Extended sequential adaptive fuzzy inference system for classification problems

Rong, Hai-Jun; Sundararajan, N.; Huang, Guang-Bin; Zhao, Guiping

doi:10.1007/s12530-010-9023-9

Cited by 80 publications

(53 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The statistical contribution, however, ignores summarization power of a rule because it does not consider how strategic a current position of rule in the feature space is [24], [38]. This hinders its capability to capture concept drift because no distance information is provided in enumerating the importance of fuzzy rules.…”

Section: ) Complexity Analysismentioning

confidence: 99%

Evolving Ensemble Fuzzy Classifier

Pratama

Pedrycz

Lughofer

2018

IEEE Trans. Fuzzy Syst.

101

View full text Add to dashboard Cite

Abstract-The concept of ensemble learning offers a promising avenue in learning from data streams under complex environments because it addresses the bias and variance dilemma better than its single-model counterpart and features a reconfigurable structure, which is well-suited to the given context. While various extensions of ensemble learning for mining nonstationary data streams can be found in the literature, most of them are crafted under a static base-classifier and revisits preceding samples in the sliding window for a retraining step. This feature causes computationally prohibitive complexity and is not flexible enough to cope with rapidly changing environments. Their complexities are often demanding because it involves a large collection of offline classifiers due to the absence of structural complexities reduction mechanisms and lack of an online feature selection mechanism. A novel evolving ensemble classifier, namely Parsimonious Ensemble (pENsemble), is proposed in this paper. pENsemble differs from existing architectures in the fact that it is built upon an evolving classifier from data streams, termed Parsimonious Classifier (pClass). pENsemble is equipped by an ensemble pruning mechanism, which estimates a localized generalization error of a base-classifier. A dynamic online feature selection scenario is integrated into the pENsemble. This method allows for dynamic selection and deselection of input features on the fly. pENsemble adopts a dynamic ensemble structure to output a final classification decision where it features a novel drift detection scenario to grow the ensemble's structure. The efficacy of the pENsemble has been numerically demonstrated through rigorous numerical studies with dynamic and evolving data streams where it delivers the most encouraging performance in attaining a tradeoff between accuracy and complexity.

show abstract

Section: ) Complexity Analysismentioning

confidence: 99%

Evolving Ensemble Fuzzy Classifier

Pratama

Pedrycz

Lughofer

2018

IEEE Trans. Fuzzy Syst.

101

View full text Add to dashboard Cite

show abstract

“…In some existing literature, the conditions of adding new rules are based on the model output error and the distance of the current sample to the existing rules [7], [8], and the information potential that depends on the current sample's spatial proximity to all other data points [2], [9]- [11]. In this paper, the correntropy in (15) together with the distance between the current sample to the existing rules is used as the criteria to determine whether a rule needs to be generated. Rule adding of the CEFNS is presented as follows.…”

Section: B Rules Evolvingmentioning

confidence: 99%

“…The FNSs with static structure can hardly handle nonstationary processes, in which the modeling performance is degraded. To address this issue, intensive research work have been concentrated on developing Rong evolving FNSs (EFNSs) aiming to design FNSs with a higher level of flexibility and autonomy [2], [4]- [15]. EFNSs evolve their structure according to the information from the dynamic processes and thus are able to deal with data shift and drift due to changes in the operating environment over time.…”

Section: Introductionmentioning

confidence: 99%

“…This limits its application. To improve the performance of SAFIS, the author has developed an extended SAFIS (ESAFIS) [15] based on the modified influence of a rule in which the uniform distribution of the input data is not necessary. In [19], a multivariable Gaussian evolving fuzzy modeling system (eMG) is developed using a recursive clustering algorithm inspired by the idea of participatory learning.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Correntropy-Based Evolving Fuzzy Neural System

Bao

Rong

Angelov

et al. 2018

IEEE Trans. Fuzzy Syst.

Self Cite

View full text Add to dashboard Cite

Abstract-In this paper, a correntropy-based evolving fuzzy neural system (correntropy-EFNS) is proposed for approximation of nonlinear systems. Different from the commonly used meansquare error criterion, correntropy has a strong outliers rejection ability through capturing the higher moments of the error distribution. Considering the merits of correntropy, this paper brings contributions to build EFNS based on the correntropy concept to achieve a more stable evolution of the rule base and update of the rule parameters instead of the commonly used meansquare error criterion. The correntropy-EFNS (CEFNS) begins with an empty rule base and all rules are evolved online based on the correntropy criterion. The consequent part parameters are tuned based on the maximum correntropy criterion where the correntropy is used as the cost function so as to improve the nonGaussian noise rejection ability. The steady-state convergence performance of the CEFNS is studied through the calculation of the steady-state excess mean square error (EMSE) in two cases: i) Gaussian noise; and ii) non-Gaussian noise. Finally, the CEFNS is validated through a benchmark system identification problem, a Mackey-Glass time series prediction problem as well as five other real-world benchmark regression problems under both noise-free and noisy conditions. Compared with other evolving fuzzy neural systems, the simulation results show that the proposed CEFNS produces better approximation accuracy using the least number of rules and training time and also owns superior non-Gaussian noise handling capability.

show abstract

“…SAFIS uses a distance criterion in conjunction with an influence measure of the new rules created to develop and update the rule base. Further instances of evolving methods include [30,31,32].…”

Section: Introductionmentioning

confidence: 99%

Adaptive Fuzzy C-Regression Modeling for Time Series Forecasting

Maciel¹,

Lemos²,

Ballini³

et al. 2015

Proceedings of the 2015 Conference of the International Fuzzy Systems Association and the European Society for Fuzzy Logic and

View full text Add to dashboard Cite

The aim of the 2015 IFSA-EUSFLAT International Time Series Competition, Computational Intelligence in Forecasting (CIF), is to evaluate the performance of computational intelligence-based approaches to forecast time series of different nature. The participants must propose a unique consistent methodology for all time series. This paper suggests an adaptive fuzzy c-regression modeling approach (aFCR) for time series forecasting. The aFCR is a fuzzy clustering with affine prototypes modeling approach to develop fuzzy functional rule-based models. The approach uses participatory learning to adapt the model structure as it processes data as a stream of time series values. Computational experiments show that the aFCR forecaster is an effective tool to forecast time series.

show abstract

Extended sequential adaptive fuzzy inference system for classification problems

Cited by 80 publications

References 31 publications

Evolving Ensemble Fuzzy Classifier

Evolving Ensemble Fuzzy Classifier

Correntropy-Based Evolving Fuzzy Neural System

Adaptive Fuzzy C-Regression Modeling for Time Series Forecasting

Contact Info

Product

Resources

About