Effective Out-of-Distribution (OOD) detection is criti-cal for ensuring the reliability of semantic segmentation models, particularly in complex road environments where safety and accuracy are paramount. Despite recent advancements in large language models (LLMs), notably GPT-4, which significantly enhanced multimodal reasoning through Chain-of-Thought (CoT) prompting, the application of CoT-based visual reasoning for OOD semantic segmentation remains largely unexplored. In this paper, through extensive analyses of the road scene anomalies, we identify three challenging scenarios where current state-of-the-art OOD segmentation methods consistently struggle: (1) densely packed and overlapping objects, (2) distant scenes with small objects, and (3) large foreground-dominant objects. To address the presented challenges, we propose a novel CoT-based framework targeting OOD detection in road anomaly scenes. Our method leverages the extensive knowledge and reasoning capabilities of foundation models, such as GPT-4, to enhance OOD detection through improved image understanding and prompt-based reasoning aligned with observed problematic scene attributes. Extensive experiments show that our framework consistently outperforms state-of-the-art methods on both standard benchmarks and our newly defined challenging subset of the RoadAnomaly dataset, offering a robust and interpretable solution for OOD semantic segmentation in complex driving environments.

Open-vocabulary semantic segmentation attempts to classify and outline objects in an image using arbitrary text labels, including those unseen during training. Self-supervised learning resolves numerous visual and linguistic processing problems when effectively trained. This study investigates simple yet efficient methods for adapting previously learned foundation models for open-vocabulary semantic segmentation tasks. Our research proposes "Beyond-Labels", a lightweight transformer-based fusion module that uses a small amount of image segmentation data to fuse frozen visual representations with language concepts. This strategy allows the model to leverage the extensive knowledge of pre-trained models without requiring significant retraining, making the approach data-efficient and scalable. Furthermore, we capture positional information in images using Fourier embeddings, improving generalization and enabling smooth and consistent spatial encoding. We perform thorough ablation studies to examine the main components of our proposed method. On the standard benchmark PASCAL-5i, the method performs better despite being trained on frozen vision and language representations. Index Terms: Beyond-Labels, open-vocabulary semantic segmentation, Fourier embeddings, PASCAL-5i

Estimating uncertainty from deep neural networks is a widely used approach for detecting out-of-distribution (OoD) samples, which typically exhibit high predictive uncertainty. However, conventional methods such as Monte Carlo (MC) Dropout often focus solely on either model or data uncertainty, failing to align with the semantic objective of OoD detection. To address this, we propose the Free-Energy Posterior Network, a novel framework that jointly models distributional uncertainty and identifying OoD and misclassified regions using free energy. Our method introduces two key contributions: (1) a free-energy-based density estimator parameterized by a Beta distribution, which enables fine-grained uncertainty estimation near ambiguous or unseen regions; and (2) a loss integrated within a posterior network, allowing direct uncertainty estimation from learned parameters without requiring stochastic sampling. By integrating our approach with the residual prediction branch (RPL) framework, the proposed method goes beyond post-hoc energy thresholding and enables the network to learn OoD regions by leveraging the variance of the Beta distribution, resulting in a semantically meaningful and computationally efficient solution for uncertainty-aware segmentation. We validate the effectiveness of our method on challenging real-world benchmarks, including Fishyscapes, RoadAnomaly, and Segment-Me-If-You-Can.

Interpretation of resolved polarized images of black holes by the Event Horizon Telescope (EHT) requires predictions of the polarized emission observable by an Earth-based instrument for a particular model of the black hole accretion system. Such predictions are generated by general relativistic radiative transfer (GRRT) codes, which integrate the equations of polarized radiative transfer in curved spacetime. A selection of ray-tracing GRRT codes used within the EHT collaboration is evaluated for accuracy and consistency in producing a selection of test images, demonstrating that the various methods and implementations of radiative transfer calculations are highly consistent. When imaging an analytic accretion model, we find that all codes produce images similar within a pixel-wise normalized mean squared error (NMSE) of 0.012 in the worst case. When imaging a snapshot from a cell-based magnetohydrodynamic simulation, we find all test images to be similar within NMSEs of 0.02, 0.04, 0.04, and 0.12 in Stokes I, Q, U , and V respectively. We additionally find the values of several image metrics relevant to published EHT results to be in agreement to much better precision than measurement uncertainties.

