2787 papers
The paper introduces Causal Transformer (CT), a self-attention architecture that overcomes the dimensionality limitations of existing gradient-based methods to enable robust causal discovery in the 200–500 variable regime, successfully identifying biologically validated cancer driver networks on consumer hardware where previous approaches fail.
This study utilizes a suite of machine learning algorithms, with Logistic Regression emerging as the most effective model, to classify high-risk paratransit drivers in Gazipur, Bangladesh, based on socioeconomic and operational factors, thereby offering transportation authorities a data-driven tool for targeted safety interventions.
This paper proposes CCM-YOLO, a lightweight and accurate photovoltaic defect detection model utilizing visible light imaging that integrates novel decoupled downsampling, mixed attention, and heterogeneous convolution modules to significantly reduce computational cost while outperforming state-of-the-art methods in both precision and model efficiency.
This paper proposes a two-stage framework combining Fashion LoRA guidance and Garment Region-Reweighted Flow Matching to address the structural and textural degradation of garments in ancient paintings, significantly improving restoration quality and providing auditable, structure-constrained digital candidates for cultural heritage preservation.
This paper introduces RAGtio, a modular, open-source framework built on Haystack and Qdrant that enables systematic, reproducible evaluation of hybrid RAG retrieval pipelines in the biomedical domain through dual assessment modes and accessible interfaces for both technical and non-technical users.
This paper introduces XCognitive SwarmNet, a cognitive swarm-based multi-agent framework that integrates collaborative reasoning, explainable decision intelligence, and digital twin simulation to significantly enhance the accuracy, quality, efficiency, and transparency of AI-driven software engineering and business automation for Industry 4.0.
This paper introduces Delegation-Aware Runtime Contracts (DARC), a formal framework that treats authority transfer in LLM multi-agent systems as a machine-checkable process to enforce constraint preservation, capability revocation, and state recovery through deterministic runtime mediation, thereby addressing the safety risks of treating delegation as mere natural language.
This paper presents a two-pass framework for hand-drawn diagram parsing that combines a multi-head graph-evidence network with a deterministic assembler to effectively bridge the gap between pixel-level visual evidence and accurate structural graph recovery, achieving high performance in node detection, connector tracing, and directed link reconstruction.
This paper investigates the impact of different data partitioning strategies on Echo State Network performance for predicting Dynamic Aperture in hadron storage rings, specifically exploring the automation of the training-test split boundary using the derivative of analytical scaling laws derived from the Nekhoroshev theorem.
This paper proposes a reliability-weighted spectral allocation method that automatically determines task-specific parameter counts and locations using gradient-based evidence, enabling parameter-efficient visual fine-tuning with full spectral cores that outperform linear probing while requiring no user-specified rank budgets.