Search

We introduce Posterior Distillation Sampling (PDS), a novel optimization method for parametric image editing based on diffusion models. Existing optimization-based methods, which leverage the powerful 2D prior of diffusion models to handle various parametric images, have mainly focused on generation. Unlike generation, editing requires a balance between conforming to the target attribute and preserving the identity of the source content. Recent 2D image editing methods have achieved this balance by leveraging the stochastic latent encoded in the generative process of diffusion models. To extend the editing capabilities of diffusion models shown in pixel space to parameter space, we reformulate the 2D image editing method into an optimization form named PDS. PDS matches the stochastic latents of the source and the target, enabling the sampling of targets in diverse parameter spaces that align with a desired attribute while maintaining the source's identity. We demonstrate that this optimization resembles running a generative process with the target attribute, but aligning this process with the trajectory of the source's generative process. Extensive editing results in Neural Radiance Fields and Scalable Vector Graphics representations demonstrate that PDS is capable of sampling targets to fulfill the aforementioned balance across various parameter spaces., Comment: Project page: https://posterior-distillation-sampling.github.io/

As large language models become increasingly integrated into daily life, detecting implicit toxicity across diverse contexts is crucial. To this end, we introduce LifeTox, a dataset designed for identifying implicit toxicity within a broad range of advice-seeking scenarios. Unlike existing safety datasets, LifeTox comprises diverse contexts derived from personal experiences through open-ended questions. Experiments demonstrate that RoBERTa fine-tuned on LifeTox matches or surpasses the zero-shot performance of large language models in toxicity classification tasks. These results underscore the efficacy of LifeTox in addressing the complex challenges inherent in implicit toxicity. We open-sourced the dataset\footnote{\url{https://huggingface.co/datasets/mbkim/LifeTox}} and the LifeTox moderator family; 350M, 7B, and 13B., Comment: 11 pages, 5 figures, NAACL 2024

Let $N$ be a positive integer and $\Gamma$ be a subgroup of $\mathrm{SL}_2(\mathbb{Z})$ containing $\Gamma_1(N)$. Let $K$ be an imaginary quadratic field and $\mathcal{O}$ be an order of discriminant $D_\mathcal{O}$ in $K$. Under some assumptions, we show that $\Gamma$ induces a form class group of discriminant $D_\mathcal{O}$ (or of order $\mathcal{O}$) and level $N$ if and only if there is a certain canonical model of the modular curve for $\Gamma$ defined over a suitably small number field. In this way we can find an interesting link between two different subjects, which will be useful in the study of certain quadratic Diophantine equations in terms of primes $p$., Comment: 18 pages, The title has been changed

Medical image segmentation, which is essential for many clinical applications, has achieved almost human-level performance via data-driven deep learning technologies. Nevertheless, its performance is predicated upon the costly process of manually annotating a vast amount of medical images. To this end, we propose a novel framework for robust semi-supervised medical image segmentation using diagonal hierarchical consistency learning (DiHC-Net). First, it is composed of multiple sub-models with identical multi-scale architecture but with distinct sub-layers, such as up-sampling and normalisation layers. Second, with mutual consistency, a novel consistency regularisation is enforced between one model's intermediate and final prediction and soft pseudo labels from other models in a diagonal hierarchical fashion. A series of experiments verifies the efficacy of our simple framework, outperforming all previous approaches on public benchmark dataset covering organ and tumour., Comment: Accepted to IEEE EMBC 2024 (46th Annual International Conference of the IEEE Engineering in Medicine & Biology Society)

When two galaxies merge, they often produce a supermassive black hole binary (SMBHB) at their center. Numerical simulations with stars and cold dark matter show that SMBHBs typically stall out at a distance of a few parsecs apart and take billions of years to coalesce. This is known as the final parsec problem. We suggest that ultralight dark matter (ULDM) halos around SMBHBs can generate dark matter waves due to dynamical friction. These waves can effectively carry away orbital energy from the black holes, rapidly driving them together. To test this hypothesis, we performed numerical simulations of black hole binaries inside ULDM halos. Due to gravitational cooling and quasi-normal modes, the loss-cone problem can be avoided. The decay time scale gives lower bounds on masses of the ULDM particles and SMBHBs comparable to observational data. Our results imply that ULDM waves can lead to the rapid orbital decay of black hole binaries., Comment: Accepted for publication in Physics Letters B

