Pith. sign in

REVIEW 3 major objections 5 minor 32 references

From Model-Based and Adaptive Control to Evolving Fuzzy Control

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Evolving fuzzy control adapts both the structure and parameters of its model online, closing a gap that classical adaptive control leaves open.

desk verdict A competent short review whose 'fill the gap' conclusion overreaches; useful as an orientation piece, not as evidence. read the letter →

arxiv 2506.06594 v1 pith:PQRZOOGU submitted 2025-06-07 eess.SY cs.AIcs.LGcs.SY

Daniel Leite , Igor Škrjanc , Fernando Gomide This is my paper
classification eess.SYcs.AIcs.LGcs.SY
keywords evolvingfuzzysystemsadaptivecontrolstructuraladaptationnonstationaryenvironmentsrule-basedlearningdatastreams
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This position paper, written for the sixtieth anniversary of fuzzy set theory, traces the arc from model-based control through adaptive control to evolving fuzzy control and argues that the last step is the one that closes a real gap. Classical adaptive control--whether direct or indirect, in the model-reference or self-tuning tradition--adjusts parameters online but keeps the structure of the plant model and controller fixed. Evolving fuzzy systems instead grow, shrink, and rewrite their rule bases as data arrive, so model structure and parameters adapt together, often from an empty starting point and without a prior plant model. The paper maintains that this combined structural and parametric adaptation is what makes them suited to nonstationary and poorly modeled processes, and it frames safety, interpretability, and controlled evolution as the open problems that stand between the idea and dependable use.

What carries the argument

The load-bearing object is the evolving fuzzy system: an incremental rule-based model whose rule base--the collection of fuzzy rules and their membership functions--is created and modified online as each data point arrives, rather than fixed during an offline design. Within each rule, recursive estimation updates local parameters, so structure and parameters change simultaneously. In the control setting this mechanism is coupled with parallel distributed compensation, a scheme in which each local rule of the evolving plant model comes with a matching local controller rule; the cited stability analysis treats the closed loop with Lyapunov arguments and bounded inputs. This combination is what lets the system start from scratch, add or remove rules as operating conditions change, and still carry formal guarantees.

What would settle it

Run a benchmark where a plant switches abruptly between two operating regimes; if a fixed-structure adaptive controller (self-tuning regulator or model reference) matches or beats an evolving fuzzy controller that starts with no rules on both transient and steady-state error, the claimed gap closes. Conversely, showing that any existing adaptive scheme with online switching of model structure already achieves structural adaptation would undercut the paper's dichotomy.

Watch

Extended reading notes

Core claim

The paper's central claim is that simultaneous structural and parametric adaptation is the missing capability that separates modern control practice from nonstationary reality. It characterizes classical adaptive schemes as online parameter estimators wrapped around a controller of fixed structure: model-reference adaptive control tunes gains to follow a reference model, and self-tuning regulators re-estimate plant parameters and recompute a controller, but neither changes the form of the model or the rule set. Evolving fuzzy systems are presented as the natural extension, with an incremental rule base in which rules are created, updated, and merged from streaming data; a concrete instance described in the paper couples an evolving functional-rule model of the plant with an evolving fuzzy controller under a parallel distributed compensation scheme, with bounded control inputs and Lyapunov-based stability arguments. The authors conclude that this closes the gap left by adaptive control and makes evolving fuzzy control a promising path for complex, time-varying processes, provided safety and interpretability mechanisms are added.

Load-bearing premise

The conclusion holds only if the cited prior work--some of it by the same research groups--genuinely demonstrates stable, online structural evolution for fuzzy control, and if adaptive control is fairly represented as fixed-structure parameter tuning rather than as a family that includes switching and multiple-model schemes.

