Interpretability assessment of fuzzy knowledge bases: A cointension based approach

Computing with words (CWW) relies on linguistic representation of knowledge that is processed by operating at the semantical level defined through fuzzy sets. Linguistic representation of knowledge is a major issue when fuzzy rule based models are acquired from data by some form of empirical learning. Indeed, these models are often requested to exhibit interpretability, which is normally evaluated in terms of structural features, such as rule complexity, properties on fuzzy sets and partitions. In this paper we propose a different approach for evaluating interpretability that is based on the notion of cointension. The interpretability of a fuzzy rule-based model is measured in terms of cointension degree between the explicit semantics, defined by the formal parameter settings of the model, and the implicit semantics conveyed to the reader by the linguistic representation of knowledge. Implicit semantics calls for a representation of user's knowledge which is difficult to externalise. Nevertheless, we identify a set of properties -- which we call “logical view” -- that is expected to hold in the implicit semantics and is used in our approach to evaluate the cointension between explicit and implicit semantics. In practice, a new fuzzy rule base is obtained by minimising the fuzzy rule base through logical properties. Semantic comparison is made by evaluating the performances of the two rule bases, which are supposed to be similar when the two semantics are almost equivalent. If this is the case, we deduce that the logical view is applicable to the model, which can be tagged as interpretable from the cointension viewpoint. These ideas are then used to define a strategy for assessing interpretability of fuzzy rule-based classifiers (FRBCs). The strategy has been evaluated on a set of pre-existent FRBCs, acquired by different learning processes from a well-known benchmark dataset. Our analysis highlighted that some of them are not cointensive with user's knowledge, hence their linguistic representation is not appropriate, even though they can be tagged as interpretable from a structural point of view.

[1]  Lotfi A. Zadeh,et al.  The concept of a linguistic variable and its application to approximate reasoning - II , 1975, Inf. Sci..

[2]  Giovanna Castellano,et al.  A neuro-fuzzy network to generate human-understandable knowledge from data , 2002, Cognitive Systems Research.

[3]  Paulo Fazendeiro,et al.  A Working Hypothesis on the Semantics/Accuracy Synergy , 2005, EUSFLAT Conf..

[4]  A ZadehLotfi.  Toward a theory of fuzzy information granulation and its centrality in human reasoning and fuzzy logic , 1997 .

[5]  Andri Riid,et al.  Interpretability improvement of fuzzy systems: Reducing the number of unique singletons in zeroth order Takagi-Sugeno systems , 2010, International Conference on Fuzzy Systems.

[6]  Rudolf Kruse,et al.  Obtaining interpretable fuzzy classification rules from medical data , 1999, Artif. Intell. Medicine.

[7]  Lotfi A. Zadeh,et al.  Fuzzy logic = computing with words , 1996, IEEE Trans. Fuzzy Syst..

[8]  Joos Vandewalle,et al.  Constructing fuzzy models with linguistic integrity from numerical data-AFRELI algorithm , 2000, IEEE Trans. Fuzzy Syst..

[9]  Lotfi A. Zadeh,et al.  Is there a need for fuzzy logic? , 2008, NAFIPS 2008 - 2008 Annual Meeting of the North American Fuzzy Information Processing Society.

[10]  Roberto Guerrieri,et al.  An enhanced two-level Boolean synthesis methodology for fuzzy rules minimization , 1995, IEEE Trans. Fuzzy Syst..

[11]  John Wang,et al.  Encyclopedia of Data Warehousing and Mining , 2005 .

[12]  Yeung Yam,et al.  Trade-off between approximation accuracy and complexity: TS controller design via HOSVD based complexity minimization , 2003 .

[13]  Brigitte Charnomordic,et al.  Generating an interpretable family of fuzzy partitions from data , 2004, IEEE Transactions on Fuzzy Systems.

[14]  Mo-Yuen Chow,et al.  Heuristic constraints enforcement for training of and knowledge extraction from a fuzzy/neural architecture. I. Foundation , 1999, IEEE Trans. Fuzzy Syst..

[15]  Francisco Herrera,et al.  Integration of an Index to Preserve the Semantic Interpretability in the Multiobjective Evolutionary Rule Selection and Tuning of Linguistic Fuzzy Systems , 2010, IEEE Transactions on Fuzzy Systems.

