Science – Page 3 – Minority Mindset

Predicting Power Grid Failures Before They Happen

28.03.2026 by qfx

Standard deviation as an aggregation function, combined with features extracted from the frequency domain, proves most effective in accurately predicting faults within the system under study.

A new simulation-based approach dramatically improves the accuracy of fault prediction in distribution systems, offering utilities crucial lead time for preventative maintenance.

Fragile Pegs: Green Bonds and Systemic Risk in Stablecoins

28.03.2026 by qfx

The analysis demonstrates that the USDC stablecoin maintains a remarkably stable peg to the dollar, exhibiting minimal deviation even under fluctuating market conditions - a testament to its robust design and the mechanisms employed to ensure price stability, despite the inherent volatility of the broader cryptocurrency landscape.

A new analysis reveals that backing stablecoins with illiquid green bonds can amplify market stress and trigger prolonged de-pegging events.

Can Search Engines Outsmart the Spammers?

28.03.2026 by qfx

Traditional search engines contrast with LLM-enhanced search engines (LLMSE) in their approach to information retrieval, with the latter leveraging large language models to move beyond keyword matching towards a more nuanced understanding of user intent and contextual relevance.

A new study examines how the integration of large language models into search affects its vulnerability to manipulative SEO tactics.

Building a Digital Twin for 6G Networks

28.03.2026 by qfx

Researchers are exploring AI-powered world models to predict and optimize the performance of future wireless systems.

Predicting the Future, with Confidence

28.03.2026 by qfx

The proposed D2D model accurately replicates the evolution of probability density functions-observed through numerical simulation of the Lorenz63 system and represented by red data points- mirroring the behavior of the underlying dynamical system across successive lead times.

A new neural network framework moves beyond single-value predictions to model the full range of possible outcomes in complex dynamical systems.

Predicting the Future of AI Compute: A New Approach to GPU Workload Forecasting

28.03.2026 by qfx

PRISM demonstrates superior performance relative to established baselines, indicating an advancement in the evaluated metric.

Accurately forecasting demand for GPUs is critical for efficient AI infrastructure, and researchers have developed a new framework that dramatically improves prediction accuracy.

Building Trust in Collaborative AI

28.03.2026 by qfx

The system integrates an Agentic Trust Control Loop into the Federated Learning process, establishing a cyclical relationship where trust informs agent behavior and subsequent learning refines that trust - a feedback mechanism designed to optimize collaborative intelligence.

A new approach to federated learning empowers industrial networks to proactively manage trust and improve system stability.

Scaling Up Weather Prediction with Neural Networks

27.03.2026 by qfx

A new study reveals how systematically increasing the size of neural networks, combined with strategic training techniques, can dramatically improve weather forecasting accuracy.

Seeing is Believing: AI-Powered Crop Mapping with Transparent Insights

27.03.2026 by qfx

The systematic obscuring of features within the top fifteen most salient elements reveals that the system’s reliance on these elements is not uniform; certain features exhibit a disproportionate influence on overall performance, suggesting a brittle architecture vulnerable to even minor perturbations in those key areas.

A new framework combines the power of deep learning with explainable AI to achieve near-perfect crop classification and provide farmers with actionable data.

When Less Data Means More Resilience

27.03.2026 by qfx

$Epistemic compression fundamentally differs from likelihood maximization by employing rate reduction-quantified as [latex]\Delta R[/latex]-as a geometric sieve that collapses high-variance noise onto the underlying low-dimensional manifold, effectively orthogonalizing class subspaces and recovering the invariant causal structure, in contrast to the unconstrained feature space and brittle representations produced by fitting noise instances to separate classes.$

As AI models grow in complexity, a surprising strategy for improving performance in unpredictable conditions is gaining traction: deliberately limiting the information they process.