cs papers | Gist.Science

How Reliable and Explainable Are Machine Learning Models for Dementia Detection? A Leakage-Aware Evaluation

This study demonstrates that while machine learning models can achieve high dementia detection performance on the OASIS-2 dataset when rigorously evaluated with participant-level cross-validation to prevent data leakage, their apparent reliance on MRI features is inconsistent and often overshadowed by the Mini-Mental State Examination (MMSE), highlighting the critical need for independent validation to ensure clinical explainability and generalizability.

Md. Hasibur Rahman, Md Jamil Hasan, Md. Sajjad Hossain, Md Sultanul Islam Ovi2026-09-09

💻 computer science

Intensity and Landmark Based Registration Under Stain Normalization: A Comparative Study on Differently Stained Histopathology Images

This study demonstrates that a robust landmark-based registration technique utilizing thin-plate spline interpolation outperforms intensity-based methods like Elastix and ANTs in aligning histopathology images with varying stains, particularly when combined with deep learning-based stain normalization, despite the ongoing challenge of accurate landmark annotation.

Muhammad Talha Ali, Anja Keskinarkaus, Tapio Seppänen, Md Ziaul Hoque2026-09-09

💻 computer science

Precision, Physics, and Confabulation: Softmax Temperature as an Analog of Dopaminergic Precision-Weighting in LLM Hallucination

This pilot study proposes a formal mathematical analogy between dopamine-mediated precision-weighting in computational psychiatry and softmax temperature in LLMs, finding that while lowering temperature increases the confidence of hallucinations (reducing hedging) in ungrounded conditions without affecting grounded accuracy, the results remain statistically underpowered to definitively confirm the theoretical link.

Kunal Dhanda2026-09-09

💻 computer science

MUSiNET: A Complex Network Model to Explore Underlying Relationship among Musics

This paper introduces MUSiNET, a novel complex network model that integrates intrinsic and extrinsic musical features to map global song relationships, revealing structural insights about cultural integration and popularity dynamics that offer a robust, data-driven alternative to traditional music classification for applications like recommendation and discovery.

Avijit Gayen, Avik Mondal, Praswasti Sharma, Angshuman Jana2026-09-09

💻 computer science

MASS-ScanDMM: A Panoramic Image Scanpath Prediction Method Integrating Multiple Attention Mechanisms and Spherical Semantic Enhancement

This paper proposes MASS-ScanDMM, a panoramic image scanpath prediction method that integrates Multiple Attention Scene Perception (MASP) and Spherical Information Enhancement Perception (SIEP) mechanisms to achieve superior accuracy and generalization across multiple datasets.

Jianwei Li, Yuxin Chen, Hongjue Chen, Sixi Chen2026-09-09

💻 computer science

Bias Detection and Mitigation in Extractive Text Summarization

The paper introduces BiasZero, a fairness-aware framework for extractive text summarization that integrates multi-dimensional bias detection, counterfactual data augmentation, and adversarial debiasing to significantly reduce bias intensity by up to 40% while maintaining high summary quality with less than 1% ROUGE degradation.

Kavitha Sooda, Tejeshwara B M, Naga Pragnathmik Gurrala, Mohammed Zeeshan Umar, Ayman Khan, Simon Kevin Mohan2026-09-09

💻 computer science

Reputation-Driven Federated Learning: Enhancing Trust in Healthcare Data Sharing with Blockchain

This paper proposes a secure, blockchain-based federated learning architecture for healthcare that integrates Attribute-based Access Control with a calculated trust attribute and a monitoring mechanism to ensure data privacy and detect malicious nodes during model sharing.

Jahan Zeb Shahid, Stelvio Cimato2026-09-09

💻 computer science

Tensor Erasing-Based Data Augmentation for Fully Automated Brain Tumor MRI Segmentation

This paper proposes a novel Tensor Erasing data augmentation method that enhances the robustness and generalization of deep learning models for fully automated 3D brain tumor MRI segmentation by simulating image corruption through random voxel erasure, thereby effectively addressing the challenge of scarce annotated datasets.

dagui qin, yu wang2026-09-09

💻 computer science

SOFE: SDN-based Task Offloading Technique for Fog-IoEV Network

This paper proposes SOFE, an SDN-based task offloading technique for Fog-IoEV networks that utilizes a binary linear programming model to optimize task assignment to fog nodes or parked vehicles based on load, mobility, energy, and deadlines, thereby reducing response time and extending electric vehicle lifespans compared to existing methods.

Ahmed Jawad Kadhim Al-Shaibany, Jaber Ibrahim Naser, Mustafa Noaman Kadhim, Dhulfiqar Zoltán Aalwahab2026-09-09

💻 computer science

A Reliable Multimodal Conversational Emotion Recognition Model Based on a Dual-Stage Attention Mechanism

This paper proposes DSA-MERC, a reliable multimodal conversational emotion recognition model that utilizes a dual-stage attention mechanism combining Bi-Directional MultiAttention and an audio-guided leader-follower network with multi-scale optimization, alongside confidence and transfer learning strategies, to effectively integrate multimodal features and mitigate label noise for improved accuracy and robustness.

Xinheng Wang, Lun Xie, Xurong Xie, Xiaolan Peng, Zhiliang Wang2026-09-08

← Previous Next →