312.5602
12.
RamicicM.BonariniA. (2017a). Attention-based experience replay in deep Q-learning. In Proceedings of the 9th International Conference on Machine Learning and Computing (pp. 476–481). New York, NY: ACM.
13.
RamicicM.BonariniA. (2017b). Entropy-based prioritized sampling in deep Q-learning. In 2nd International Conference on Image, Vision and Computing (ICIVC) (pp. 1068–1072). New York, NY, USA: IEEE.
14.
RamicicM.BonariniA. (2019). Selective perception as a mechanism to adapt agents to the environment: An evolutionary approach. IEEE Transactions on Cognitive and Developmental Systems. Retrieved from https://ieeexplore.ieee.org/document/8630035
15.
RecchiaT.ChungJ.PochirajuK. (2014). Performance of heterogeneous robot teams with personality adjusted learning. Biologically Inspired Cognitive Architectures, 7, 87–97.
16.
RumbellT.BarndenJ.DenhamS.WennekersT. (2012). Emotions in autonomous agents: Comparative analysis of mechanisms and functions. Autonomous Agents and Multi-agent Systems, 25, 1–45.
17.
SalichsM. A.MalfazM. (2006). Using emotions on autonomous agents: The role of happiness, sadness and fear. Integrative Approaches to Machine Consciousness, Part of AISB, 6, 157–164.
18.
SatoS.NozawaA.IdeH. (2004). Characteristics of behavior of robots with emotion model. IEEJ Transactions on Electronics, Information and Systems, 124, 1390–1395.
ShanahanM. (2006). A cognitive architecture that combines internal simulation with a global workspace. Consciousness and Cognition, 15, 433–449.
21.
SunR.WilsonN. (2014a). A model of personality should be a cognitive architecture itself. Cognitive Systems Research, 29, 1–30.
22.
SunR.WilsonN. (2014b). Roles of implicit processes: Instinct, intuition, and personality. Mind & Society, 13, 109–134.
23.
SuttonR. S. (1990). Integrated architectures for learning, planning, and reacting based on approximating dynamic programming. In Machine Learning Proceedings 1990 (pp. 216–224). Morgan Kaufmann.
24.
SuttonR. S.BartoA. G. (1998). Reinforcement learning: An introduction. Cambridge, MA: The MIT press.
25.
SuttonR. S.SzepesváriC.GeramifardA.BowlingM. P. (2012). Dyna-style planning with linear function approximation and prioritized sweeping. Arxiv Preprint Arxiv:1206.3285. Retrieved from https://arxiv.org/abs/1206.3285
26.
TsankovaD. D. (2009). Emotional intervention on an action selection mechanism based on artificial immune networks for navigation of autonomous agents. Adaptive Behavior, 17, 135–152.
27.
ZhaiJ.LiuQ.ZhangZ.ZhongS.ZhuH.ZhangP.SunC. (2016). Deep Q-learning with prioritized sampling. In HiroseA.OzawaS.DoyaK.IkedaK.LeeM.LiuD. (Eds.), Neural Information Processing (Vol. 9947). ICONIP 2016. Lecture Notes in Computer Science. Cham, Switzerland: Springer.
28.
ZiemkeT.JirenhedD. A.HesslowG. (2005). Internal simulation of perception: A minimal neuro-robotic model. Neurocomputing, 68, 85–104.
