The Price of Playing Follow the Leader
New research reveals how competitive experimentation in dynamic pricing can inadvertently push prices higher, even without explicit collusion.
Science
Seeing Through the Storm: Ensuring Reliable Object Detection for Self-Driving Cars
As autonomous vehicles navigate increasingly complex environments, maintaining reliable object detection in challenging weather conditions is paramount for safety and performance.
Untangling Neural Network Dynamics
![Feature learning performance in a linear single hidden layer network demonstrates strong agreement between theoretical predictions-specifically, the Neural Network Gaussian Process (NNGP) and the Li & Sompolinsky approach-and direct simulation, as evidenced by comparable mean discrepancies [latex]\langle\Delta\alpha\rangle[/latex] across a parameter space defined by [latex]P=80[/latex], [latex]N=100[/latex], [latex]d=200[/latex], and an Ising task with [latex]p=0.1[/latex] and a regulator of [latex]\kappa=0.01[/latex], validated through Langevin dynamics with one million training steps and twenty thousand samples.](https://arxiv.org/html/2602.12855v1/x14.png)
New research bridges the gap between Gaussian processes and recurrent neural networks to reveal the underlying principles governing deep learning.
Smarter Stockrooms: How AI is Reinventing Inventory Control
New research reveals that combining the power of artificial intelligence with established operations research and human insight yields significant improvements in managing supply and demand.
Decoding Starlight: AI Predicts Molecular Fingerprints in Space
![The comparison of infrared spectra-predicted by a Graph Neural Network and computed via Density Functional Theory for a selection of pericondensed molecules-demonstrates a compelling correspondence, even after applying a Gaussian broadening with a Full Width at Half Maximum of [latex]10\text{\,}{\mathrm{cm}}^{-1}[/latex], suggesting the model’s capacity to approximate complex molecular vibrational characteristics.](https://arxiv.org/html/2602.12560v1/x4.png)
Researchers have harnessed the power of graph neural networks to rapidly predict the infrared spectra of polycyclic aromatic hydrocarbons, the molecules that permeate interstellar space.
Can Neural Networks Reason About Graphs?
New research explores the fundamental limits and capabilities of graph neural networks when applied to algorithmic tasks on graph structures.
Designing Circuits with AI: A New Approach to Analog and Mixed-Signal Creation
Researchers have developed a reinforcement learning framework that automates the design of complex analog and mixed-signal circuits, leveraging simulation feedback for optimized performance.
Predicting Bitcoin’s Next Move: A Deep Learning Fusion
A new hybrid model combines the strengths of temporal fusion transformers, attention-based recurrent networks, and gradient boosting to enhance the accuracy of Bitcoin price predictions.
Seismic Signals: Hunting for Earthquake Foreshadows with AI
![The architecture systematically reduces high-dimensional input data-initially [latex]512 \times 32 \times 32 \times 15[/latex]-to a single value through 3D convolutional layers, leveraging temporal dilation-specifically powers of 2-to halve the dimensionality with each application, ultimately producing an uncalibrated anomaly score that distills complex input into a concise metric.](https://arxiv.org/html/2602.12408v1/x3.png)
A new deep learning framework analyzes evolving patterns in earthquake data to pinpoint subtle anomalies that could indicate changing seismic risk.
Beyond Prediction: Building Recommendation Systems That Know What They Don’t Know
A new framework explicitly models uncertainty in generative recommendation to create more robust, trustworthy, and risk-aware suggestions.