Researchers have developed a novel approach to assessing the resilience of complex networks by modeling higher-order interactions, offering a more accurate prediction of cascading failures.
New research reveals that simply pairing humans with large language models doesn’t guarantee improved accuracy, highlighting critical flaws in current AI-integration training pipelines.
A novel method for generating privacy-preserving synthetic financial datasets enables more effective and explainable fraud detection through collaborative data sharing.
![Data streams from networked sensors converge upon a dual-model machine learning system-an [latex]LSTM[/latex] for forecasting and a Random Forest for anomaly detection-with results surfaced through a real-time Streamlit dashboard, establishing a closed-loop system for monitoring and preemptive alerts.](https://arxiv.org/html/2512.21801v1/leak_detection_architecture.drawio.png)
A new IoT framework leverages machine learning to forecast and detect leaks in liquid cooling systems, minimizing downtime and maximizing energy efficiency.
A new framework analyzes public Telegram channels to identify emerging cyber threats before they materialize, offering a proactive defense against malicious actors.
As natural language processing becomes increasingly integrated into critical systems, organizations need robust protocols to ensure these models are secure, compliant, and reliable.
![The correlation structure reveals a shared underlying pattern between synthetically generated data-built upon a [latex]Gaussian-Bernoulli[/latex] model-and data originating from real-world observations, suggesting the model effectively captures essential relationships present in the observed phenomena.](https://arxiv.org/html/2512.21823v1/gaussian_corr.png)
A new approach leverages conditional Restricted Boltzmann Machines to identify structural changes in financial time series beyond traditional volatility analysis.
![The research details a surgical video analysis pipeline-SpikeSurgSeg-which leverages a pretrained, spike-driven video encoder employing layer-wise tube masking for reconstruction, and integrates this with a spike-driven memory readout and feature pyramid network, ultimately achieving surgical scene segmentation with an SNN-based model distinguished by its two spike-driven CNN blocks and two spike-driven spatiotemporal Transformers exhibiting linear space-time computational complexity-[latex]O(N)[/latex].](https://arxiv.org/html/2512.21284v1/x2.png)
A new framework leverages spiking neural networks and video transformers to deliver accurate and energy-efficient segmentation of surgical video feeds.
![A diffusion-based system extracts multi-level features from traffic imagery by progressively introducing noise, then leveraging a U-Net architecture to denoise and identify optimal feature layers-a process refined through [latex]K[/latex]-means clustering for efficient fine-tuning-and ultimately fusing adjacent network layer features to represent both detailed and abstract traffic patterns.](https://arxiv.org/html/2512.21144v1/x2.png)
A new approach leverages the power of diffusion models and large language models to accurately identify network traffic even within the limitations of resource-constrained IoT devices.
A new analysis reveals the Lightning Network maintains surprising resilience over time, even as it becomes increasingly fragmented and concentrated in its structure.