Non-monotonic causal discovery with Kolmogorov-Arnold Fuzzy Cognitive Maps

Jose L. Salmeron

cs.AIApr 6, 2026

Non-monotonic causal discovery with Kolmogorov-Arnold Fuzzy Cognitive Maps

Authors: Jose L. Salmeron

Organizations: School of Engineering, CUNEF Universidad, Madrid, Spain

Abstract

Fuzzy Cognitive Maps constitute a neuro-symbolic paradigm for modeling complex dynamic systems, widely adopted for their inherent interpretability and recurrent inference capabilities. However, the standard FCM formulation, characterized by scalar synaptic weights and monotonic activation functions, is fundamentally constrained in modeling non-monotonic causal dependencies, thereby limiting its efficacy in systems governed by saturation effects or periodic dynamics. To overcome this topological restriction, this research proposes the Kolmogorov-Arnold Fuzzy Cognitive Map (KA-FCM), a novel architecture that redefines the causal transmission mechanism. Drawing upon the Kolmogorov-Arnold representation theorem, static scalar weights are replaced with learnable, univariate B-spline functions located on the model edges. This fundamental modification shifts the non-linearity from the nodes' aggregation phase directly to the causal influence phase. This modification allows for the modeling of arbitrary, non-monotonic causal relationships without increasing the graph density or introducing hidden layers. The proposed architecture is validated against both baselines (standard FCM trained with Particle Swarm Optimization) and universal black-box approximators (Multi-Layer Perceptron) across three distinct domains: non-monotonic inference (Yerkes-Dodson law), symbolic regression, and chaotic time-series forecasting. Experimental results demonstrate that KA-FCMs significantly outperform conventional architectures and achieve competitive accuracy relative to MLPs, while preserving graph- based interpretability and enabling the explicit extraction of mathematical laws from the learned edges.

Figures & tables

Panel of Figure 1 — Fig. 1: Causal relationship comparison

Fig. 2: Kolmogorov-Arnold Fuzzy Cognitive Map architecture.

Fig. 3: Comparative modeling of the non-monotonic Yerkes-Dodson law. The scatter plot represents the synthetic ground truth exhibiting an inverted U-shape. The dashed blue line indicates the standard FCM prediction. The green dashed line shows the MLP results. The solid red line shows the function learned by the KA-FCM edge.

Model	MSE	Max. Abs. Error	Std. Dev. Error
Standard FCM	$3.96\times 10^{-1}$	$0.971$	$0.556$
MLP	$1.18\times 10^{-4}$	$0.025$	$0.009$
KA-FCM	$\mathbf{4.10\times 10^{-5}}$	$\mathbf{0.015}$	$\mathbf{0.006}$

TABLE I: Performance and stability analysis (Experiment I)

Parameter	Optimal value	Correlation w/ Error
Grid size ( $G$ )	$4$	$+0.05$ (minimal)
Learning rate ( $\eta$ )	$0.1$	$\mathbf{-0.61}$ (critical)
Epochs	$611$	$-0.16$ (low)

TABLE II: Hyperparameter sensitivity (Experiment I)

Model	MSE	Max. Abs. Error	Std. Dev. Error
Standard FCM	$4.26\times 10^{-1}$	$1.119$	$0.654$
MLP (Baseline)	$4.37\times 10^{-4}$	$0.082$	$0.011$
KA-FCM	$\mathbf{3.40\times 10^{-8}}$	$\mathbf{0.0004}$	$\mathbf{0.0002}$

TABLE III: Performance and stability analysis (Experiment II)

Fig. 4: Learned edge functions comparison.

Parameter	Optimal value	Correlation w/ Error
Grid size ( $G$ )	$19$	$\mathbf{-0.32}$ (significant)
Learning rate ( $\eta$ )	$0.1$	$\mathbf{-0.40}$ (high)
Epochs	$1500$	$-0.22$ (moderate)

TABLE IV: Hyperparameter sensitivity (Experiment II)

Model	MAPE (%)	Max. Abs. Error	Std. Dev. Error
Standard FCM	$41.37\%$	$0.162$	$0.058$
MLP	$27.67\%$	$0.097$	$0.027$
KA-FCM	$\mathbf{13.66\%}$	$\mathbf{0.072}$	$\mathbf{0.023}$

TABLE V: Performance and stability analysis (Experiment III - Mackey-Glass)

Fig. 5: One-step-ahead forecasting performance on the chaotic Mackey-Glass time series ( $\tau=17$ ).

