Unlocking Tensor Decomposition with Aletheia
![The computational cost of verifying mathematical proofs-specifically, those generated by a semiautonomous system-scales relative to the complexity of the problem, as demonstrated by the inference cost per proof being benchmarked against the solution to Erdős-1051 [latex] [/latex].](https://arxiv.org/html/2602.21201v1/FP_compute.png)
A new approach, Aletheia, delivers a faster, more scalable solution for a key challenge in tensor decomposition by autonomously tackling the FirstProof problem.
Science
Yields Don’t Lie: Cracks in Crypto’s Efficient Market Theory
New research reveals that staking and lending returns in Ethereum-based cryptoasset markets defy established economic principles, suggesting persistent inefficiencies.
Beyond Spotting Fakes: A New Approach to Image Authenticity
![The SimLBR framework enhances generalization by training a model to discern real images from subtly perturbed fakes, achieved through sampling either true or fabricated labels and blending limited information from these fakes within a pretrained latent space-a process that effectively tightens the decision boundary around the distribution of unperturbed real images and improves robustness against unseen generative models, as expressed by [latex] p(x) [/latex].](https://arxiv.org/html/2602.20412v1/sec/img/framework_aayushapprove_final.png)
Researchers are shifting the focus from directly identifying AI-generated images to better understanding what constitutes a ‘real’ image, leading to more reliable detection methods.
Predicting the Plug-In Future: Forecasting EV Charging Demand
As electric vehicle adoption accelerates, accurately forecasting charging demand is crucial for grid stability and efficient energy management.
Unmasking AI’s Secrets: How Models Reveal Their Training Data
New research demonstrates a method for reliably identifying whether a language model was trained on a specific piece of data, raising critical privacy concerns.
Taming High-Frequency Waves with Deep Learning
![The study demonstrates that a finite-difference multi-grid domain decomposition learning (FD-MGDL) approach, when applied to the two-dimensional Helmholtz equation [latex] \equation{} [/latex], exhibits performance sensitivity to structural variations, as evidenced by comparative training loss curves.](https://arxiv.org/html/2602.20719v1/MGDL-123.jpg)
A new framework combines the precision of traditional methods with the adaptive power of deep neural networks to efficiently solve challenging wave propagation problems.
Predicting City Movement: A New AI Approach
Researchers have developed a novel framework that uses artificial intelligence to realistically generate urban mobility patterns, offering potential benefits for traffic simulation and autonomous driving.
Seeing Clearly in Space: AI Targets Camera Faults
A new deep learning approach enables spacecraft cameras to autonomously identify and mitigate straylight, enhancing onboard processing and system resilience.
GANs Guide the Search: A New Era for Bayesian Sampling
Researchers are harnessing the power of generative adversarial networks to dramatically improve the speed and accuracy of Markov Chain Monte Carlo methods for complex statistical inference.
Uncovering Solar Secrets: A New Way to Find the Unexpected
Researchers have developed a machine learning tool to efficiently identify rare and unusual events in vast amounts of solar observation data, promising new discoveries in solar physics.