Tackling Travel Behaviour: An approach based on Fuzzy Cognitive Maps

Bello

Vanhoof

2014

IDA

Self Cite

Fuzzy Cognitive Maps (FCM) may be defined as Recurrent Neural Networks that allow causal reasoning. According to the transformation function used for updating the activation value of concepts they can be characterized as discrete or continuous. It is remarkable that FCM having discrete neurons never exhibit chaotic states, but this premise cannot be guaranteed for FCM having continuous concepts. On the other hand, complex Sigmoid FCM resulting from experts or learning algorithms often show chaotic or cyclic patterns, therefore leading to confusing interpretation of the investigated system. The first contribution of this paper is focused on explaining why most studies on FCM stability are not applicable to FCM used on classification or decision-making tasks. Next we describe a non-direct learning methodology based on Swarm Intelligence for improving the system stability once the causal weight estimation is done. The objective here is to find a specific threshold function for each map neuron simulating an external stimulus, instead of using the same transformation function for all concepts. At the end, we can compute more stable maps, so better consistency in hidden patterns is achieved. G. Nápoles et al. / How to improve the convergence on sigmoid FCM?decision-making tasks, risk analysis, prediction, text categorization, pattern recognition, management, and classification. However, estimating parameters that characterize the whole system (e.g. the causal weight matrix) may be tedious for humans, leading to inefficient models. In order to increase the reliability of FCM-based models several learning algorithms for tuning such parameters have been introduced [10], although most of them are focused on computing causal relations between map concepts.On the other hand, existing approaches suppose that FCM are closed systems and they do not consider external influences, while other factors such as the FCM stability are frequently ignored. As far as known, there is no existence of any learning method for enhancing the system stability once the system causality is established. For example, let us suppose a FCM resulting from experts where causal connections may be partially modified (e.g. we know the direction of causalities and an approximation of their values that should be preserved). Can we expect lineal stability in the final map? If not, how to improve the stability of the system without affecting causal connections estimated by experts?In the literature a few researches concerning FCM convergence have been proposed. For instance, Kosko [4] developed an analytic method based on Liapounov functions for reaching stable solutions on Feedback Standard Additive Models (SAM -which share several of the FCM characteristics). Unfortunately, Kosko concluded that such conditions cannot be extended to FCM due to the large number of feedback links involved in FCM-based models. More recently, other analytical methods were introduced (see Section 3), but we believe that such approaches are mostly useful for stabilizing FCM used in...

Section: Stability On Non-discrete Fuzzy Cognitive Mapsmentioning

confidence: 99%

How to improve the convergence on sigmoid Fuzzy Cognitive Maps?

Bello

Vanhoof

2014

IDA

Self Cite

“…Therefore the FCM theory seems to be an appropriate choice. Recently, M. León et al [25,26] proposed a novel modeling based on Cognitive Mapping theory to characterize mental representations of transportation experts.…”

Section: Second Case Study: Travel Behavior Analysis Using Fcm Theorymentioning

confidence: 99%

“…From the machine learning point of view, the proposed topology leads to a complex multi-objective regression problem, where the expected utility that reports the selection of a specific transport mode should be estimated. However, as a result from [26], we know that some concepts are severely susceptible to expert's criteria and they are often superfluous or contradictory. Then, finding central nodes could reduce the map complexity, allowing the knowledge extraction tasks.…”

Section: Proposed Fcm Topologymentioning

confidence: 99%

Learning of Fuzzy Cognitive Maps for simulation and knowledge discovery

Procedings of the Fourth International Workshop on Knowledge Discovery, Knowledge Management and Decision Support

Grau

Pérez-García

et al. 2013

Self Cite

In recent years Fuzzy Cognitive Maps (FCM) has become a useful Soft Computing technique for modeling and simulation. They are connectionist and recurrent structures involving concepts describing the system behavior, and causal connections. This paper describes two abstract models based on Swarm Intelligence for learning parameters characterizing FCM, which is a central issue on this field. At the end, we obtain accurate maps, allowing the simulation of the system and also the extraction of relevant knowledge associated with underlying patterns.

“…FCMs have received increasing attention among researchers and both practical and theoretical results have been introduced. Some representative application fields include: decision making [5], system control [6], engineering [7], protein modeling [8], transport management [9], intrusion detection [10], etc. Also, Papakostas et al [11] introduced FCM-based classifiers as light grey box models, being used for classification tasks.…”

Section: Introductionmentioning

confidence: 99%

FCM Expert: Software Tool for Scenario Analysis and Pattern Classification Based on Fuzzy Cognitive Maps

Int. J. Artif. Intell. Tools

Espinosa

Grau

et al. 2018

Self Cite

Fuzzy Cognitive Maps (FCMs) have become a suitable and proven knowledge-based methodology for systems modeling and simulation. This technique is especially attractive when modeling systems characterized by ambiguity, and/or non-trivial causalities among its variables. The rich literature that is found related to FCMs reports very clearly many successful studies solved through the use of FCMs; however, when it comes to software implementations, where domain experts can design FCM-based systems, run simulations or perform more advanced experiments, not much is found or documented. The few existing implementations are not proficient in providing options for experimentation. Therefore, we believe that a gap exists, specifically between the theoretical advances and the development of accurate, transparent and sound FCM-based systems; and we advocate for the creation of more complete and exible software products. The goal of this paper is to introduce “FCM Expert”, a software tool for fuzzy cognitive modeling, where we focus on scenario analysis and pattern classification. The main features of FCM Expert rely on Machine Learning algorithms to compute the parameters that might define a model, optimize its network topology and improve the system convergence without losing information. Also, FCM Expert allows performing WHAT-IF simulations and studying the system behavior through a friendly, intuitive and easy-to-use graphical user interface.