Tracking Market Stress with Food Prices
New research shows how volatility estimates derived from local food price data can serve as an early warning system for economic instability in developing nations.
Science
Securing the Rise of Intelligent Agents
As AI systems gain autonomy, a robust security framework is crucial, and this review lays the groundwork for formally verifying their trustworthiness.
Beyond Averages: Quantile Modeling for Smarter Risk Forecasting
A new approach to modeling financial return dynamics leverages the full distribution of potential outcomes, offering more accurate predictions of extreme losses.
Predicting the Future with Clarity: A New Approach to Time-Series Forecasting
Researchers have developed a novel method that enhances both the accuracy and interpretability of time-series predictions, offering significant improvements for applications like early warning systems and predictive maintenance.
Beyond the Data: Harnessing Collective Intelligence for Causal Understanding
A new approach to causal learning argues that integrating human expertise and AI can overcome limitations of purely data-driven methods.
Modeling Cancer to Predict Drug Response
A new framework combines the strengths of machine learning and symbolic reasoning to more accurately forecast how colorectal cancers will respond to treatment.
Can Subtle Shifts Predict the Big One? A New Approach to Earthquake Forecasting
![Model rescaling, evaluated using the Brier Skill Score (BSS) as a function of a multiplicative scaling factor and the mean number of earthquakes [latex]\overline{n\_{\mathrm{eq}}}[/latex], demonstrates that empirical BSS-based rescaling-obtained by fitting along the ridge of positive skill-and logit-based prior correction consistently outperform uncorrected methods, with the latter further refined by an offset calibrated to optimize BSS, as evidenced by comparative performance across both training and independent test epochs and substantiated by the distribution of [latex]\overline{n\_{\mathrm{eq}}}[/latex] for all space-time samples and those culminating in earthquakes with [latex]M\_{W} \geq 5[/latex].](https://arxiv.org/html/2603.03079v1/2603.03079v1/x1.png)
A new study explores whether deep learning, combined with analysis of the Gutenberg-Richter b-value, can offer a marginal improvement in predicting earthquake occurrences.
Decoding the Heart’s Signals: AI Predicts Structural Disease from Standard ECGs
![The additive model estimates entry-wise functions - termed “Partial Effect” and represented as [latex]\hat{f}_{j}(x)[/latex] - alongside corresponding “Predictive Value” expressed as [latex]\sigma\{\hat{h}_{j}(\mathbf{X})\}[/latex], where [latex]\hat{h}_{j}(\mathbf{X})[/latex] defines the function’s contribution, to discern patterns within datasets labeled with conditions such as inferior myocardial infarction (IMI), non-diagnostic T abnormalities (NDT), atrial fibrillation (AFIB), and left ventricular hypertrophy (LVH).](https://arxiv.org/html/2603.02616v1/2603.02616v1/x5.png)
A new approach leverages the power of artificial intelligence to identify indicators of structural heart disease directly from routine electrocardiograms.
Beyond Keywords: A New Dataset Connects Financial News to Market Movements
Researchers have created a large-scale dataset, FinTexTS, designed to improve the link between financial news and stock price predictions.
Securing the Agent Ecosystem: A New Approach to Skill Supply Chains
As AI agents become increasingly reliant on external skills, ensuring the integrity of those skills is paramount, and this research introduces a novel framework to address this growing security challenge.