Institution profile

Babes-Bolyai University

Academic institutioneurope · ro
Official website
Research library24linked papers
Opportunities0open roles
Selected work

Representative Papers

CHiPS: Character Histograms and Positional Signals for Lightweight Authorship Attribution in Romanian Texts

Jul 24, 2026

This work proposes CHiPS, a lightweight, preprocessing-free, and leakage-resistant authorship attribution method tailored for Romanian text. CHiPS uniquely transforms character positional signals into frequency-domain features (FFT12-LR) and combines them with character histogram-based modeling (CH-SVM) to capture stylistic patterns—eliminating the need for tokenization, syntactic parsing, or pretrained language models. The approach introduces a novel decision-level fusion mechanism designed to be leakage-safe, achieving high transparency and controllability with minimal feature engineering. Evaluated on the ROST and ROSTories-cleaned datasets, CHiPS attains accuracy and macro F1 scores of 0.9310/0.9341 and 0.8919/0.8708, respectively, demonstrating strong performance under stringent data integrity constraints.

0 citationsRead paper

Post-Operative Glioma Segmentation via Loss Stabilization, Normalization and Subspace Attention

Jul 23, 2026

This study addresses the significant performance degradation of post-operative glioma segmentation under cross-institutional clinical protocols due to domain shift. To mitigate this issue, the authors propose brain-mask percentile normalization combined with voxel-level contrastive learning to stabilize training dynamics. Furthermore, they introduce a Subspace-Aware Class Attention (SACA) module to recalibrate bottleneck features and enhance sensitivity to contrast-enhancing tumor regions. Integrated into the nnU-Net framework, the proposed method achieves a Dice coefficient of 0.94 for whole lesion segmentation on the MU-GLIOMA-POST dataset. The SACA-augmented variant reduces boundary error to an HD95 of 2.92 mm and yields a relative 9.1% improvement in sensitivity to enhancing tumor regions.

0 citationsRead paper
Recent publications

Latest Papers

CHiPS: Character Histograms and Positional Signals for Lightweight Authorship Attribution in Romanian Texts

Jul 24, 2026

This work proposes CHiPS, a lightweight, preprocessing-free, and leakage-resistant authorship attribution method tailored for Romanian text. CHiPS uniquely transforms character positional signals into frequency-domain features (FFT12-LR) and combines them with character histogram-based modeling (CH-SVM) to capture stylistic patterns—eliminating the need for tokenization, syntactic parsing, or pretrained language models. The approach introduces a novel decision-level fusion mechanism designed to be leakage-safe, achieving high transparency and controllability with minimal feature engineering. Evaluated on the ROST and ROSTories-cleaned datasets, CHiPS attains accuracy and macro F1 scores of 0.9310/0.9341 and 0.8919/0.8708, respectively, demonstrating strong performance under stringent data integrity constraints.

0 citationsRead paper

Post-Operative Glioma Segmentation via Loss Stabilization, Normalization and Subspace Attention

Jul 23, 2026

This study addresses the significant performance degradation of post-operative glioma segmentation under cross-institutional clinical protocols due to domain shift. To mitigate this issue, the authors propose brain-mask percentile normalization combined with voxel-level contrastive learning to stabilize training dynamics. Furthermore, they introduce a Subspace-Aware Class Attention (SACA) module to recalibrate bottleneck features and enhance sensitivity to contrast-enhancing tumor regions. Integrated into the nnU-Net framework, the proposed method achieves a Dice coefficient of 0.94 for whole lesion segmentation on the MU-GLIOMA-POST dataset. The SACA-augmented variant reduces boundary error to an HD95 of 2.92 mm and yields a relative 9.1% improvement in sensitivity to enhancing tumor regions.

0 citationsRead paper