3031 papers
This paper presents a DenseNet-121-based deep learning system that achieves 96.52% accuracy in classifying four types of brain tumors from Bangladeshi MRI datasets, while employing Grad-CAM and related techniques to ensure model transparency and trustworthiness.
This paper proposes IDW-ADAFormer, an iTransformer-based framework that integrates adaptive Inverse-Distance Weighting decomposition, a kernel-based alignment diversity attention module, and an AQI-aware loss function to achieve accurate and interpretable medium-term forecasting of spatially distributed particulate matter concentrations.
This paper proposes a Pareto hybrid artificial fish swarm algorithm that integrates improved feeding mechanisms and simulated annealing to simultaneously optimize assembly sequence planning and assembly line balancing by minimizing balancing loss, switching times, and beat time.
This study evaluates the performance of the Apriori algorithm on large-scale synthetic transactional datasets to demonstrate that while lowering support thresholds increases rule diversity, it significantly raises computational costs, ultimately highlighting the need to balance algorithmic depth with efficiency through optimal threshold selection.
This paper proposes a Population State-Aware Cooperative Adaptive Differential Evolution (PSAC-DE) algorithm featuring a unified state controller to solve a four-objective arterial traffic signal coordination model, demonstrating significant improvements in delay, stops, green wave bandwidth, and carbon emissions compared to mainstream optimizers in simulated Hangzhou scenarios.
The paper proposes FreqPrompt-AD, a training-free framework that enhances zero-shot industrial anomaly detection and segmentation by leveraging frequency-guided local semantic prompting to overcome the limitations of global image-text alignment in CLIP-style models, achieving state-of-the-art performance on multiple benchmarks.
This paper introduces a measurement-theoretic framework based on utility axioms to evaluate the rationality of Large Language Models, revealing that many models exhibit systematic violations of preference transitivity that undermine decision coherence, even as some patterns mirror human behavior.
TopoLeaf is a training-free, zero-shot visual anomaly detection framework for agricultural cyber-physical systems that combines geometric distance and local persistent homology on foundation model features to robustly identify novel plant diseases and structural deviations across diverse crop domains without requiring annotated data.
This paper presents a systematic review of 80 studies (2010–2026) analyzing the evolving arms race between AI-driven cyber threat detection and adversarial evasion techniques, offering a unified taxonomy, visual analysis of trends, and a critical assessment of challenges and future directions within Zero Trust security environments.
This paper introduces TimeSynth, a controlled benchmarking framework that exposes the limitations of standard pointwise metrics in evaluating health-signal digital twins by revealing significant phase and frequency errors in leading models, thereby advocating for architecture-driven model selection based on temporal fidelity diagnostics.