3031 papers
This paper introduces GutenbergLM, a 109.5M-parameter decoder-only transformer trained from scratch on a temporally balanced Project Gutenberg corpus with era-conditioning tokens, demonstrating that explicit temporal stratification enables the model to generate distinct, period-appropriate writing styles across Early, Middle, and Modern literary eras.
This paper introduces Internal Combustion Engine Optimization (ICEO), a novel metaheuristic algorithm inspired by the four-stroke engine cycle and governed by an adaptive Engine Control Unit, which significantly outperforms state-of-the-art methods in solving high-dimensional, nonlinear optimization problems by effectively balancing exploration and exploitation to avoid premature convergence.
This paper reviews the transformative potential and challenges of AI and IoT in Industry 4.0 supply chains through global and regional case studies, ultimately proposing a scalable framework that integrates pilot projects with governance mechanisms to achieve both productivity and sustainability.
This paper proposes a cryptographic containment architecture that secures agentic LLM systems against prompt injection by enforcing a high-entropy, token-authenticated syntactic boundary between instructions and data, thereby rendering injection attempts structurally inert with negligible failure rates and minimal runtime overhead.
This paper proposes AWMCEL, a novel multi-view ensemble framework that combines discriminant least squares regression with adaptive weighting and -norm regularization to automatically assign optimal weights to classifiers based on their performance, thereby effectively leveraging complementary information from diverse data views without complex parameter tuning.
This paper proposes a source-aware reasoning framework for large language models that extends the inference scope to include data producers and their upstream context, enabling more accurate discrimination between genuine system abnormalities and data artifacts while significantly improving detection accuracy and reducing temporal errors.
This paper reviews key image denoising algorithms, analyzes the performance of Gaussian, linear, and non-linear isotropic smoothing techniques, and concludes that both linear and non-linear smoothing offer effective solutions for mitigating noise-induced information loss.
This paper introduces a fully reproducible, open-source retrieval-augmented pipeline that combines fine-tuned and zero-shot large language models to achieve GPT-5.1-level performance in low-resource Latin translation, validated through new task-ready training corpora and challenging out-of-domain benchmarks.
This paper proposes DuTLR-Net, a deep unrolled tensor low-rank network that integrates physical priors via a Cross-Scale Manifold Alignment module and an ADMM-based iterative optimization backbone to achieve superior, interpretable hyperspectral image classification with high spatial-spectral consistency, even in small-sample scenarios.
This paper proposes a sustainable blockchain framework for carbon markets that combines a hybrid Proof-of-Stake and Proof-of-Work consensus mechanism for small-to-medium applications with the Osmosis DeFi platform for large-scale operations, demonstrating significant improvements in energy efficiency, throughput, and latency compared to traditional systems.