We present a novel method for identifying cosmic web filaments using the IllustrisTNG (TNG100) cosmological simulations and investigate the impact of filaments on galaxies. We compare the use of cosmic density field estimates from the Delaunay Tessellation Field Estimator (DTFE) and the Monte Carlo Physarum Machine (MCPM), which is inspired by the slime mold organism, in the DisPerSE structure identification framework. The MCPM-based reconstruction identifies filaments with higher fidelity, finding more low-prominence/diffuse filaments and better tracing the true underlying matter distribution than the DTFE-based reconstruction. Using our new filament catalogs, we find that most galaxies are located within 1.5-2.5 Mpc of a filamentary spine, with little change in the median specific star formation rate and the median galactic gas fraction with distance to the nearest filament. Instead, we introduce the filament line density, Sigma_fil(MCPM), as the total MCPM overdensity per unit length of a local filament segment, and find that this parameter is a superior predictor of galactic gas supply and quenching. Our results indicate that most galaxies are quenched and gas-poor near high-line density filaments at z<=1. At z=0, quenching in log(M*/Msun)>10.5 galaxies is mainly driven by mass, while lower-mass galaxies are significantly affected by the filament line density. In high-line density filaments, satellites are strongly quenched, whereas centrals have reduced star formation, but not gas fraction, at z<=0.5. We discuss the prospect of applying our new filament identification method to galaxy surveys with SDSS, DESI, Subaru PFS, etc. to elucidate the effect of large-scale structure on galaxy formation., Comment: Accepted for publication in ApJ. Data available at https://doi.org/10.5281/zenodo.10929036

Lead (Pb2+) toxification in society is one of the most concerning public health crisis that remains unaddressed. The exposure to Pb2+ poisoning leads to a multitude of enduring health issues, even at the part-per-billion scale (ppb). Yet, public action dwarfs its impact. Pb2+ poisoning is estimated to account for 1 million deaths per year globally, which is in addition to its chronic impact on children. With their ring-shaped cavities, crown ethers are uniquely capable of selectively binding to specific ions. In this work, for the first time, the synergistic integration of highly-scalable silicon photonics, with crown ether amine conjugation via Fischer esterification in an environmentally-friendly fashion is demonstrated. This realises a photonic platform that enables the in-situ, highly-selective and quantitative detection of various ions. The development dispels the existing notion that Fischer esterification is restricted to organic compounds, laying the ground for subsequent amine conjugation for various crown ethers. In this work, the platform is engineered for Pb2+ detection, demonstrating a large dynamic detection range of 1 - 262000 ppb with high selectivity against a wide range of relevant ions. These results indicate the potential for the pervasive implementation of the technology to safeguard against ubiquitous lead poisoning in our society.

Forecasting sea ice concentration (SIC) and sea ice velocity (SIV) in the Arctic Ocean is of great significance as the Arctic environment has been changed by the recent warming climate. Given that physical sea ice models require high computational costs with complex parameterization, deep learning techniques can effectively replace the physical model and improve the performance of sea ice prediction. This study proposes a novel multi-task fully conventional network architecture named hierarchical information-sharing U-net (HIS-Unet) to predict daily SIC and SIV. Instead of learning SIC and SIV separately at each branch, we allow the SIC and SIV layers to share their information and assist each other's prediction through the weighting attention modules (WAMs). Consequently, our HIS-Unet outperforms other statistical approaches, sea ice physical models, and neural networks without such information-sharing units. The improvement of HIS-Unet is more significant to when and where SIC changes seasonally, which implies that the information sharing between SIC and SIV through WAMs helps learn the dynamic changes of SIC and SIV. The weight values of the WAMs imply that SIC information plays a more critical role in SIV prediction, compared to that of SIV information in SIC prediction, and information sharing is more active in marginal ice zones (e.g., East Greenland and Hudson/Baffin Bays) than in the central Arctic.