Social Action in Socially Situated Agents

Barnes, Chloe M.; Ekárt, Anikó; Lewis, Peter R.

doi:10.1109/saso.2019.00021

Cited by 9 publications

(22 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Agents use a neural network learning architecture introduced by Barnes et al (2019), which is a simplified version of other RCT agent learning architectures proposed by Robinson et al (2007) and Borg et al (2011). The first of the two artificial neural networks in this learning architecture is a feedforward deliberative network that determines target locations to move to.…”

Section: The River Crossing Taskmentioning

confidence: 99%

“…Activations can also be highly negative, indicating that the location is to be avoided (e.g., a successful agent may learn to avoid Figure 1. A simplified RCT environment inspired by Robinson et al (2007), adapted from Barnes et al (2019). The goal of the agent is to collect the resource object on the opposite side of the river; to do this, it must learn to build a bridge in the river with a stone.…”

Section: The River Crossing Taskmentioning

confidence: 99%

“…Task); agents employed a hybrid neural architecture to express reactive and deliberative behaviours that enabled navigation and interaction with the environment. Further, Barnes et al (2019) explored how the pursuit of individual goals was affected when agents coevolved in a multi-agent, gamified version of the task (RCD), in which agents shared an environment with an unknown other. Additionally, a simple technique inspired by the theory of social action was shown to mitigate evolutionary volatility arising from actions of other agents within the environment, without the need for agents to be aware of the existence of others; this unintended influence should be considered when designing systems that are colocated or operate in shared environments.…”

Section: What Can We Learn From River Crossing Tasks?mentioning

confidence: 99%

“…The RCT was originally developed to explore how agents learn to solve increasingly complex problems in tasks; initially this was done using a novel neural network architecture that separated reactive from deliberative processes. Since then, the testbed has been extended to explore the effect of social learning strategies ( Jolley et al, 2016), learning by imitation (Borg et al, 2011), and social action (Barnes et al, 2019), as well as how agents learn in a 3D world (Stanton & Channon, 2015). One of the difficult features of the RCT and its extensions is that agents must learn sub-tasks in order to achieve their goal.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Explaining Evolutionary Agent-Based Models via Principled Simplification

2021

Self Cite

View full text Add to dashboard Cite

Understanding how evolutionary agents behave in complex environments is a challenging problem. Agents can be faced with complex fitness landscapes derived from multi-stage tasks, interaction with others, and limited environmental feedback. Agents that evolve to overcome these can sometimes access greater fitness, as a result of factors such as cooperation and tool use. However, it is often difficult to explain why evolutionary agents behave in certain ways, and what specific elements of the environment or task may influence the ability of evolution to find goal-achieving behaviours; even seemingly simple environments or tasks may contain features that affect agent evolution in unexpected ways. We explore principled simplification of evolutionary agent-based models, as a possible route to aiding their explainability. Using the River Crossing Task (RCT) as a case study, we draw on analysis in the Minimal River Crossing (RC-) Task testbed, which was designed to simplify the original task while keeping its key features. Using this method, we present new analysis concerning when agents evolve to successfully complete the RCT. We demonstrate that the RC- environment can be used to understand the effect that a cost to movement has on agent evolution, and that these findings can be generalised back to the original RCT. Then, we present new insight into the use of principled simplification in understanding evolutionary agents. We find evidence that behaviour dependent on features that survive simplification, such as problem structure, are amenable to prediction; while predicting behaviour dependent on features that are typically reduced in simplification, such as scale, can be invalid.

show abstract

Section: The River Crossing Taskmentioning

confidence: 99%

Section: The River Crossing Taskmentioning

confidence: 99%

Section: What Can We Learn From River Crossing Tasks?mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Explaining Evolutionary Agent-Based Models via Principled Simplification

2021

Self Cite

View full text Add to dashboard Cite

show abstract

“…Continuing the theme of broader views on self-adaptive systems, the study of socio-technical, purely social, or democratic legal systems as self-adaptive or self-integrating examples is a long-running sub-area of the community. Barnes, Ekárt, and Lewis explore the idea of social action in socially situated agents [18], considering scenarios in which two (generally human) systems are introduced into the same world and how this impacts the goals and actions of those systems. This kind of study can lead to fresh insight on how technical systems can similarly negotiate their actions in unknown environments with other potentially competing actors.…”

Section: Altmentioning

confidence: 99%

A Survey of Methodology in Self-Adaptive Systems Research

Porter

Filho

Dean

2020

2020 IEEE International Conference on Autonomic Computing and Self-Organizing Systems (ACSOS)

View full text Add to dashboard Cite

Major research venues on autonomic and selfadaptive systems have been active for 16 years, exploring and building on the seminal vision of autonomic computing in 2003. We study the current trajectory and progress of the research field towards this vision, surveying on the research questions that are asked by researchers and the methodological practice that they employ in order to answer these questions. We survey contributions under this lens across the three main venues for primary research in autonomic and self-adaptive systems work: ICAC, SASO, and SEAMS. We examine the last three years of contributions from each venue, totaling 210 publications, to gain an understanding of the dominant current research questions and methodological practice-and what this shows us about the progress of the field. Our major findings include the general research questions focusing one level below the highest autonomy vision; methodological practice being split almost evenly between real-world experiments and simulation; a general a high positive results bias in publications; and a potential concern over low levels of repeatability across most contributions.

show abstract