Institution profile

Université catholique de Louvain

Academic institutioneurope · be
Official website
Research library180linked papers
Opportunities0open roles
Selected work

Representative Papers

Network Simulator-Centric Compositional Testing

Mar 04, 2025Formal Techniques for (Networked and) Distributed Systems

Addressing the challenges of protocol verification under time-varying network properties—namely, non-reproducible testing and difficult fault attribution—this paper proposes a modular, compositional testing paradigm centered on the programmable network emulator CORE/INET. We introduce the first simulator-native compositional testing architecture, which decouples protocol components from test logic, enabling decomposable, incremental validation of cross-layer interactions and precise fault localization. By integrating formal specifications (TLA+/SysML) with differential fuzz testing, our approach uncovers 17 previously undisclosed timing-sensitive defects across the TCP/QUIC/HTTP/3 protocol stack. Evaluation demonstrates a 3.8× improvement in verification efficiency and reduces false positive rates to 4.2%.

2 citationsRead paper

A universal vision transformer for fast calorimeter simulations

Jan 07, 2026

This work addresses the high computational cost of traditional simulation tools like Geant4 in modeling particle showers within complex, high-dimensional calorimeter geometries. The authors propose a general-purpose generative model based on the Vision Transformer (ViT), which is, to the best of their knowledge, the first successful application of ViT to fast simulation of both electromagnetic and hadronic showers, accommodating both regular and irregular detector geometries. Leveraging large-scale pretraining followed by fine-tuning on target detectors, the model generates full showers in 10–100 milliseconds on a single GPU, achieving orders-of-magnitude speedup. The generated showers exhibit statistical fidelity indistinguishable from Geant4 across multiple metrics while demonstrating strong generalization across diverse geometric configurations.

1 citationsRead paper

Context-aware Adaptive Visualizations for Critical Decision Making

Nov 14, 2025

Existing information visualization dashboards lack real-time awareness of users’ cognitive states and adaptive capabilities, hindering critical decision-making under high cognitive load. This paper introduces Symbiotik—the first adaptive visualization framework integrating neurophysiological signals (EEG) with reinforcement learning to dynamically optimize visual encoding, layout, and interaction logic based on real-time cognitive load decoding. Methodologically, we design a lightweight EEG feature extraction model and a policy-gradient-based interface control algorithm, enabling millisecond-scale closed-loop feedback. In a controlled study with 120 participants, Symbiotik significantly improved task completion efficiency (+28.6%) and subjective immersion (p < 0.01), while demonstrating cross-domain transferability. This work establishes a reusable methodology for real-time adaptive visualization, with direct applicability to domains such as neuromarketing and human-AI collaboration.

1 citationsRead paper

Telling Human and Machine Handwriting Apart

Oct 01, 2025IEEE Transactions on Systems, Man, and Cybernetics: Systems

This work proposes a shallow recurrent neural network–based approach to directly distinguish between human-written and machine-generated handwriting samples—including those produced by GANs, Transformers, and diffusion models—using raw pen-tip trajectories, thereby enabling authentic user verification and defense against automated attacks. It presents the first systematic evaluation of the discriminability of diverse handwriting synthesis methods in human–machine classification, eliminating the need for handcrafted features by integrating kinematic modeling grounded in the Sigma-lognormal framework. The method achieves an average AUC of 98.3% and an equal error rate of only 1.4% across ten public datasets. Remarkably, it maintains high performance on 90% of test sets using merely 10% of the training data and demonstrates strong robustness in cross-domain scenarios.

1 citationsRead paper

The Volterra Stein-Stein model with stochastic interest rates

Mar 03, 2025

This paper addresses the fragmentation between Markovian and non-Markovian frameworks in financial derivative pricing and hedging, along with the scarcity of analytical solutions. We propose a unified framework wherein correlated Gaussian Volterra processes jointly drive volatility and interest rates, yielding the first Volterra extension of the Stein–Stein model incorporating stochastic interest rates. Despite its non-Markovian nature, the model retains strong analytical tractability: it admits closed-form expressions for zero-coupon bonds and interest rate caps/floors, and delivers a semi-explicit characteristic function for the log-forward index—derived via Fredholm resolvents and determinants. Methodologically, we integrate Volterra stochastic calculus, Fredholm theory, and infinite-dimensional Riccati equations to enable efficient pricing and calibration. Empirical analysis demonstrates that the model accurately captures the “humped” term structure of ATM volatility and the concave log-moneyness slope observed in S&P 500 options.

1 citationsRead paper
Recent publications

Latest Papers

Translating finite-domain integer constraint models to CP/SMT/ILP/PB/SAT solvers with CPMpy

Aug 15, 2026

This study addresses the high cost and error-proneness of model refactoring caused by paradigm disparities among constraint solvers. We propose a modular automated translation framework based on CPMpy that employs a layered waterfall architecture to uniformly handle sub-expression negation and auxiliary variable generation while optimizing linearization strategies for nonlinear operators. This approach enables seamless translation from high-level models to low-level paradigms, including CP, SMT, and ILP. Experimental results demonstrate that the framework effectively eliminates manual rewriting and that its optimized linearization significantly enhances ILP and PB solving performance. Consequently, this work provides an efficient, flexible, and standardized solution for the automatic benchmarking of multi-paradigm solvers in combinatorial optimization.

0 citationsRead paper