Search

Although 3D object editing has the potential to significantly influence various industries, recent research in 3D generation and editing has primarily focused on converting text and images into 3D models, often overlooking the need for fine-grained control over the editing of existing 3D objects. This paper introduces a framework that employs a pre-trained regressor, enabling continuous, precise, attribute-specific modifications to both the stylistic and geometric attributes of vehicle 3D models. Our method not only preserves the inherent identity of vehicle 3D objects, but also supports multi-attribute editing, allowing for extensive customization without compromising the model's structural integrity. Experimental results demonstrate the efficacy of our approach in achieving detailed edits on various vehicle 3D models., Comment: 18 pages, 12 figures

We present the findings of a comprehensive and detailed analysis of merger tree data from ultra-high-resolution cosmological $N$-body simulations. The analysis, conducted with a particle mass resolution of $5 \times 10^3 h^{-1} M_{\odot}$ and a halo mass resolution of $10^7 h^{-1} M_{\odot}$, provides sufficient accuracy to suppress numerical artefacts. This study elucidates the dynamical evolution of subhaloes associated with the Milky Way-like host haloes. Unlike more massive dark matter haloes, which have been extensively studied, these subhaloes follow a distinct mass evolution pattern: an initial accretion phase, followed by a tidal stripping phase where mass is lost due to the tidal forces of the host halo. The transition from accretion to stripping, where subhaloes reach their maximum mass, occurs around a redshift of $z\simeq1$. Smaller subhaloes reach this point earlier, while larger ones do so later. Our analysis reveals that over 80 per cent of subhaloes have experienced mass loss, underscoring the universality of tidal stripping in subhalo evolution. Additionally, we derived the eccentricities and pericentre distances of subhalo orbits from the simulations and compare them with those of nearby satellite galaxies observed by the Gaia satellite. The results demonstrate a significant alignment between the orbital elements predicted by the cold dark matter model and the observed data, providing robust support for the model as a credible candidate for dark matter., Comment: 7 pages, 5 figures, accepted for publication in PASJ

Foams exhibit absorptive properties and are widely used for cleaning contaminants, oil recovery, and selective mineral extraction via froth flotation. Although foam absorption has historically been linked to equilibrium osmotic pressure, empirical observations show that drainage occurs at levels much lower than theoretical predictions. In this study, we investigate the effective osmotic pressure in foams. The experimental findings indicate that effective osmotic pressure is primarily influenced by short-term dynamics and is governed by the yield stress rather than equilibrium osmotic pressure. Gravity-induced flow drives bubble movement downward, subjecting the lower bubbles to increased pressure. Once the pressure exerted by the solution exceeds the yield stress, internal bubble rearrangement occurs, and drainage is promoted by the relaxation of inter-bubble repulsive forces. In contrast, below the yield stress, these repulsive forces suppress drainage. While foams have traditionally been modeled as immobile porous media, the reality involves kinematic coupling between the solution and the bubbles, which establishes yield stress as the key factor in determining effective osmotic pressure. This framework is also relevant to the dynamics of soft jammed systems, such as blood flow in vessels, emulsions, and biological tissues, offering significant advancements in the understanding of soft jammed system behavior., Comment: 17 pages, 9 figures

In this study, we aim to develop a domestic service robot (DSR) that, guided by open-vocabulary instructions, can carry everyday objects to the specified pieces of furniture. Few existing methods handle mobile manipulation tasks with open-vocabulary instructions in the image retrieval setting, and most do not identify both the target objects and the receptacles. We propose the Dual-Mode Multimodal Ranking model (DM2RM), which enables images of both the target objects and receptacles to be retrieved using a single model based on multimodal foundation models. We introduce a switching mechanism that leverages a mode token and phrase identification via a large language model to switch the embedding space based on the prediction target. To evaluate the DM2RM, we construct a novel dataset including real-world images collected from hundreds of building-scale environments and crowd-sourced instructions with referring expressions. The evaluation results show that the proposed DM2RM outperforms previous approaches in terms of standard metrics in image retrieval settings. Furthermore, we demonstrate the application of the DM2RM on a standardized real-world DSR platform including fetch-and-carry actions, where it achieves a task success rate of 82% despite the zero-shot transfer setting. Demonstration videos, code, and more materials are available at https://kkrr10.github.io/dm2rm/.

