Mapping Flash Flood Risk Beyond the Grid
A new approach leverages terrain connectivity and advanced machine learning to improve flash flood susceptibility mapping in the mountainous region of Himachal Pradesh.
Science
Beyond Flat Networks: Hyperbolic Geometry Improves Bitcoin Transaction Analysis
New research shows that modeling the complex structure of Bitcoin transactions with hyperbolic graph neural networks significantly outperforms traditional Euclidean approaches.
Decoding Complex Systems: From Event Data to Root Cause
A new framework leverages the power of artificial intelligence to automatically diagnose faults in intricate, high-dimensional event sequences.
Predicting Sepsis Earlier with AI and Collaborative Data
A new framework leverages federated learning, knowledge graphs, and temporal transformers to improve early sepsis detection across multiple intensive care units.
Predicting Crisis: Can AI Make Sense of Geopolitical Chaos?
New research reveals that artificial intelligence systems demonstrate surprisingly adept strategic reasoning when analyzing unfolding international crises, though reliable prediction remains a significant challenge.
Simulating Disaster on Social Media: An Agentic Approach
![A system of interacting agents iteratively refines synthetic crisis-related tweets: a generator creates content, a compliance evaluator assesses its suitability, and a feedback augmenter relays evaluations back to the generator, with each accepted tweet accumulating over [latex]n[/latex] rounds to form a dataset [latex]\mathcal{D}\_{syn}[/latex], demonstrating a cyclical process of refinement inherent in complex systems.](https://arxiv.org/html/2603.13625v1/x1.png)
Researchers are using AI-powered workflows to generate realistic synthetic tweet datasets, overcoming the challenges of accessing real-time social media data during crises.
Catching the Shift: Predicting System Collapse with AI
A new approach combines the power of machine learning with key indicators to forecast critical transitions in complex systems before they happen.
Mapping the Blackout: AI Predicts Hurricane Recovery Times
A new graph neural network model leverages spatial relationships to forecast how long power will be out after major storms.
When Cyberattacks Go Viral: Modeling Firm Risk and Insurance Impact
![The data reveals a temporal pattern of newly infected firms across size categories from May to July 2024, quantified as [latex] I_{k} [/latex], indicating the daily incidence of infection spreading through the business landscape.](https://arxiv.org/html/2603.15369v1/x4.png)
A new stochastic model connects the spread of cyberattacks within organizations to their growth and the potential cascading effects on insurance portfolios.
Navigating Shifting Markets: A Causal Approach to Smarter Trading
![A normalization process-applied to financial indicators including Money Flow Index, Relative Strength Index, Bollinger Bands percentage, and Moving Average Convergence Divergence for both EURUSDT and BTCUSDT-centers the data by subtracting a 5000-minute rolling median [latex]I~^{(k)}\_{t}=I^{(k)}\_{t}-m^{(k)}\_{t}[/latex], then scales it using the rolling Median Absolute Deviation [latex]s^{(k)}\_{t}[/latex] to produce a dimensionless, comparable metric [latex]Z^{(k)}\_{t}=\tilde{I}^{(k)}\_{t}/s^{(k)}\_{t}[/latex] that facilitates direct cross-indicator analysis and aggregation.](https://arxiv.org/html/2603.13638v1/x1.png)
A new strategy leverages causal inference and forward-looking indicators to adapt to the ever-changing dynamics of financial markets and improve portfolio performance.