Science – Page 13 – Minority Mindset

When AI Forgets What It Knows: Adapting to New Realities

14.12.2025 by qfx

$During test-time adaptation, a novel approach blends geometric proximity with predictive confidence to create hybrid targets, subtly guiding features toward plausible classifier weights while actively repelling negatives through the loss function $\mathcal{L}\_{\text{NC}}$.$

A new study reveals how neural networks lose accuracy when faced with unfamiliar data, and introduces a method to realign their internal representations for better performance.

From Text to Knowledge: A Smarter Way to Classify Data

14.12.2025 by qfx

The pipeline transforms textual documents into a graph structure, initially extracting representations and defining relationships between nodes, then leverages a large language model to label a selection of these nodes before propagating those labels across the entire graph using a graph neural network.

A new approach combines the power of language models with graph networks to achieve accurate text classification, even when labeled data is limited.

Fast and Accurate Aerodynamic Prediction with Kernelized Neural Surrogates

14.12.2025 by qfx

A new approach combines the strengths of kernel methods and neural networks to dramatically accelerate aerodynamic simulations while maintaining high accuracy.

Beyond Data Hunger: Cultivating Language Technology for All

14.12.2025 by qfx

A new approach to building inclusive language tools prioritizes ethical data creation and community involvement for languages often left behind.

Can AI Spot the Phish? The Hidden Weaknesses in Language Model Security

14.12.2025 by qfx

LLM-PEA presents a framework wherein large language models are leveraged to enhance prompting effectiveness through a process of prompt engineering and analysis, ultimately optimizing interactions and outcomes.

Despite impressive accuracy, new research reveals that large language models are surprisingly vulnerable to cleverly crafted phishing attacks.

Quantized Confidence: Shrinking Neural Networks Without Losing Certainty

14.12.2025 by qfx

A Bayesian Neural Network extends the standard neural network paradigm by replacing deterministic parameters with probability distributions, allowing for a mathematically rigorous treatment of uncertainty and enabling probabilistic inference.

New research demonstrates that Bayesian Neural Networks can be dramatically compressed for efficient deployment without sacrificing their crucial ability to estimate prediction uncertainty.

Smarter, Faster Networks: Optimizing Early-Exit Architectures for Edge AI

14.12.2025 by qfx

The proposed AEBNAS framework systematically designs early-exit neural networks through a three-stage process: encoding architectural possibilities, precisely calibrating exit thresholds, and optimizing for a balance between predictive accuracy and computational efficiency-quantified by the number of MACs-thereby achieving a provable trade-off between performance and resource consumption.

Researchers have developed a new framework to automatically design efficient neural networks that can deliver faster performance on resource-constrained devices.

Beyond Accuracy: Forecasting with Confidence in Critical Time Series

14.12.2025 by qfx

SoTra bypasses typical probabilistic modeling by directly translating predicted probability distributions-$p_t$-into continuous embeddings, facilitating fully differentiable trajectory generation and ultimately enabling point forecasts decoded through application-specific risk minimization rather than conventional distance metrics.

A new approach to time series forecasting prioritizes managing risk and uncertainty, crucial for applications like clinical decision support where errors can have serious consequences.

Predicting System Failure: A New Path to Robust Control

14.12.2025 by qfx

This review connects the ability to forecast and diagnose faults in complex systems to a fundamental property called ‘pre-normality’, offering new strategies for design and control.

Keeping AI Safe While It Learns

13.12.2025 by qfx

Standard fine-tuning of large language models rapidly erodes previously established safety alignments, with this degradation accelerating when exposed to maliciously crafted user data, while adapted continual learning strategies demonstrably preserve those safety constraints even under adversarial conditions.

New research shows how to prevent large language models from losing their safety guardrails as they are continuously updated with new information.