In this study, we propose a methodology to extract, index, and visualize ``climate change narratives'' (stories about the connection between causal and consequential events related to climate change). We use two natural language processing methods, BERT (Bidirectional Encoder Representations from Transformers) and causal extraction, to textually analyze newspaper articles on climate change to extract ``climate change narratives.'' The novelty of the methodology could extract and quantify the causal relationships assumed by the newspaper's writers. Looking at the extracted climate change narratives over time, we find that since 2018, an increasing number of narratives suggest the impact of the development of climate change policy discussion and the implementation of climate change-related policies on corporate behaviors, macroeconomics, and price dynamics. We also observed the recent emergence of narratives focusing on the linkages between climate change-related policies and monetary policy. Furthermore, there is a growing awareness of the negative impacts of natural disasters (e.g., abnormal weather and severe floods) related to climate change on economic activities, and this issue might be perceived as a new challenge for companies and governments. The methodology of this study is expected to be applied to a wide range of fields, as it can analyze causal relationships among various economic topics, including analysis of inflation expectation or monetary policy communication strategy.

Hydrocarbon dust is one of the dominant components of interstellar dust, which mainly consists of polycyclic aromatic hydrocarbons and aliphatic hydrocarbons. While hydrocarbon dust is thought to be processed in interstellar radiation fields or shocks, detailed processing mechanisms are not completely understood yet. We investigate the processing of hydrocarbon dust by analyzing the relation between the luminosities emitted by hydrocarbon dust and the total infrared luminosities $(L_{\mathrm{IR}})$ for 138 star-forming galaxies at redshift $z\ <\ 0.3$. Using near-infrared 2.5-5 $\mathrm{\mu m}$ spectra obtained with AKARI, we derived the luminosities of the aromatic hydrocarbon feature at 3.3 $\mathrm{\mu m}$ ($L_\mathrm{aromatic}$) and the aliphatic hydrocarbon feature at 3.4-3.6 $\mathrm{\mu m}$ ($L_\mathrm{aliphatic}$). We also derived $L_\mathrm{IR}$ and the radiation field strength by modeling the spectral energy distributions of the 138 galaxies with AKARI, WISE and IRAS photometry data. We find that galaxies with higher $L_\mathrm{IR}$ tend to exhibit lower $L_\mathrm{aliphatic}/L_\mathrm{aromatic}$ ratios. Furthermore, we find that there is an anti-correlation between $L_\mathrm{aliphatic}/L_\mathrm{aromatic}$ ratios and the radiation field strength, and also that the galaxies with low $L_\mathrm{aliphatic}/L_\mathrm{aromatic}$ ratios are dominated by merger galaxies. These results support that hydrocarbon dust is processed through photodissociation in strong radiation fields and/or shocks during merging processes of galaxies; the $L_\mathrm{aliphatic}/L_\mathrm{aromatic}$ ratio is likely to decrease in such harsh interstellar conditions since the aliphatic bonds are known to be chemically weaker than the aromatic bonds., Comment: 10 pages, 11 figures, accepted for publication in PASJ

The dark matter haloes associated with galaxies have hitherto established strong correlations within a range of observed parameters, known as scaling relations of dark matter haloes. The origin of these scaling relations still contains significant ambiguities and requires comprehensive exploration for complete understanding. Utilising the correlation between the concentration and mass of dark matter haloes inferred from cosmological $N$-body simulations based on the cold dark matter paradigm ($c$-$M$ relation), we derive theoretical scaling relations among other physical quantities such as the surface mass density, the maximum circular velocity, and the scale radius of the dark matter halo. By comparing theoretical and observed scaling relations at various mass scales, it is found that the scaling relations observed in dwarf galaxies and galaxies originate in the $c$-$M$ relation of the dark matter halo. We predict that this theoretical scaling relation is also established in galaxy clusters. Moreover, we propose a novel theoretical scaling relation that incorporates the effects of the cusp-to-core transition, which is supposed to occur in cold dark matter haloes. Our discussion concludes with the exploration of potential observational verification of the cusp-to-core transition process in dark matter haloes., Comment: 18 pages, 4 figures, accepted for publication in PASJ, the code to transit the properties of the NFW halo to ones of the Burkert halo is available on https://github.com/YukaKANEDA/cusp-to-core_transition_model