Editorial extensions

If this is right

  • Controllers built this way can operate entirely online from an empty rule base, removing the offline identification and training phase that is infeasible for unstable or poorly understood plants.
  • Because rules and parameters co-adapt, evolving fuzzy control can track changes that shift the plant into entirely new operating regimes, not just slow coefficient drift.
  • The stability analysis for the evolving fuzzy model-based controller shows that structural adaptation and mathematical guarantees need not be opposed, encouraging further formal work.
  • Practical deployment will need fallback controllers, human oversight, bounded actuation, or restricted exploration during early learning, especially in safety-critical plants.
  • The paper identifies shadow-mode validation and human review of added rules as required mechanisms before evolving control is adopted in practice.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The clean split between fixed-structure adaptive control and structurally evolving fuzzy systems ignores switching and multiple-model adaptive schemes, which already change the active model structure online; testing the paper's thesis against those schemes would show whether the claimed gap is a real one.
  • A direct benchmark that would test the thesis: on a plant with abrupt regime changes, compare a self-tuning or model-reference controller against an evolving fuzzy controller starting from scratch, measuring transient error, settling time, and rule-count growth; the paper does not report such comparisons.
  • The same structure-plus-parameter update loop transfers to forecasting, clustering, and classification of data streams, so the paper's argument implies the broader claim that nonstationary learning problems generally benefit from structural evolution.
  • One testable design principle suggested by the review is to gate structural changes through a risk metric computed in shadow mode, so a new rule is only activated when it reduces predicted error or preserves stability; this could be implemented and evaluated directly.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. This brief position/review paper, written for the 60th anniversary of fuzzy set theory, traces the historical development of fuzzy and adaptive control and argues that evolving fuzzy systems—which incrementally update both the structure and parameters of rule-based models from data streams—fill a gap left by conventional adaptive control (which tunes parameters under fixed structures). The paper reviews classical adaptive control schemes (MRAC, STR), introduces evolving fuzzy modeling and control, and closes by listing open challenges such as safety, interpretability, and principled structural evolution, with the conclusion that evolving fuzzy systems 'fill this gap' between adaptive control and nonstationary environments.

Significance. The paper's contribution is conceptual: it offers a compact, readable historical framing and an explicit research agenda. The strongest feature is the candid discussion in Section II.C of the safety risks of closing the loop with an evolving controller and the proposal of fallback control, human-in-the-loop supervision, and shadow-mode validation. These are valuable points for the community. However, the paper is not a technical contribution and provides no independent verification, systematic comparison, or error analysis. Its central claim that evolving fuzzy systems uniquely fill a structural-adaptation gap is carried primarily by self-cited prior work ([6], [7]) and by a sharp dichotomy between parameter adaptation and structural evolution that omits relevant adaptive control methods. If the paper is to be accepted as a survey/position piece, the supporting evidence and the framing of the gap need to be strengthened.

major comments (3)
  1. [Section II.A and Conclusion] The claim that conventional adaptive control operates under fixed structures is overstated. Supervisory adaptive control, multiple-model adaptive control, gain scheduling, and switching among local controllers are established techniques that alter the controller's structure (or the local model set) online in response to plant changes; they are not merely coefficient tuning. The statement in Section II.A that 'adaptive systems allow real-time adjustment of model coefficients and controller gains but still assume fixed structures' ignores these schemes. Since the conclusion's 'fill this gap' depends on this dichotomy, the paper should either discuss these alternative structural-adaptation mechanisms explicitly and justify why they fail to meet the same need, or temper the conclusion.
  2. [Section II.B] The main concrete evidence that evolving fuzzy control delivers structural adaptation with stability guarantees is reference [6], a paper sharing co-authors with this submission; reference [7] is also self-authored. Section II.B states that a practical example in [6] 'formally ensured' Lyapunov stability and bounded control inputs, but no independent confirmation or external follow-up is cited. Given the paper's role as a survey/position statement, this reliance on self-citations is insufficient to establish a general result. The authors should broaden the evidence base with independent studies or clearly flag the example as an illustration rather than a general guarantee.
  3. [Section II.C and Conclusion] The paper's own Section II.C concedes that 'closing the loop with a controller that evolves from scratch is risky' and that structural changes 'should be validated in shadow or standby mode.' This concession is in tension with the conclusion that evolving fuzzy systems 'fill this gap by supporting both parametric and structural adaptation... often without requiring prior process knowledge.' The conclusion should be qualified to reflect that structural evolution is a promising but not-yet-safe capability, and the proposed safety strategies (fallback control, conservative initialization, bounded output policies) should be tied to concrete mechanisms or references.
minor comments (5)
  1. [Acknowledgment] The acknowledgment contains a misspelling: 'North Rhyne-Westphalia' should be 'North Rhine-Westphalia.'
  2. [Section II.A] The phrase 'on the basis the observed system behavior' should read 'on the basis of the observed system behavior.'
  3. [Figures] The paper refers to Fig. 1 and Fig. 2, but the figures are not included in the arXiv text version; ensure they are present in the final camera-ready version.
  4. [Reference [21]] Reference [21] is listed as an early indirect adaptive control work, but the connection to 'digital process control applications' is not explained in the text; consider a brief clarifying phrase.
  5. [Title and structure] The title suggests a progression 'from model-based and adaptive control to evolving fuzzy control,' but the body is a chronological and conceptual survey rather than a demonstration of that trajectory; the rhetorical framing could be aligned more explicitly with the content.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: this is a position/review paper whose claims are supported by external prior work, not by derivations that reduce to their own inputs.

