Holding AI-Based Systems Accountable in the Public Sector: A Systematic Review

Public Performance & Management Review, Год журнала: 2025, Номер unknown, С. 1 - 34

Опубликована: Фев. 23, 2025

Язык: Английский

---

AI-Driven Design Optimization for Sustainable Buildings: A Systematic Review

Energy and Buildings, Год журнала: 2025, Номер unknown, С. 115440 - 115440

Опубликована: Фев. 1, 2025

Язык: Английский

---

Detection of various gastrointestinal tract diseases through a deep learning method with ensemble ELM and explainable AI

Expert Systems with Applications, Год журнала: 2024, Номер 256, С. 124908 - 124908

Опубликована: Июль 30, 2024

The rising prevalence of gastrointestinal (GI) tract disorders worldwide highlights the urgent need for precise diagnosis, as these diseases greatly affect human life and contribute to high mortality rates. Fast identification, accurate classification, efficient treatment approaches are essential addressing this critical health issue. Common side effects include abdominal pain, bloating, discomfort, which can be chronic debilitating. Nausea vomiting also frequent, leading difficulties in maintaining adequate nutrition hydration. current study intends develop a deep learning (DL)-based approach that automatically classifies GI diseases. For first time, GastroVision dataset with 8000 images 27 different was utilized work design computer-aided diagnosis (CAD) system. This presents novel lightweight feature extractor compact size minimum number layers named Parallel Depthwise Separable Convolutional Neural Network (PD-CNN) Pearson Correlation Coefficient (PCC) selector. Furthermore, robust classifier Ensemble Extreme Learning Machine (EELM), combined pseudo inverse ELM (ELM) L1 Regularized (RELM), has been proposed identify more precisely. A hybrid preprocessing technique, including scaling, normalization, image enhancement techniques such erosion, CLAHE, sharpening, Gaussian filtering, employed enhance representation improve classification performance. consists twenty-four only 0.815 million parameters 9.79 MB model size. PD-CNN-PCC-EELM extracts features, reduces computational overhead, achieves excellent performance on multiclass images. achieved highest precision, recall, f1, accuracy, ROC-AUC, AUC-PR values 88.12 ± 0.332 %, 87.75 0.348 87.12 0.324 98.89 92 respectively, while testing time 0.000001 s. comparative utilizes 10-fold cross-validation, ablation various state-of-the-art (SOTA) transfer (TL) models extractors. Then, PCC EELM integrated TL generate predictions, notably terms real-time processing capability; significantly outperforms other models. Moreover, explainable AI (XAI) methods, SHAP (Shapley Additive Explanations), heatmap, guided Grad-Cam (Gradient-weighted Class Activation Mapping), Grad-CAM, Saliency mapping, have explore interpretability decision-making capability model. Therefore, provides practical intelligence increasing confidence diagnosing real-world scenarios.

Язык: Английский

---

Automated and explainable machine learning for monitoring lipid and protein oxidative damage in mutton using hyperspectral imaging

Food Research International, Год журнала: 2025, Номер 203, С. 115905 - 115905

Опубликована: Фев. 1, 2025

Язык: Английский

---

On Explainability of Reinforcement Learning-Based Machine Learning Agents Trained with Proximal Policy Optimization That Utilizes Visual Sensor Data

Applied Sciences, Год журнала: 2025, Номер 15(2), С. 538 - 538

Опубликована: Янв. 8, 2025

In this paper, we address the issues of explainability reinforcement learning-based machine learning agents trained with Proximal Policy Optimization (PPO) that utilizes visual sensor data. We propose an algorithm allows effective and intuitive approximation PPO-trained neural network (NN). conduct several experiments to confirm our method’s effectiveness. Our proposed method works well for scenarios where semantic clustering scene is possible. approach based on solid theoretical foundation Gradient-weighted Class Activation Mapping (GradCAM) Classification Regression Tree additional proxy geometry heuristics. It excels in explanation process a virtual simulation system video relatively low resolution. Depending convolutional feature extractor network, obtains 0.945 0.968 accuracy black-box model. The has important application aspects. Through its use, it possible estimate causes specific decisions made by due current state observed environment. This estimation makes determine whether as expected (decision-making related model’s observation objects belonging different classes environment) detect unexpected, seemingly chaotic behavior might be, example, result data bias, bad design reward function or insufficient generalization abilities publish all source codes so can be reproduced.

Язык: Английский

---

Clever Hans in the Loop? A Critical Examination of ChatGPT in a Human-in-the-Loop Framework for Machinery Functional Safety Risk Analysis

Eng—Advances in Engineering, Год журнала: 2025, Номер 6(2), С. 31 - 31

Опубликована: Фев. 7, 2025

This paper presents a first-of-its-kind evaluation of integrating Large Language Models (LLMs) within Human-In-The-Loop (HITL) framework for risk analysis in machinery functional safety, adhering to ISO 12100. The methodology systematically addresses LLM limitations, such as hallucinations and lack domain-specific expertise, by embedding expert oversight ensure reliable compliant outputs. Applied four diverse industrial case studies—motorized gates, autonomous transport vehicles, weaving machines, rotary printing presses—this study assesses the applicability ChatGPT routine tasks central safety workflows, hazard identification assessment. results demonstrated substantial improvements: during HITL involvement subsequent iterations assessment with feedback, complete agreement ground truth was achieved across all use cases. also identified additional scenarios edge cases, enriching analysis. Efficiency gains were notable, time efficiency rated at 4.95 out 5, on average, studies. Overall accuracy (4.7 5) usability (4.8 ratings robustness ensuring practical Likert scale evaluations reflected high confidence refined outputs, emphasizing critical role enhancing both trust usability. highlights importance prompt design, revealing that longer initial prompts improve accuracy, while shorter iterative maintain without compromising efficiency. process further ensures outputs align standards requirements. underscores transformative potential generative AI activities rigorous human safety-critical, regulated industries.

Язык: Английский

---

Machine learning approach for analyzing feature importance in alternative fuel vehicle selection

Travel Behaviour and Society, Год журнала: 2025, Номер 39, С. 100987 - 100987

Опубликована: Янв. 22, 2025

Язык: Английский

---

Enhancing explainability in real-world scenarios: Towards a robust stability measure for local interpretability

Expert Systems with Applications, Год журнала: 2025, Номер 274, С. 126922 - 126922

Опубликована: Фев. 24, 2025

Язык: Английский

---

Local Universal Rule-based eXplainer (LUX)

SoftwareX, Год журнала: 2025, Номер 30, С. 102102 - 102102

Опубликована: Фев. 27, 2025

Язык: Английский

---

Trustworthy Artificial Intelligence for Industrial Operations and Manufacturing: Principles and Challenges

Springer series in advanced manufacturing, Год журнала: 2025, Номер unknown, С. 179 - 197

Опубликована: Янв. 1, 2025

Язык: Английский

---