Parameter	Optimal value	Correlation w/ Error (MAPE)
Grid size ( $G$ )	$19$	$-0.07$ (slight)
Learning rate ( $\eta$ )	$0.05$	$\mathbf{-0.62}$ (critical)
Epochs	$1277$	$-0.28$ (moderate)

TABLE VI: Hyperparameter sensitivity (Experiment III)

Explore similar work

Jul 13, 2026

DAG-FM: A Foundation Model for Causal Discovery under Heterogeneous Causal Mechanisms

Causal discovery from observational tabular data remains fundamentally challenging, primarily due to the heterogeneity of underlying causal mechanisms and the high-dimensional combinatorial search space of Directed Acyclic Graphs (DAGs). In this paper, we propose DAG-FM, a novel foundation model architecture that amortizes causal discovery. Unlike direct matrix prediction, DAG-FM decomposes the causal discovery process into two auto-regressive stages using two specialized Transformer-based sub-modules: a leaf-node predictor and a parent-node predictor. To effectively model complex row-column interactions, we adopt a robust tabular interaction block to output feature-wise representations. Crucially, to handle diverse and unknown Functional Causal Model (FCM) assumptions in real-world scenarios, we introduce Mixture-of-Leaf-Experts (MoLE), allowing the model to dynamically route and adapt to identifiable mechanism families. Through an iterative inference algorithm, DAG-FM seamlessly extracts causal orderings and constructs valid DAGs. Extensive experiments demonstrate that DAG-FM achieves state-of-the-art performance on both synthetic benchmarks and complex real-world datasets, significantly outperforming traditional classical algorithms and recent foundation models in both accuracy and scalability.

Yikang Chen, Zhengkang Guan, Haoyuan Qian +3

1Zhejiang University · 2Tsinghua University

Causal Foundation Models Tabular Foundation Models Structural Causal Models Causal Discovery

Ask PDF

Jun 11, 2026

CausalMoE: A Billion-Scale Multimodal Foundation Model for Granger Causal Discovery with Pattern-Routed Heterogeneous Experts

Granger Causal Discovery (GCD) is fundamental for analyzing temporal dependencies in complex systems. However, existing neural GCD methods predominantly rely on a "one-size-fits-all" paradigm, struggling to capture distribution shifts and dynamic regime changes inherent in real-world time series. This often leads to entangled representations and spurious causal graphs. In this paper, we propose CausalMoE, a billion-scale multimodal Granger causal foundation model that explicitly models patch-level heterogeneity. CausalMoE introduces a Pattern-Routed Mixture of Heterogeneous Experts, which dynamically identifies latent temporal patterns and routes patches to specialized domain experts, effectively decoupling regime-specific mechanisms from shared dynamics. To ensure interpretable graph recovery, we design a Causality-Aware Self-Attention mechanism operating across variables, yielding sparse Granger causal graphs via proximal optimization. Furthermore, CausalMoE is the first to integrate LLMs and VLMs to align numerical signals with textual and visual priors, regularizing causal estimation in complex scenarios. Extensive experiments demonstrate that CausalMoE establishes a new state-of-the-art on fully supervised benchmarks, while effectively generalizing to few-shot settings where traditional methods fail.

Bo Liu, Di Dai, Jingwei Liu +5

State Key Laboratory of General Artificial Intelligence, School of Intelligence Science and Technology, Peking University Beijing, China · National Institute of Health Data Science, and Institute for Artificial Intelligence, Peking University Beijing, China

Time-Series Causal Discovery Mixture-of-Experts Models Time-Series Foundation Models

Ask PDF

May 6, 2026

Counterfactual identifiability beyond global monotonicity: non-monotone triangular structural causal models

Structural causal models provide a unified semantics for interventions and counterfactuals, but most identifiability results rely on restrictive assumptions like global monotonicity, which are often violated in embodied interaction, where the same exogenous perturbation can induce opposite responses under different contact contexts. We ask what structure still suffices once global monotonicity is dropped. We introduce non-monotone triangular structural causal models (NM-TM-SCM), which retain triangular recursion but replace global monotonicity with mechanism-wise invertibility and context-independent inverse transport. We prove that these conditions are equivalent to exogenous isomorphism and imply complete counterfactual identifiability, and we give a counterexample showing that local invertibility alone is insufficient. We instantiate the theory in CausalInverter, with triangular invertible layers, orientation gates, and transport-stability regularization. On synthetic non-monotonic mechanisms, the structural bias yields systematic counterfactual gains as non-monotonicity increases. On MuJoCo Door, our model achieves perfect event-level counterfactual recovery, lowers continuous angle error relative to a Transformer baseline, and delivers substantially more stable recovery than Transformer and conditional-flow predictors. On MuJoCo Push, where non-monotonicity is weaker, the same low-data predictors remain competitive or better, consistent with a bias-variance boundary. These results identify a broader identifiable regime between globally monotone triangular models and unconstrained black-box world models.

Pengcheng Tan, Jiang Chen, Dehui Du

Shanghai Key Laboratory of Trustworthy Computing, East China Normal University

Structural Causal Models Causal Counterfactual Generation Counterfactual Prediction Counterfactual Learning Embodied World Models

Ask PDF