full rationale

The paper is an invited-style position/review, not a derivation. It does not fit parameters, derive equations, or present a formal result; therefore the classic circularity patterns (self-definitional equivalence, fitted inputs called predictions, uniqueness theorems imported from self-citations, ansatz smuggling) do not apply. The central claim that evolving fuzzy systems combine structural and parametric adaptation is definitionally connected to the term 'evolving fuzzy systems,' but the paper does not pretend to derive this from first principles; it presents it as a characterization of an established research area and supports it with both independent foundational citations (e.g., [8]–[10], [27]) and self-authored survey/example citations ([6], [7]). Self-citation is present, notably for the concrete stability-oriented example in Section II.B citing [6], but a review's citation of prior archival work by overlapping authors is not circular: the present paper does not reduce its argument to an unverified self-citation, and the prior work is external to this manuscript. The paper also contains an explicit limitation statement in Section II.C conceding that closing the loop with an evolving controller is risky and that structural changes should be validated in shadow mode, which weighs against the strength of its own conclusion but is not a circularity defect. The absence of a discussion of switching or multiple-model adaptive control is a completeness/support concern, not a circularity concern. Accordingly, no circular step can be exhibited, and the score is 0.

Assumptions & free parameters 0 free parameters · 2 assumptions · 0 invented entities

The paper is a review, so the ledger contains no free parameters or invented entities. The only underlying assumptions are the validity of fuzzy set theory as a modeling framework and the correctness of the cited state-of-the-art results, several of which are authored by the present authors.

assumptions (2)
  • domain assumption The reported results in the cited evolving fuzzy systems literature (especially [6] and [7]) are correct and representative.
    The review's positive characterization of evolving fuzzy systems in Section II.B is built on these citations, several of which are authored by the present authors.
  • domain assumption Fuzzy set theory and rule-based interpolation provide a viable basis for nonlinear identification and control.
    This is the foundational premise of the entire review, introduced in Section I.

how reviews work

0 comments
Cite this review

Pith. "Pith review of From Model-Based and Adaptive Control to Evolving Fuzzy Control." pith.science (2026). https://pith.science/paper/PQRZOOGU

@misc{pith2026250606594,
  author       = {Pith},
  title        = {Pith review of: From Model-Based and Adaptive Control to Evolving Fuzzy Control},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PQRZOOGU}},
  note         = {Machine review of arXiv:2506.06594}
}
read the original abstract

Evolving fuzzy systems build and adapt fuzzy models - such as predictors and controllers - by incrementally updating their rule-base structure from data streams. On the occasion of the 60-year anniversary of fuzzy set theory, commemorated during the Fuzz-IEEE 2025 event, this brief paper revisits the historical development and core contributions of classical fuzzy and adaptive modeling and control frameworks. It then highlights the emergence and significance of evolving intelligent systems in fuzzy modeling and control, emphasizing their advantages in handling nonstationary environments. Key challenges and future directions are discussed, including safety, interpretability, and principled structural evolution.

