Science – Page 49 – Minority Mindset

Predicting Market Shifts: A New Approach to Causal Signals

01.01.2026 by qfx

$A consistent threshold of [latex]\theta=0.06[/latex] applied to both a forward-oriented causal observable [latex]\mathscr{F}(t)[/latex] and the raw composite signal [latex]\mathscr{F}_{0}(t)[/latex] demonstrates comparable performance-as measured by cumulative returns relative to buy-and-hold-and yields a similar frequency of state changes (trades) when applied to one-minute EURUSDT data.$

This review details a method for building causally valid predictive signals from financial time series, offering improved performance in specific market conditions.

Adapting to Change: Smarter IoT Botnet Detection

01.01.2026 by qfx

A new approach leverages latent space representation and graph neural networks to reliably identify IoT botnets even as attack patterns evolve.

Predictive Power: AI for Smarter Transformers

01.01.2026 by qfx

The simulation environment explores the relationship between residual magnetic flux and the closing angle of a transformer’s core, demonstrating how these inputs directly influence the magnitude of the resulting peak inrush current[10].

This review explores how artificial intelligence, particularly neural networks, is being used to monitor, diagnose, and optimize the performance of critical power transformer infrastructure.

Predictive Power: AI Meets Transformer Reliability

01.01.2026 by qfx

A Bayesian Physics-Informed Neural Network (PINN) approach is proposed to model probabilistic spatiotemporal thermal transformations, integrating Bayesian inference with the underlying physics to quantify uncertainty and improve predictive capability.

New research demonstrates how integrating physics-based models with artificial intelligence significantly enhances the accuracy and trustworthiness of transformer health monitoring.

Can AI Play Venture Capitalist?

31.12.2025 by qfx

Unlike prior approaches focused on isolated decision-making, this work models the interplay between startups and investors by simultaneously considering the viewpoints of all involved parties, offering a more holistic understanding of collective behavior.

A new framework simulates the dynamics of venture capital investment using interacting AI agents to predict startup success with greater accuracy.

Keeping Credit Models Current: An Adaptive Approach to Risk Forecasting

31.12.2025 by qfx

The architecture anticipates deployment on AWS, acknowledging that any chosen structure merely seeds a future landscape of inevitable compromise.

As digital lending scales, maintaining the accuracy of credit risk models requires continuous adaptation and robust monitoring.

Decoding Automated Vehicle Accidents: A Data-Driven Approach

31.12.2025 by qfx

Analysis of incident data reveals a correlation between pre-crash speed, automation level, and injury severity, with a disproportionate number of automated vehicle (AV) crashes occurring in the top ten urban centers-suggesting that while automation exists, its effectiveness diminishes as speed increases and is concentrated in specific, high-traffic environments.

New research analyzes over 2,500 crash records to identify key patterns and contributing factors in vehicles with varying levels of automation.

Predicting the Wireless Future with AI

31.12.2025 by qfx

Wireless traffic prediction is achieved through a two-stage process, beginning with spatially-aware clustering and culminating in the application of the TIDES framework-a system leveraging large language models for spatial-temporal learning.

A new approach leverages the power of artificial intelligence to anticipate network traffic and optimize performance for next-generation wireless systems.

Fortifying Future Networks: A Resilience Blueprint

31.12.2025 by qfx

$Resilience against attack vectors isn’t a static property, but emerges from a dynamic interplay between vulnerabilities, network state [latex]X_{t}[/latex], and three classes of defense-proactive hardening, responsive adaptation, and retrospective learning from performance outcomes [latex]Y_{t}[/latex]-that continuously refine the system’s ability to withstand compromise.$

This review explores the evolving threat landscape and emerging strategies for building cyber-resilient next-generation networks.

When Federated Learning Falls Apart: Why Data Splits Break AI Models

31.12.2025 by qfx

The aggregated global model successfully replicates the precise neural network structure learned by individual clients-as demonstrated by the prevalence of shared connections-indicating effective knowledge consolidation without introducing representational drift during the learning process.

New research reveals that performance drops in decentralized learning aren’t just random, but a result of internal network structures disintegrating under non-ideal data conditions.