arXiv:2608.21142v1 Announce Type: new Abstract: Structured pruning reduces the size and inference cost of large language models (LLMs) by removing weight columns, but the resulting output error can degrade accuracy. Existing training-free compensation methods use an additive bias or a single orthog
arXiv:2608.21147v1 Announce Type: new Abstract: The cyclic structure of physiological processes offers a natural prior for self-supervised representation learning, and the cardiac cycle provides a particularly well-defined setting in which to exploit it. We derive a phase-equivariant self-supervise
arXiv:2608.21172v1 Announce Type: new Abstract: Federated fine-tuning enables large language models to adapt on edge devices without centralizing private data, but practical deployments must address hardware instability and adversarial update corruption together. Thermally constrained clients may t
arXiv:2608.21186v1 Announce Type: new Abstract: Surgical procedures such as laparoscopic appendectomy are complex, high-stakes processes, yet formalizing their workflows for decision support remains a significant challenge. Inducing probabilistic planning domain models in this setting is particular
arXiv:2608.21199v1 Announce Type: new Abstract: Transformer-based tabular foundation models such as TabPFN achieve strong predictive performance but incur quadratic computational cost with context length. On the other hand, subquadratic SSM-based alternatives such as Hydra trade away accuracy for e
arXiv:2608.21207v1 Announce Type: new Abstract: Imputing physiological time series (arterial blood pressure, blood glucose, etc.) is essential for addressing the missingness that pervades clinical data. Yet modern imputation methods perform poorly in this domain: a recent benchmark found that simpl
arXiv:2608.21251v1 Announce Type: new Abstract: Operational telemetry can be jointly anomalous while every individual stream stays inside its familiar range. TRACE-C is an auditable strictly-prior rank-calibrated detector for aligned multi-stream telemetry: same-regime rolling median/MAD residuals
arXiv:2608.21277v1 Announce Type: new Abstract: State-of-the-art multivariate time-series forecasters can model complex temporal and cross-variable dependencies, yet their opaque representations provide limited insight into why a particular forecast is produced. This lack of transparency restricts
Researchers at CERN have created microscopic versions of the early Universe by colliding surprisingly small atomic nuclei at nearly the speed of light. The collisions produced quark-gluon plasma, the ultra-hot matter believed to have filled the cosmos shortly after the Big Bang. Even more intriguing
Researchers have used AI to uncover the DNA signature of a key genetic “switch” involved in turning genes on. After analyzing about 500,000 DNA sequences, the model identified the initiator in roughly 60% of human genes. The breakthrough could help predict the effects of harmful mutations and eventu
Scientists have developed an ultrathin coating that could dramatically improve how efficiently heat is transferred during condensation. By turning tiny polymer structures once considered “defects” into places where water droplets can form, then helping those droplets detach quickly, the new surface
Astronomers have found stars that repeatedly skim past supermassive black holes, surviving each encounter while producing a new burst of light. In some systems, those flares mysteriously fade with every return. Researchers now think the key may be stars that were already spinning extremely fast befo
Freezing the liquid core of an optical fiber produced an extreme environment where light and sound interact more than 1,000 times more strongly than in ordinary fibers. Researchers used the effect to create optoacoustic memory, potentially paving the way for lower-energy photonic computers and advan
Astronomers using JWST have discovered that massive early galaxies contain far more small, faint stars than expected. That hidden population could make some of these galaxies three to four times more massive than previous estimates. The finding makes it even harder to explain how enormous, mature ga
The U.S. power grid is at an inflection point, with factors ranging from blistering summer heat to data center power demand that is potentially set to double from 2025 to 2027, an uptick in wildfires and aging transmission infrastructure. But a significant effort to reshape the grid is also underway
Scientists watched a light-triggered hidden state form inside a material in only 30 femtoseconds, revealing a step that had never been seen before. The material first entered a fleeting electronic state in which its bonds reorganized in a repeating pattern, followed by tiny atomic shifts. This ultra
Two very different types of quantum particles may be able to form stable droplets that hold themselves together, challenging decades of conventional thinking. The prediction could soon be tested experimentally and may reveal an unexpectedly rich world of new quantum phases.
For decades, scientists have puzzled over why certain atomic nuclei unexpectedly produce large numbers of low-energy gamma rays. A new experiment traced the effect to magnetic changes inside the nucleus, where protons and neutrons effectively flip their tiny internal magnets. The discovery could sha
Lawrence Livermore National Laboratory (LLNL) is partnering with Oxylus, a Yale spinout focused on CO2 electrolysis technology, to develop and, potentially, commercialize a reactor that could convert carbon waste to methanol without alkali cations. LLNL won $2 million from the DOE’s Industrial Tec
Extreme experiments have revealed how diamond behaves at crushing pressures beyond those inside Neptune and Uranus, resolving a decades-old conflict between theory and observation. The results could help scientists boost fusion energy output while revealing more about the exotic diamond rain hidden