Institution profile

Bielefeld University

Academic institutioneurope · de
Official website
Research library233linked papers
Opportunities0open roles
Selected work

Representative Papers

PolySHAP: Extending KernelSHAP with Interaction-Informed Polynomial Regression

Jan 26, 2026

This work addresses the computational intractability of exact Shapley value estimation, which requires exponentially many model evaluations and thus does not scale to high-dimensional feature spaces. To overcome this limitation, the authors propose PolySHAP, a novel approximation method that replaces the linear assumption in KernelSHAP with higher-order polynomial regression to better capture nonlinear feature interactions. Theoretical analysis reveals that second-order PolySHAP is equivalent to pairwise sampling, thereby providing rigorous justification for this previously heuristic approach. Extensive experiments across multiple benchmark datasets demonstrate that PolySHAP achieves significantly improved accuracy and consistency in Shapley value estimation compared to existing methods.

1 citations1 influentialRead paper

Influence of prior and task generated emotions on XAI explanation retention and understanding

May 15, 2025

This study investigates how prior affective states (happiness/fear) and task-elicited emotional responses influence users’ comprehension and retention of feature importance explanations in eXplainable AI (XAI). Method: Employing an affect induction paradigm, we conducted multimodal assessment—including heart rate variability (HRV), Facial Action Coding System (FACS) analysis, subjective self-reports, and risk propensity questionnaires—to quantify affective states. Contribution/Results: (1) Prior affect does not impair memory retention of explanations but induces confirmation bias in feature comprehension; (2) Task-elicited, attitude-congruent physiological arousal—specifically triggered by salient features—significantly disrupts comprehension without affecting overall memory; (3) We provide the first empirical evidence that affect differentially modulates the cognitive processing pathways underlying XAI explanation *comprehension* versus *memory*. These findings establish critical cognitive mechanisms for designing affect-aware XAI systems.

1 citationsRead paper

State-space models can learn in-context by gradient descent

Oct 15, 2024arXiv.org

Existing work lacks a theoretical explanation for how structured state space models (SSMs) support in-context learning (ICL) via gradient descent. Method: The authors explicitly construct a single-layer, gated SSM architecture with multiplicative input/output gating, capable of exactly simulating implicit linear and nonlinear model behavior under one- to multi-step gradient updates. Contribution/Results: This construction establishes a formal theoretical connection between SSMs and linear self-attention, identifying multiplicative gating as a critical inductive bias enabling large-model-like expressivity in recurrent architectures. Empirical validation confirms that randomly initialized models, after training, yield parameters closely matching analytical solutions; moreover, the proposed model successfully reproduces ICL capabilities on both linear and nonlinear regression tasks—demonstrating that gradient-based adaptation emerges intrinsically from the SSM’s structure and gating mechanism.

1 citationsRead paper
Recent publications

Latest Papers