cs papers | Gist.Science

Longitudinal Component Level Analysis of Neural Network Training Dynamics Beyond Accuracy and Loss

This study introduces a non-intrusive framework for analyzing epoch-level neural network component dynamics, revealing that models with similar accuracy often exhibit distinct internal behaviors and demonstrating that descriptor-guided pruning can outperform random pruning while confirming that internal metrics offer complementary diagnostic value without yet surpassing magnitude-based criteria.

Sharon Yalov Handzel, Din Kosberg2026-09-23

💻 computer science

Explainable AI (XAI) for Credit Scoring Model Validation

This research proposes and empirically validates a comprehensive XAI-driven framework that integrates post-hoc explanation techniques and quantitative quality metrics into the credit scoring model lifecycle to balance predictive performance with regulatory compliance, transparency, and stakeholder trust in digital banking.

Albert Schulle2026-09-23

💻 computer science

Mosaic: Modular Orchestration of Specialised AI Components

The paper proposes MOSAIC, a modular architecture that replaces expensive frontier language models with a strict hierarchy of hardcoded orchestration logic and specialized smaller components to achieve comparable accuracy at a significantly reduced inference cost through mechanisms like model-scale substitution, early termination, and cumulative partial resolution.

NABIL ISLAM2026-09-23

💻 computer science

Fish monitoring in river by deploying IoT-UWSNs

This paper presents an AI-driven fish monitoring system that utilizes IoT-enabled underwater sensor networks and convolutional neural networks to automatically identify fish species in real-time, thereby enhancing aquatic ecosystem management and biodiversity conservation despite ongoing challenges in dataset expansion and hardware optimization.

Muhammad Tahir2026-09-23

💻 computer science

Privacy-Preserving Continual Learning for Detecting Concept Drift in Distributed Artificial Intelligence Systems

This paper proposes and evaluates a privacy-preserving continual learning framework for distributed AI systems that effectively detects concept drift and mitigates catastrophic forgetting using differentially private signals and regularized rehearsal, demonstrating that accurate drift detection is feasible at moderate privacy budgets despite a modest increase in detection delay.

SAI DOONDI KOTHAPALLI2026-09-23

💻 computer science

When high AUROC does not travel: internal-to-external performance differences in machine-learning models for antimicrobial resistance

This meta-research study reveals that machine learning models for antimicrobial resistance frequently exhibit a significant drop in performance when moving from internal to external validation, with the magnitude of this gap varying widely and precluding the use of a universal correction factor for reported AUROCs.

Dripto Roy2026-09-23

💻 computer science

Do Clinical Notes Improve ICU Mortality Prediction? A Controlled Comparison of Structured and Multimodal EHR Fusion Strategies

This study demonstrates that while clinical notes do not consistently improve overall ICU mortality discrimination beyond structured EHR variables, they can enhance mortality recall when integrated via specific multimodal fusion strategies, underscoring the necessity of rigorous baseline comparisons and multi-dimensional evaluation in clinical AI.

Ashish Katyal2026-09-23

💻 computer science

Quantum Superposition over Near Optimal Seeds for Maximum Independent Set on Dense Graphs

This paper presents a quantum variational algorithm that leverages uniform superpositions of near-optimal seeds and interference-based post-selection to solve Maximum Independent Set problems on dense graphs up to 400 nodes, significantly outperforming standard VQE and classical heuristics on hard instances where previous methods stall.

Kalyan Dasgupta, Sumanta Mukherjee, Dhriti Verma, Surya Shravan Kumar Sajja, Abhishek Singh, Dzung Phan, Jayant Kalagnan (…)2026-09-23

💻 computer science

Uncertainty-Aware Dual-Attention Temporal Convolutional Networks for Remaining Useful Life Prediction

This paper proposes an Uncertainty-Aware Dual-Attention Temporal Convolutional Network (UA-DA-TCN) framework that integrates multi-scale temporal convolutions, dual attention mechanisms, and Monte Carlo Dropout to achieve accurate, interpretable, and uncertainty-aware Remaining Useful Life predictions for aircraft engines, outperforming existing baselines on the NASA C-MAPSS dataset.

Mubashar Abbas2026-09-23

💻 computer science

A multi-model study of diagnostic faithfulness in AI-generated histopathology images

This study evaluates seven text-to-image models on a large cohort of histopathology cases and finds that while the best-performing models approach diagnostic interpretability, they frequently omit critical features and fail to capture clinically meaningful quality, indicating that current generative AI and evaluation metrics are not yet ready for responsible adoption in oncology.

Hong-Yu Zhou, Qinxin Wang, Jiachen Ji, Xihai Zhao2026-09-23

← Previous Next →