Figures

Figures reproduced from arXiv: 2506.06594 by the authors.

Figure 1
Figure 1. Indirect adaptive control scheme The use of adaptive control is based on the assumption that, for any possible values of the coefficients of the plant model, there exists a controller—with fixed structure and com￾plexity—capable of meeting the design specifications through appropriate tuning of its parameters [24]. In this context, the task of adaptation is to determine suitable values for the controller parameters.… view at source ↗
Figure 2
Figure 2. Evolving system—plant model or controller: Online adaptation of [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

32 extracted references · 32 canonical work pages

  1. [25]

    A Historical Perspective of Adaptive Control and Learning

    A. Annaswamy, A. Fradkov, A historical perspective of adaptive control and learning. https://doi.org/10.48550/arXiv.2108.11336, 2022

    work page Pith review arXiv doi:10.48550/arxiv.2108.11336 2022

  2. [6]

    Leite, R

    D. Leite, R. Palhares, V . Campos, F. Gomide, Evolving granular fuzzy model-based control of nonlinear dynamic systems. IEEE Transactions on Fuzzy Systems, 23(4), 923-938, 2015

    work page 2015

  3. [7]

    ˇSkrjanc, J

    I. ˇSkrjanc, J. A. Iglesias, A. Sanchis, D. Leite, E. Lughofer, F. Gomide, Evolving fuzzy and neuro-fuzzy approaches in clustering, regression, identification, and classification: A survey. Information Sciences, 490, 344-368, 2019

    work page 2019

  4. [1]

    L. A. Zadeh. Fuzzy sets. Information and Control, 8(3), 338–353, 1965

    work page 1965

  5. [2]

    Belohlavek, J

    R. Belohlavek, J. Dauben, J. Klir, Fuzzy Logic and Mathematics: A Historical Perspective. New York, NY: Oxford, 2017

    work page 2017

  6. [3]

    Kosko, Fuzzy Logic: A Dynamical System Approach to Machine Intelligence, Englewood Cliffs, NJ: Prentice-Hall, 1992

    B. Kosko, Fuzzy Logic: A Dynamical System Approach to Machine Intelligence, Englewood Cliffs, NJ: Prentice-Hall, 1992

    work page 1992

  7. [4]

    Ying, Fuzzy Control and Modeling: Analytical Foundations and Applications

    H. Ying, Fuzzy Control and Modeling: Analytical Foundations and Applications. New York, NY: IEEE Press, 2000

    work page 2000

  8. [5]

    Tanaka, H

    K. Tanaka, H. O. Wang, Fuzzy Control Systems Design and Analysis: A Linear Matrix Inequality Approach. J. Wiley & Sons, Hoboken, 2004

    work page 2004

Show all 32 references
  1. [8]

    Kasabov, Q

    N. Kasabov, Q. Song, DENFIS: Dynamic evolving neural-fuzzy in- ference system and its application for time-series prediction. IEEE Transactions on Fuzzy Systems, 10, 144–154, 2002

    2002

  2. [9]

    Angelov, Evolving Rule-based Models: A Tool for Design of Flexible Adaptive Systems

    P. Angelov, Evolving Rule-based Models: A Tool for Design of Flexible Adaptive Systems. Heidelberg: Springer, 2002

    2002

  3. [10]

    Angelov, D

    P. Angelov, D. Filev, An approach to online identification of Tak- agi–Sugeno fuzzy models. IEEE Transactions on Systems, Man, and Cybernetics B, 34, 484–498, 2004

    2004

  4. [11]

    Sayed-Mouchaweh, E

    M. Sayed-Mouchaweh, E. Lughofer (Eds.), Learning in Non-Stationary Environments: Methods and Applications. Springer, New York, 2012

    2012

  5. [12]

    Khalil, Nonlinear Systems

    H. Khalil, Nonlinear Systems. Upper Saddle River, NJ: Prentice Hall, 2002

    2002

  6. [13]

