BNAIC 2017 Session: Agent Systems, Wednesday 8 November 2017, 13.00h

Session chair: Harmen de Weerd

////BNAIC 2017: Session report “Agent Systems 2”

The Agent Systems session was opened by Bram Wiggers from the University of Groningen. In his paper “The origin of mimicry: Deception or merely coincidence?”, he presents an agent model for the evolution of mimicry. Mimicry refers to a situation in which two species of prey animals share a phenotype, while only one of them is dangerous to eat for predators. In this situation, predators avoid eating both species, which allows the harmless prey to benefit from the presence of the dangerous prey. Wiggers showed how mimicry could evolve through deception, where harmless prey evolve to have the same phenotype as the dangerous prey, but also through coincidence.

Next, Daniel Formolo presented his joint work with Natalie van der Wal on “Simulating Collective Evacuations with Social Elements”. He argued that while current simulations for evacuations are helpful, but they do not model human behaviour that corresponds to reality. In particular, current models lack socio-cultural modeling such as trips and falls, fear, social contagion, helping behaviour, or familiarity with the environment. He showed that especially travelling in groups reduces evacuation time. Formolo concluded by saying that the model with socio-cultural modeling has the potential to predict human behaviour in emergency situations more accurately than the current standard.

In the final presentation of the session, Jens Nevens presented his work with Katrien Beuls, titled “The Effect of Tutor Feedback in Language Acquisition Models”. Nevens argued that an important aspect of teaching and learning is referent learning, which is learning what physical object is referred to by a given word. In particular, he investigated the role of social feedback, such as pointing or gazing at the object. In his agent model for a color guessing game, Nevens compared cross-situational learning, in which no feedback is given, with interactive learning, in which social feedback is always given. The results show that social feedback influences learning a language in two ways: it allows the learner to understand the referent of a given word more quickly, and also better. In addition, the model showed that the tutor’s strategy for selecting topics greatly influences the learner’s learning.

By |2018-03-01T20:36:09+00:00March 1st, 2018|Event reports|0 Comments