Researchers have developed a novel diffusion model to map and interpret the complex internal states of large neural networks, offering new ways to control and analyze their decision-making processes.
![Specular reflections, proving difficult to replicate in forged faces due to the mathematical foundations of the Phong illumination model [latex]I = r + a\cos\theta [/latex], are leveraged by the proposed SRI-Net to identify inconsistencies and expose manipulations undetectable by spatial or frequency-based forgery detection methods.](https://arxiv.org/html/2602.06452v1/x1.png)
Researchers are leveraging inconsistencies in how light reflects off faces to develop a more robust and generalizable method for detecting manipulated videos.
![The study demonstrates a robust linear relationship between the [latex]W1W1[/latex] magnitude and stellar mass for galaxies lacking spectral energy distribution estimates, with propagated uncertainties on [latex]log_{10}(M_{\star})[/latex] typically around ±0.05 dex-though peaking at intermediate masses where dust content and star-formation histories introduce greater diversity.](https://arxiv.org/html/2602.06492v1/figs/predicted_err_vs_mass.png)
A new study shows that artificial neural networks can accurately determine the stellar masses of galaxies using only readily available observational data, offering a faster alternative to complex modeling.
New research reveals that machine learning can effectively identify promising investment opportunities hidden within SEC filings related to insider purchases of small-cap stocks.
New research leverages the power of artificial intelligence to reveal how people prioritize their financial goals, mirroring Maslow’s classic hierarchy of needs.
New research demonstrates that advanced language agents can negotiate as effectively as – and sometimes better than – human MBAs, opening up new frontiers in AI evaluation.
New research reveals that even before training, the architecture of transformer models introduces systematic biases affecting how they process information.
A new analysis explores how advanced time-series models, enriched with external data like weather patterns, are improving the accuracy of electricity demand predictions.
Researchers are leveraging flow matching to build generative models that learn more interpretable and controllable latent representations.
![Spatially-adaptive mixture-of-experts leverage Sobel edge detection [latex] \nabla I [/latex] to dynamically refine feature maps, enabling a nuanced understanding of image structure and localized processing within a neural network.](https://arxiv.org/html/2602.05100v1/x2.png)
A novel deep learning architecture combines the power of U-Nets with the interpretability of fuzzy logic to deliver both accurate and understandable edge detection.