Observations of galaxies and primordial radiation suggest that the Universe is made mostly of non-luminous dark matter. Several types of new fundamental particles have been proposed as candidates for dark matter such as weakly interacting massive particles (WIMPs) but no definitive signal has been seen despite concerted efforts by many collaborations. One exception is the much-debated claim by the DAMA collaboration of a statistically significant annual modulation in the event rate of their experiment with a period and phase consistent with that expected from WIMP dark matter. Several groups have been working to develop experiments with the aim of reproducing DAMA's results using the same target medium. Here we report results from the initial operation of the COSINE-100 experiment. COSINE-100 uses sodium iodide as the target medium-the same medium as DAMA-and is designed to carry out a model-independent test of DAMA's claim. Initial data based on the first 59.5 days indicate that there is no excess of events over the expected background, confirming that DAMA's annual modulation signal is in severe tension with results from other experiments under the assumption of dark matter having spin independent interactions and the Standard Halo Model. COSINE-100 is now taking data to study the presence of dark matter-induced annual modulation in the event rate of the sodium iodide detectors.

We present results of a search for bosonic super-weakly interacting massive particles (BSW) as keV scale dark matter candidates that is based on an exposure of 97.7 kg$\cdot$year from the COSINE experiment. In this search, we employ, for the first time, Compton-like as well as absorption processes for pseudoscalar and vector BSWs. No evidence for BSWs is found in the mass range from 10 $\mathrm{keV/c}^2$ to 1 $\mathrm{MeV/c}^2$, and we present the exclusion limits on the dimensionless coupling constants to electrons $g_{ae}$ for pseudoscalar and $\kappa$ for vector BSWs at 90% confidence level. Our results show that these limits are improved by including the Compton-like process in masses of BSW, above $\mathcal{O}(100\,\mathrm{keV/c}^2)$.

Neutrino elastic scattering observation with NaI (NEON) is an experiment designed to detect neutrino-nucleus coherent scattering using reactor electron antineutrinos. NEON is based on an array of six NaI(Tl) crystals with a total mass of 13.3 kg, located at the tendon gallery that is 23.7 m away from a reactor core with a thermal power of 2.8 GW in the Hanbit nuclear power complex. The installation of the NEON detector was completed in December 2020, and since May 2021, the detector has acquired data at full reactor power. Based on the observed light yields of the NaI crystals of approximately 22, number of photoelectrons per unit keV electron-equivalent energy (keVee), and 6 counts/kg/keV/day background level at 2-6 keVee energy, coherent elastic neutrino-nucleus scattering observation sensitivity is evaluated as more than 3$\sigma$ assuming one-year reactor-on and 100 days reactor-off data, 0.2 keVee energy threshold, and 7 counts/keV/kg/day background in the signal region of 0.2-0.5 keVee. This paper describes the design of the NEON detector, including the shielding arrangement, configuration of NaI(Tl) crystals, and associated operating systems. The initial performance and associated sensitivity of the experiment are also presented.

COSINE-100 is a dark matter direct detection experiment with 106 kg NaI(Tl) as the target material. 210Pb and daughter isotopes are a dominant background in the WIMP region of interest and are detected via beta decay and alpha decay. Analysis of the alpha channel complements the background model as observed in the beta/gamma channel. We present the measurement of the quenching factors and Monte Carlo simulation results and activity quantification of the alpha decay components of the COSINE-100 NaI(Tl) crystals. The data strongly indicate that the alpha decays probabilistically undergo two possible quenching factors but require further investigation. The fitted results are consistent with independent measurements and improve the overall understanding of the COSINE-100 backgrounds. Furthermore, the half-life of 216Po has been measured to be 143.4 +/- 1.2 ms, which is consistent with and more precise than recent measurements.

We report measurements of annual and diurnal modulations of the cosmic-ray muon rate in the Yangyang underground laboratory (Y2L) using 952 days of COSINE-100 data acquired between September 2016 and July 2019. A correlation of the muon rate with the atmospheric temperature is observed and its amplitude on the muon rate is determined. The effective atmospheric temperature and muon rate variations are positively correlated with a measured effective temperature coefficient of $\alpha_{T}$ = 0.80 $\pm$ 0.11. This result is consistent with a model of meson production in the atmosphere. We also searched for a diurnal modulation in the underground muon rate by comparing one-hour intervals. No significant diurnal modulation of the muon rate was observed.

The system of a cold atomic gas in an optical lattice is governed by two factors: nonlinearity originating from the interparticle interaction, and the periodicity of the system set by the lattice. The high level of controllability associated with such an arrangement allows for the study of the competition and interplay between these two, and gives rise to a whole range of interesting and rich nonlinear effects. This review covers the basic idea and overview of such nonlinear phenomena, especially those corresponding to extended states. This includes "swallowtail" loop structures of the energy band, Bloch states with multiple periodicity, and those in "nonlinear lattices", i.e., systems with the nonlinear interaction term itself being a periodic function in space.