    D. A. Copp, J. P. Hespanha, Simultaneous nonlinear model predictive control and state estimation. Automatica, 77, 143–154, 2017

    2017

  7. [14]

    Bernal, A

    M. Bernal, A. Sala, Z. Lendek, T. Guerra, Analysis and Synthesis of Nonlinear Control Systems. Cham, CH: Springer, 2022

    2022

  8. [15]

    Aseltine, R

    A. Aseltine, R. Mancini, C. Saturne, A survey of adaptive control systems. IRE Transactions on Automatic Control, 6, 102–108, 1958

    1958

  9. [16]

    Feldbaum, Dual control theory

    A. Feldbaum, Dual control theory. Automation and Remote Control, 21, 1240–1249, 1960

    1960

  10. [17]

    Mishkin, L

    E. Mishkin, L. Brown, Adaptive Control Systems. New York, NY: McGraw-Hill, 1961

    1961

  11. [18]

    Truxal, Adaptive control

    J. Truxal, Adaptive control. Proceedings of IFAC, 1, 386–392, 1963

    1963

  12. [19]

    Ioannou, B

    P. Ioannou, B. Fidan, Adaptive Control Tutorial. Philadelphia, PA: SIAM, 2006

    2006

  13. [20]

    Whitaker, J

    H. Whitaker, J. Yamron, A. Kezer, Design of a model-reference adaptive control system for aircraft. Tech. Report R-164, MIT Instrumentation Laboratory, Cambridge, MA, 1958

    1958

  14. [21]

    Kalman, Design of self-optimizing control systems

    R. Kalman, Design of self-optimizing control systems. Transactions of ASME Journal of Basic Engineering, 80, 468–478, 1958

    1958

  15. [22]

    Bla ˇziˇc, I

    S. Bla ˇziˇc, I. ˇSkrjanc, D. Matko, Globally stable direct fuzzy model reference adaptive control. Fuzzy Sets and Systems, 139(1), 3–33, 2003

    2003

  16. [23]

    ˇSkrjanc, S

    I. ˇSkrjanc, S. Bla ˇziˇc, D. Matko, Model-reference fuzzy adaptive control as a framework for nonlinear system control. Journal of Intelligent & Robotic Systems, 36(3), 331–347, 2003

    2003

  17. [24]

    Landau, R

    I. Landau, R. Lozano, M. Saad, A. Karimi, Adaptive Control: Algo- rithms, Analysis and Applications. New York, NY: Springer, 2011

    2011

  18. [26]

    Takagi, M

    T. Takagi, M. Sugeno, Fuzzy identification of systems and its applica- tions to modeling and control. IEEE Transactions on Systems, Man, and Cybernetics, 15(1), 116–132, 1985

    1985

  19. [27]

    Angelov, N

    P. Angelov, N. Kasabov, D. Filev, Guest editorial evolving fuzzy systems: Preface to the special section. IEEE Transactions on Fuzzy Systems, 16, 1391–1392, 2008

    2008

  20. [28]

    Fritzke, Growing cell structures—A self-organizing network for unsupervised and supervised learning

    B. Fritzke, Growing cell structures—A self-organizing network for unsupervised and supervised learning. Neural Netw, 7, 1441–1460, 1994

    1994

  21. [29]

    Kordon, Inferential sensors as potential application area of intelligent evolving systems

    A. Kordon, Inferential sensors as potential application area of intelligent evolving systems. Proceedings of IEEE International Symposium on Evolving Fuzzy Systems, UK, 2006

    2006

  22. [30]

    Zadeh, On the definition of adaptivity

    L. Zadeh, On the definition of adaptivity. Proceedings of the IEEE, 51, 469–470, 1963

    1963

  23. [31]

    J. Hill, G. McMurthy, K. Fu, A computer-simulated on-line experiment in learning control systems. Simulation, 4, 104–116, 1965

    1965

  24. [32]

    Saridis, Self-organizing control and applications to trainable manip- ulators and learning prostheses

    G. Saridis, Self-organizing control and applications to trainable manip- ulators and learning prostheses. IFAC Proceedings, 8, 632–640, 1975

    1975

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.