[16]  Beatrice Lazzerini,et al.  Learning concurrently partition granularities and rule bases of Mamdani fuzzy systems in a multi-objective evolutionary framework , 2009, Int. J. Approx. Reason..

[17]  Mo-Yuen Chow,et al.  Heuristic constraints enforcement for training of and rule extraction from a fuzzy/neural architecture. II. Implementation and application , 1999, IEEE Trans. Fuzzy Syst..

[18]  Robert Babuška,et al.  A multi-objective evolutionary algorithm for fuzzy modeling , 2001, Proceedings Joint 9th IFSA World Congress and 20th NAFIPS International Conference (Cat. No. 01TH8569).

[19]  Detlef D. Nauck,et al.  Measuring interpretability in rule-based classification systems , 2003, The 12th IEEE International Conference on Fuzzy Systems, 2003. FUZZ '03..

[20]  Witold Pedrycz,et al.  Optimization of fuzzy models , 1996, IEEE Trans. Syst. Man Cybern. Part B.

[21]  Uzay Kaymak,et al.  Similarity measures in fuzzy rule base simplification , 1998, IEEE Trans. Syst. Man Cybern. Part B.

[22]  Lotfi A. Zadeh.  Toward Human Level Machine Intelligence - Is It Achievable? The Need for a Paradigm Shift , 2008 .

[23]  D. Tikk,et al.  Exact trade-off between approximation accuracy and interpretability: solving the saturation problem for certain FRBSs , 2003 .

[24]  Moshe Sipper,et al.  Fuzzy CoCo: Balancing Accuracy and Interpretability of Fuzzy Models by Means of Coevolution , 2003 .

[25]  Lotfi A. Zadeh,et al.  Toward a theory of fuzzy information granulation and its centrality in human reasoning and fuzzy logic , 1997, Fuzzy Sets Syst..

[26]  José M. Alonso,et al.  KBCT: a knowledge extraction and representation tool for fuzzy logic based systems , 2004, 2004 IEEE International Conference on Fuzzy Systems (IEEE Cat. No.04CH37542).

[27]  Anna Maria Fanelli,et al.  Interpretability constraints for fuzzy information granulation , 2008, Inf. Sci..

[28]  Witold Pedrycz,et al.  Granular computing: an introduction , 2001, Proceedings Joint 9th IFSA World Congress and 20th NAFIPS International Conference (Cat. No. 01TH8569).

[29]  John R. Anderson,et al.  MACHINE LEARNING An Artificial Intelligence Approach , 2009 .

[30]  J. Casillas.  Interpretability issues in fuzzy modeling , 2003 .

[31]  Lars Niklasson,et al.  Accuracy vs. comprehensibility in data mining models , 2004 .

[32]  Zhi-Hua Zhou.  Comprehensibility of Data Mining Algorithms , 2005 .

[33]  Witold Pedrycz,et al.  A Multiobjective Design of a Patient and Anaesthetist-Friendly Neuromuscular Blockade Controller , 2007, IEEE Transactions on Biomedical Engineering.

[34]  Bernhard Sendhoff,et al.  Extracting Interpretable Fuzzy Rules from RBF Networks , 2003, Neural Processing Letters.

[35]  Yaochu Jin,et al.  Fuzzy modeling of high-dimensional systems: complexity reduction and interpretability improvement , 2000, IEEE Trans. Fuzzy Syst..

[36]  José M. Alonso,et al.  Combining user's preferences and quality criteria into a new index for guiding the design of fuzzy systems with a good interpretability-accuracy trade-off , 2010, International Conference on Fuzzy Systems.

[37]  G. A. Miller.  THE PSYCHOLOGICAL REVIEW THE MAGICAL NUMBER SEVEN, PLUS OR MINUS TWO: SOME LIMITS ON OUR CAPACITY FOR PROCESSING INFORMATION 1 , 1956 .

[38]  Lotfi A. Zadeh,et al.  From Computing with Numbers to Computing with Words - from Manipulation of Measurements to Manipulation of Perceptions , 2005, Logic, Thought and Action.

[39]  R. P. Paiva,et al.  Merging and Constrained Learning for Interpretability in Neuro-Fuzzy Systems , 2003 .