Dark photons are well motivated hypothetical dark sector particles that could account for observations that cannot be explained by the standard model of particle physics. A search for dark photons that are produced by an electron beam striking a thick tungsten target and subsequently interact in a 3 kiloton-scale neutrino detector in Yemilab, a new underground lab in Korea, is proposed. Dark photons can be produced by "darkstrahlung" or by oscillations from ordinary photons produced in the target and detected by their visible decays, "absorption" or by their oscillation to ordinary photons. By detecting the absorption process or the oscillation-produced photons, a world's best sensitivity for measurements of the dark-photon kinetic mixing parameter of $\epsilon^2 > 1.5 \times 10^{-13} (6.1 \times 10^{-13})$ at the 95\% confidence level (C.L.) could be obtained for dark photon masses between 80 eV and 1 MeV in a year-long exposure to a 100 MeV-100 kW electron beam with zero ($10^3$) background events. In parallel, the detection of $e^+e^-$ pairs from decays of dark photons with mass between 1 MeV and $\sim$86 MeV would have sensitivities of $\epsilon^2 > \mathcal{O}(10^{-17}) (\mathcal{O}(10^{-16}))$ at the 95% C.L. with zero ($10^3$) background events.

There is a long standing debate about whether or not the annual modulation signal reported by the DAMA/LIBRA collaboration is induced by Weakly Interacting Massive Particles~(WIMP) in the galaxy's dark matter halo scattering from nuclides in their NaI(Tl) crystal target/detector. This is because regions of WIMP-mass vs. WIMP-nucleon cross-section parameter space that can accommodate the DAMA/LIBRA-phase1 modulation signal in the context of the standard WIMP dark matter galactic halo and isospin-conserving~(canonical), spin-independent~(SI) WIMP-nucleon interactions have been excluded by many of other dark matter search experiments including COSINE-100, which uses the same NaI(Tl) target/detector material. Moreover, the recently released DAMA/LIBRA-phase2 results are inconsistent with an interpretation as WIMP-nuclide scattering via the canonical SI interaction and prefer, instead, isospin-violating or spin-dependent interactions. Dark matter interpretations of the DAMA/LIBRA signal are sensitive to the NaI(Tl) scintillation efficiency for nuclear recoils, which is characterized by so-called quenching factors~(QF), and the QF values used in previous studies differ significantly from recently reported measurements, which may have led to incorrect interpretations of the DAMA/LIBRA signal. In this article, the compatibility of the DAMA/LIBRA and COSINE-100 results, in light of the new QF measurements is examined for different possible types of WIMP-nucleon interactions. The resulting allowed parameter space regions associated with the DAMA/LIBRA signal are explicitly compared with 90\% confidence level upper limits from the initial 59.5~day COSINE-100 exposure. With the newly measured QF values, the allowed 3$\sigma$ regions from the DAMA/LIBRA data are still generally excluded by the COSINE-100 data.

We present the development of a decay energy spectroscopy (DES) method for the analysis of radioactive impurities using magnetic microcalorimeters (MMCs). The DES system was designed to analyze radionuclides, such as Ra-226, Th-228, and their daughter nuclides, in materials like copper, commonly used in rare-event search experiments. We tested the DES system with a gold foil absorber measuring 20x20x0.05 mm^3, large enough to accommodate a significant drop of source solution. Using this large absorber and an MMC sensor, we conducted a long-term measurement over ten days of live time, requiring 11 ADR cooling cycles. The combined spectrum achieved an energy resolution of 45 keV FWHM, sufficient to identify most alpha and DES peaks of interest. Specific decay events from radionuclide contaminants in the absorber were identified. This experiment confirms the capability of the DES system to measure alpha decay chains of Ra-226 and Th-228, offering a promising method for radio-impurity evaluation in ultra-low background experiments.

We report limits on WIMP dark matter derived from three years of data collected by the COSINE-100 experiment with NaI(Tl) crystals, achieving an improved energy threshold of 0.7 keV. This lowered threshold enhances sensitivity in the sub-GeV mass range, extending the reach for direct detection of low-mass dark matter. Although no excess of WIMP-like events was observed, the increased sensitivity enabled a model-independent comparison between the expected WIMP signal rate-based on mass limits from our data-and DAMA's reported modulation amplitude. Our findings strongly disfavor the DAMA signal as originating from WIMP interactions, fully excluding DAMA/LIBRA 3$\sigma$ allowed regions and providing enhanced WIMP mass limits by an order of magnitude in the spin-independent model compared to previous results. In the spin-dependent model, cross-section upper limits were obtained in the mass range [0.1-5.0] GeV/c$^2$, with additional sensitivity to sub-GeV WIMPs through the inclusion of the Migdal effect. These results represent substantial progress in low-mass dark matter exploration and reinforce constraints on the longstanding DAMA claim.

