Simple Beats Sophisticated: LSTMs Still Rule Stock Forecasting
New research challenges the prevailing trend toward transformer-based models, demonstrating that standard LSTMs consistently deliver superior stock price predictions.
Science
Decoding Intent: A Faster Path to Understanding Choice
New research unlocks efficient statistical methods for inferring the underlying motivations behind observed decisions in complex systems.
Uncovering Hidden Graph Structure for Better Classification
A new approach leverages frequent subgraph mining within persistent homology to improve the accuracy of graph classification tasks.
Thinking Before Acting: A Self-Reflective AI for Safer Autonomous Driving
Researchers have developed a new artificial intelligence framework that allows vehicles to critically evaluate their planned actions and adjust course before executing them, dramatically improving safety and adaptability.
Beyond Centralized Forecasts: Adapting Traffic Prediction with Federated Learning
A new framework, AutoFed, streamlines traffic prediction across distributed data sources by intelligently sharing knowledge without requiring manual model tuning.
Mapping the Skies: Optimizing Airline Alliances for Competitive Advantage
A new analytical framework uses network science to reveal how airline partnerships can maximize both market reach and healthy competition.
Adaptive Noise Cancellation: Reinventing Image Denoising with AI Agents
A new approach leverages the power of artificial intelligence to intelligently filter noise from images, surpassing the performance of existing denoising techniques.
The Fairness Horizon: Knowing When to Stop Searching for Unbiased Algorithms
A new statistical framework offers guarantees for adaptively finding less discriminatory machine learning models, addressing the critical challenge of certifying a sufficient search for algorithmic fairness.
Spiking Networks Tackle Wireless Channel Estimation
A new approach leverages the efficiency of spiking neural networks to accurately estimate ultra-wideband (UWB) communication channels.
The Rise of Differentiable Testing
![The DEFT framework establishes a differentiable path for reasoning about physical systems, enabling gradient-based optimization of control policies directly within the physics engine by representing continuous dynamics as [latex] \dot{x} = f(x, u) [/latex], where [latex] x [/latex] denotes the system state and [latex] u [/latex] represents the control input, thus bridging the gap between learned control and provable system stability.](https://arxiv.org/html/2512.23746v1/x2.png)
A new framework, DEFT, uses gradient-based optimization to dramatically improve the detection of hard-to-find faults in integrated circuits.