[40]  Antonio A. Márquez,et al.  A multi-objective evolutionary algorithm with an interpretability improvement mechanism for linguistic fuzzy systems with adaptive defuzzification , 2010, International Conference on Fuzzy Systems.

[41]  Ulises Cortés,et al.  Non‐monotonic characterization of induction and its application to inductive learning , 1995, Int. J. Intell. Syst..

[42]  Witold Pedrycz,et al.  Logic Minimization as an Efficient Means of Fuzzy Structure Discovery , 2008, IEEE Transactions on Fuzzy Systems.

[43]  Giovanna Castellano,et al.  DCf: a double clustering framework for fuzzy information granulation , 2005, 2005 IEEE International Conference on Granular Computing.

[44]  Luis Magdalena,et al.  HILK: A new methodology for designing highly interpretable linguistic knowledge bases using the fuzzy logic formalism , 2008 .

[45]  F. Herrera,et al.  Accuracy Improvements in Linguistic Fuzzy Modeling , 2003 .

[46]  José M. Alonso,et al.  Looking for a good fuzzy system interpretability index: An experimental approach , 2009, Int. J. Approx. Reason..

[47]  Lotfi A. Zadeh,et al.  The Concepts of a Linguistic Variable and its Application to Approximate Reasoning , 1975 .

[48]  John Q. Gan,et al.  Extracting Takagi-Sugeno Fuzzy Rules with Interpretable Submodels via Regularization of Linguistic Modifiers , 2009, IEEE Transactions on Knowledge and Data Engineering.

[49]  Moshe Sipper,et al.  Fuzzy CoCo: a cooperative-coevolutionary approach to fuzzy modeling , 2001, IEEE Trans. Fuzzy Syst..

[50]  E. McCluskey.  Minimization of Boolean functions , 1956 .

[51]  Luis Magdalena,et al.  Expert guided integration of induced knowledge into a fuzzy knowledge base , 2006, Soft Comput..

[52]  Giovanna Castellano,et al.  On the Role of Interpretability in Fuzzy Data Mining , 2007, Int. J. Uncertain. Fuzziness Knowl. Based Syst..

[53]  José Valente de Oliveira,et al.  Semantic constraints for membership function optimization , 1999, IEEE Trans. Syst. Man Cybern. Part A.

[54]  Alessio Botta,et al.  Context adaptation of fuzzy systems through a multi-objective evolutionary approach based on a novel interpretability index , 2008, Soft Comput..

[55]  Hisao Ishibuchi,et al.  Analysis of interpretability-accuracy tradeoff of fuzzy systems by multiobjective fuzzy genetics-based machine learning , 2007, Int. J. Approx. Reason..

[56]  Derek A. Linkens,et al.  Elicitation and fine-tuning of fuzzy control rules using symbiotic evolution , 2004, Fuzzy Sets Syst..

[57]  John Q. Gan,et al.  Low-level interpretability and high-level interpretability: a unified view of data-driven interpretable fuzzy system modelling , 2008, Fuzzy Sets Syst..

[58]  Magne Setnes,et al.  Compact and transparent fuzzy models and classifiers through iterative complexity reduction , 2001, IEEE Trans. Fuzzy Syst..

[59]  Gary G. Yen,et al.  Quantitative measures of the accuracy, comprehensibility, and completeness of a fuzzy expert system , 2002, 2002 IEEE World Congress on Computational Intelligence. 2002 IEEE International Conference on Fuzzy Systems. FUZZ-IEEE'02. Proceedings (Cat. No.02CH37291).

[60]  Andri Riid,et al.  Interpretability of Fuzzy Systems and Its Application to Process Control , 2007, 2007 IEEE International Fuzzy Systems Conference.

[61]  Ryszard S. Michalski,et al.  A theory and methodology of inductive learning , 1993 .

[62]  Rudolf Kruse,et al.  A neuro-fuzzy approach to obtain interpretable fuzzy systems for function approximation , 1998, 1998 IEEE International Conference on Fuzzy Systems Proceedings. IEEE World Congress on Computational Intelligence (Cat. No.98CH36228).

[63]  Hannu Koivisto,et al.  Fuzzy classifier identification using decision tree and multiobjective evolutionary algorithms , 2008, Int. J. Approx. Reason..

[64]  G. G. Stokes.  "J." , 1890, The New Yale Book of Quotations.