ATABS: A technique for automatically training agent-based simulators

García‐Magariño, Iván; Palacios-Navarro, Guillermo

doi:10.1016/j.simpat.2016.04.003

Cited by 8 publications

(3 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The ABS was trained with the same input data as in the previous case study. In particular, the ABS used the Deck random generators, since this kind of generators usually facilitates the training of the ABSs as shown in our previous work about ATABS (a technique for Automatically Training ABSs) [11]. As an example, Figure 12 shows the control version of ABS-MindHeart running the Vipassana meditation program.…”

Section: Abs Developed With the Control Mechanismmentioning

confidence: 99%

TABSAOND: A technique for developing agent-based simulation apps and online tools with nondeterministic decisions

García‐Magariño

Palacios-Navarro

Lacuesta

2017

Simulation Modelling Practice and Theory

Self Cite

View full text Add to dashboard Cite

Agent-based simulators (ABSs) have successfully allowed practitioners to estimate the outcomes of certain input circumstances in several domains. Although some techniques and processes provide hints about the construction of these systems, there are still some aspects that are not discussed in the literature. In this context, the current approach presents a technique for developing ABSs. This technique especially discusses the simulation of the decision-making processes of agents and the different ways of evolving their attributes in nondeterministic scenarios. The current approach includes a framework for developing ABSs as mobile apps and online tools. This work illustrates the current approach with a case study in the field of health and welfare. This case study has also been implemented with the most similar technique from the literature for comparing both techniques. The presented technique improved the simulated outcomes in terms of similarity with the real ones, as proved with several statistical tests. The obtained ABS is more efficient and reliable for large amounts of agents (e.g. 10,000-400,000 agents). The development time was also lower. Both the framework and the implementation of the case study are freely distributed as open-source to ensure the reproducibility of the experiments and to assist practitioners in applying the current approach.

show abstract

Section: Abs Developed With the Control Mechanismmentioning

confidence: 99%

TABSAOND: A technique for developing agent-based simulation apps and online tools with nondeterministic decisions

García‐Magariño

Palacios-Navarro

Lacuesta

2017

Simulation Modelling Practice and Theory

Self Cite

View full text Add to dashboard Cite

show abstract

“…These internal parameters can be trained given certain known input conditions and certain real resulting sociograms following ATABS (a technique for Automatically Training ABSs) [20].…”

Section: Abs-socimentioning

confidence: 99%

ABS-SOCI: An Agent-Based Simulator of Student Sociograms

et al. 2017

Self Cite

View full text Add to dashboard Cite

Sociograms can represent the social relations between students. Some kinds of sociograms are more suitable than others for achieving a high academic performance of students. However, for now, at the beginning of an educative period, it is not possible to know for sure how the sociogram of a group of students will be or evolve during a semester or an academic year. In this context, the current approach presents an Agent-Based Simulator (ABS) that predicts the sociogram of a group of students taking into consideration their psychological profiles, by evolving an initial sociogram through time. This simulator is referred to as ABS-SOCI (ABS for SOCIograms). For instance, this can be useful for organizing class groups for some subjects of engineering grades, anticipating additional learning assistance or testing some teaching strategies. As experimentation, ABS-SOCI has been executed 100 times for each one of four real scenarios. The results show that ABS-SOCI produces sociograms similar to the real ones considering certain sociometrics. This similarity has been corroborated by statistical binomial tests that check whether there are significant differences between the simulations and the real cases. This experimentation also includes cross-validation and an analysis of sensitivity. ABS-SOCI is free and open-source to (1) ensure the reproducibility of the experiments; (2) to allow practitioners to run simulations; and (3) to allow developers to adapt the simulator for different environments.

show abstract

“…Then, , is set to a specific probability value if the user is a beginner or is set to a different value otherwise. The selection of these two probability values can be established in a calibration phase of the system, for example, following the technique for Automatically Training ABS (ATABS) [21].…”

Section: Behaviors Of User Agentsmentioning

confidence: 99%

ABS-SmartPriority: An Agent-Based Simulator of Strategies for Managing Self-Reported Priorities in Smart Cities

García‐Magariño¹,

Lacuesta²

2017

Wireless Communications and Mobile Computing

Self Cite

View full text Add to dashboard Cite

Smart cities still need the appropriate tools for allowing researchers to contribute in this growing field that will change the comfort and qualities of live of citizens. Smart cities can provide services such as informing of the less overcrowded tourist routes, pathfinding for avoiding traffic jams, search services for parking, collecting tree branches from streets, and the finding of the nearest available public electric bicycles or cars. The levels of urgencies differ from some situations to others. Examples of urgent matters are the fast transportation critical patients and the removal of obstacles in main roads. Since smart cities are meant to manage huge amounts of requests, priorities should be automated and managed in a flexible way for new scenarios. Smart cities may need citizens to report the service priorities. However, citizens may have different criteria and could abuse the highest priority levels hindering the service performance for really urgent matters. A novel agent-based simulation open-source framework is proposed for testing different policies for normalizing and controlling self-reported priorities, with its simulator called ABS-SmartPriority. This approach is illustrated by simulating two different policies, in which the smart policy outperformed the one used as the control mechanism.

show abstract

ATABS: A technique for automatically training agent-based simulators

Cited by 8 publications

References 18 publications

TABSAOND: A technique for developing agent-based simulation apps and online tools with nondeterministic decisions

TABSAOND: A technique for developing agent-based simulation apps and online tools with nondeterministic decisions

ABS-SOCI: An Agent-Based Simulator of Student Sociograms

ABS-SmartPriority: An Agent-Based Simulator of Strategies for Managing Self-Reported Priorities in Smart Cities

Contact Info

Product

Resources

About