Musashino University (武蔵野大学, Musashino Daigaku) is an institution of higher education in Ariake, a district in Kōtō, Tokyo, with a suburban campus in Nishitōkyō. Musashino University is uniquely focused on the ideals associated with the Hongwanji Jodo Shinshu School of Buddhism.
Nonlocal MEMS equations exhibit finite-time quenching phenomena that pose significant challenges for numerical simulation. In this paper, we study a stagewise rescaling algorithm for a two-dimensional nonlocal MEMS equation in an asymptotically constant-feedback touchdown regime. The original nonlocal equation is not exactly invariant under the A3/2–A3 scaling used below; rather, the scaling is appropriate when the reciprocal-integral feedbackK(t)=1+∫Ω(1−u(x,t))−1dxremains bounded and converges to a finite positive limit, as in the single-point touchdown profiles constructed by Duong–Zaag. In this regime the leading-order core balance is that of a local MEMS equation with an asymptotically constant coefficient. By applying a fixed-stage scaling to the deficit variable, we transform the equation into a gradient flow for a rescaled energy at a frozen amplitude and obtain an exact energy dissipation identity within each stage. We then introduce a minimizing-movement stage solver and obtain a discrete fixed-stage energy inequality.At a full-domain stage transition, the continuous rescaling is exactly energy preserving. For the discrete 12-point transfer, we derive a one-sided refinement estimate. Under a stage-uniform, grid-aligned piecewise-regularity budget for the positive interpolation polynomial, including the normal-derivative jumps across coarse-cell interfaces and the physical-boundary traces, the positive switch defect satisfiesεm,hsw≤CAm2h2.For exact samples of a C4 profile, the interface normal-derivative jump is O(h3) and the centered-Laplacian error at interface-crossing fine-grid stencils is O(h2). A separate outer-update defect retains the finite-window truncation error, yielding an exact defect balance and a quantitative almost-monotonicity estimate. We also formulate a defect-aware criterion for the nonexistence of a global admissible continuation. Finally, reproducible two-dimensional computations report complete energy histories, a manufactured smooth-profile interpolation test with observed orders four, three, and two, and the signed switch defects in the reported full-domain run. These finite tests do not by themselves verify the required stage-uniform regularity budget on an expanding sequence of rescaled domains.
Fuzzy disjunctive normal form (FDNF) gives the canonical multi-affine extension of an elementary cellular automaton (ECA) rule to the unit cube. Although it preserves the Boolean rule on binary states, its multi-affine structure can smooth high-contrast CA patterns and restrict continuous-state dynamics. We introduce generalized FDNF rules f_k^g,u,v,w(x,y,z) = g(f_k(u(x),v(y),w(z))), where the transformations g,u,v,w: [0,1] → [0,1] fix the endpoints. The identity maps recover ordinary FDNF, while threshold-like, discontinuous, non-monotone, and expanding choices yield rule-preserving fuzzy ECAs. We demonstrate, in representative rules, that the transformation shape strongly affects pattern dynamics: threshold-like maps promote ECA-like pattern recovery, parameter deformations interpolate toward FDNF-like smoothing, and discontinuities induce gap-generated regimes. Pattern changes are summarized by contrast, fuzziness, and a finite-resolution participation-type support exponent. In three-cell systems, an expanding non-monotone transformation yields stable period-six cycles for rule 210, verified by interval arithmetic, coexisting with an expanding invariant line set; rules 51 and 85 inherit one-dimensional expanding dynamics. The framework provides a rule-preserving bridge from Boolean cellular automata to fuzzy and continuous-state nonlinear dynamics.
SHapley Additive Explanations (SHAP) are widely used to interpret machine learning models in agriculture and environmental decision-making, but SHAP inherits model misspecification, confounding, and distribution shift, risking unstable and policy-misleading importance rankings. This opinion advances a concrete, model-agnostic workflow that goes beyond SHAP: pair unsupervised structure checks (e.g., feature agglomeration, highly variable feature selection) with nonlinear, nonparametric association metrics and pre-registered sensitivity/stability analyses, then audit attributions with domain-informed negative controls and decision impact tests. The novelty lies in reframing SHAP from a standalone explainer to one component of a robustness protocol explicitly designed to reduce label-driven and proxy bias. This integrated approach yields more reliable variable importance, tighter uncertainty communication, and clearer links between attribution changes and real policy choices.
Marine Pollution Bulletin increasingly applies machine learning and explainable AI to pollutant and shellfish poisoning risk, exemplified by PCA-based source apportionment and SHAP-based feature attribution. However, linear PCA may misrepresent structure in inherently nonlinear environmental data, and existing studies often treat model-derived feature importances as evidence of true associations without assessing consistency or dose-response relationships. This paper clarifies that supervised models possess two distinct accuracies: prediction and feature importance, and only prediction can be validated against ground truth. Using a Basque coastal dataset (8195 instances, 14 features) with chlorophyll-a as a proxy for paralytic shellfish poisoning risk, we introduce a leave-top1-out procedure to test ranking stability. Random Forest and XGBoost with and without SHAP show pronounced instability, indicating biased, model-dependent importances. In contrast, unsupervised and non-target-prediction methods yield perfectly stable rankings while matching or exceeding supervised performance, supporting routine stability, consistency, dose-response, and linearity checks in environmental ML studies.
This study critically evaluates the methodological approach employed by Nayak et al. in COVID-19 CT scan classification using their Dense-Res-Inception Ensemble Network with PCA dimensionality reduction. We demonstrate fundamental limitations in applying PCA to nonlinear biological imaging data, where key mathematical assumptions-including linearity, meaningful correlations, and homoscedasticity-are frequently violated. Using the MNIST dataset as a benchmark, we comparatively assessed three dimensionality reduction techniques: PCA, High Variance Gene Selection, and Feature Agglomeration (FA). Results confirm that FA significantly outperforms PCA (92.79% vs 83.76% accuracy) by preserving crucial spatial relationships within image data. This performance disparity highlights the critical importance of methodological alignment with data characteristics in medical imaging analysis. We propose that nonlinear dimensionality reduction approaches better accommodate the complex relationships inherent in biological systems, potentially enhancing both computational efficiency and diagnostic reliability in clinical applications requiring rapid assessment.