Video prediction (VP) generates future frames by leveraging spatial representations and temporal context from past frames. Traditional recurrent neural network (RNN)-based models enhance memory cell structures to capture spatiotemporal states over extended durations but suffer from gradual loss of object appearance details. To address this issue, we propose the strong recollection VP (SRVP) model, which integrates standard attention (SA) and reinforced feature attention (RFA) modules. Both modules employ scaled dot-product attention to extract temporal context and spatial correlations, which are then fused to enhance spatiotemporal representations. Experiments on three benchmark datasets demonstrate that SRVP mitigates image quality degradation in RNN-based models while achieving predictive performance comparable to RNN-free architectures.

The DAMA/LIBRA experiment has reported an annual modulation signal in NaI(Tl) detectors, which has been interpreted as a possible indication of dark matter interactions. However, this claim remains controversial, as several experiments have tested the modulation signal using NaI(Tl) detectors. Among them, the COSINE-100 experiment, specifically designed to test DAMA/LIBRA's claim, observed no significant signal, revealing a more than 3 $\sigma$ discrepancy with DAMA/LIBRA's results. Here we present COSINE-100U, an upgraded version of the experiment, which aims to expand the search for dark matter interactions by improving light collection efficiency and reducing background noise. The detector, consisting of eight NaI(Tl) crystals with a total mass of 99.1 kg, has been relocated to Yemilab, a new underground facility in Korea, and features direct PMT-coupling technology to enhance sensitivity. These upgrades significantly improve the experiment's ability to probe low-mass dark matter candidates, contributing to the ongoing global effort to clarify the nature of dark matter.

A new 2,400 L liquid scintillator has been produced for the COSINE-100 Upgrade, which is under construction at Yemilab for the next COSINE dark matter experiment phase. The linear-alkyl-benzene-based scintillator is designed to serve as a veto for NaI(Tl) crystal targets and a separate platform for rare event searches. We measured using a sample consisting of a custom-made 445 mL cylindrical Teflon container equipped with two 3-inch photomultiplier tubes. Analyses show activity levels of $0.091 \pm 0.042$ mBq/kg for $^{238}$U and $0.012 \pm 0.007$ mBq/kg for $^{232}$Th.

Very long baseline interferometry (VLBI) enables high-angular-resolution observations in astronomy and geodesy by synthesizing a virtual telescope with baselines spanning hundreds to thousands of kilometres. Achieving high instrumental phase stability in VLBI relies on the generation of high-quality, atomic-referenced RF local oscillator (LO) and RF-comb signals for the effective downconversion of celestial RF signals and precise phase calibration, respectively. As observing frequencies move into higher ranges with wider bandwidth, conventional electronic methods face significant challenges in maintaining the quality of these signals. Here, we demonstrate that an optical frequency comb (OFC) can be used as a versatile tool to generate and distribute low-noise and atomic-referenced RF-comb and RF-LO signals in the VLBI telescope. Hydrogen maser-stabilized optical pulses are transmitted over a timing-stabilized fibre link from the observatory building to the VLBI receiver system at the telescope, where photodetection converts them into the required RF-comb and RF-LO signals. In VLBI test observation, we successfully detected VLBI fringes and extracted the RF-combs characteristics in a format suitable for VLBI instrumental phase calibration. These results highlight the high potential of OFC-based technology for enhancing next-generation broadband VLBI measurements, advancing astrophysical research and facilitating intercontinental clock comparison.

We introduce a machine-learning framework termed coarse-grained all-atom force field (CGAA-FF), which incorporates coarse-grained message passing within an all-atom force field using equivariant nature of graph models. The CGAA-FF model employs grain embedding to encode atomistic coordinates into nodes representing grains rather than individual atoms, enabling predictions of both grain-level energies and atom-level forces. Tested on organic electrolytes, CGAA-FF achieves root-mean-square errors of 4.96 meV atom-1 for energy and 0.201 eV A-1 for force predictions. CGAA-FF significantly reduces computational costs, achieving about 22- and 14-fold improvements in simulation speed and memory efficiency, respectively, compared to the all-atom potential (SevenNet-0). Since this CGAA framework can be integrated into any equivariant architecture, we believe this work opens the door to efficient all-atom simulations of soft-matter systems.