While reconstructing human poses in 3D from inexpensive sensors has advanced significantly in recent years, quantifying the dynamics of human motion, including the muscle-generated joint torques and external forces, remains a challenge. Prior attempts to estimate physics from reconstructed human poses have been hampered by a lack of datasets with high-quality pose and force data for a variety of movements. We present the AddBiomechanics Dataset 1.0, which includes physically accurate human dynamics of 273 human subjects, over 70 hours of motion and force plate data, totaling more than 24 million frames. To construct this dataset, novel analytical methods were required, which are also reported here. We propose a benchmark for estimating human dynamics from motion using this dataset, and present several baseline results. The AddBiomechanics Dataset is publicly available at https://addbiomechanics.org/download_data.html., Comment: 15 pages, 6 figures, 4 tables

Establishing an automatic evaluation metric that closely aligns with human judgments is essential for effectively developing image captioning models. Recent data-driven metrics have demonstrated a stronger correlation with human judgments than classic metrics such as CIDEr; however they lack sufficient capabilities to handle hallucinations and generalize across diverse images and texts partially because they compute scalar similarities merely using embeddings learned from tasks unrelated to image captioning evaluation. In this study, we propose Polos, a supervised automatic evaluation metric for image captioning models. Polos computes scores from multimodal inputs, using a parallel feature extraction mechanism that leverages embeddings trained through large-scale contrastive learning. To train Polos, we introduce Multimodal Metric Learning from Human Feedback (M$^2$LHF), a framework for developing metrics based on human feedback. We constructed the Polaris dataset, which comprises 131K human judgments from 550 evaluators, which is approximately ten times larger than standard datasets. Our approach achieved state-of-the-art performance on Composite, Flickr8K-Expert, Flickr8K-CF, PASCAL-50S, FOIL, and the Polaris dataset, thereby demonstrating its effectiveness and robustness., Comment: CVPR 2024

In photonic quantum applications, optical routers are required to handle single photons with low loss, high speed, and preservation of their quantum states. Single-photon routing with maintained polarization states is particularly important for utilizing them as qubits. Here, we demonstrate a polarization-maintaining electro-optic router compatible with single photons. Our custom electro-optic modulator is embedded in a configuration of a Mach-Zehnder interferometer, where each optical component achieves polarization-maintaining operation. We observe the performance of the router with 2-4% loss, 20 dB switching extinction ratio, 2.9 ns rise time, and $>$ 99% polarization process fidelity to an ideal identity operation.

Domestic service robots offer a solution to the increasing demand for daily care and support. A human-in-the-loop approach that combines automation and operator intervention is considered to be a realistic approach to their use in society. Therefore, we focus on the task of retrieving target objects from open-vocabulary user instructions in a human-in-the-loop setting, which we define as the learning-to-rank physical objects (LTRPO) task. For example, given the instruction "Please go to the dining room which has a round table. Pick up the bottle on it," the model is required to output a ranked list of target objects that the operator/user can select. In this paper, we propose MultiRankIt, which is a novel approach for the LTRPO task. MultiRankIt introduces the Crossmodal Noun Phrase Encoder to model the relationship between phrases that contain referring expressions and the target bounding box, and the Crossmodal Region Feature Encoder to model the relationship between the target object and multiple images of its surrounding contextual environment. Additionally, we built a new dataset for the LTRPO task that consists of instructions with complex referring expressions accompanied by real indoor environmental images that feature various target objects. We validated our model on the dataset and it outperformed the baseline method in terms of the mean reciprocal rank and recall@k. Furthermore, we conducted physical experiments in a setting where a domestic service robot retrieved everyday objects in a standardized domestic environment, based on users' instruction in a human--in--the--loop setting. The experimental results demonstrate that the success rate for object retrieval achieved 80%. Our code is available at https://github.com/keio-smilab23/MultiRankIt., Comment: Accepted for RAL 2023

This paper focuses on the DialFRED task, which is the task of embodied instruction following in a setting where an agent can actively ask questions about the task. To address this task, we propose DialMAT. DialMAT introduces Moment-based Adversarial Training, which incorporates adversarial perturbations into the latent space of language, image, and action. Additionally, it introduces a crossmodal parallel feature extraction mechanism that applies foundation models to both language and image. We evaluated our model using a dataset constructed from the DialFRED dataset and demonstrated superior performance compared to the baseline method in terms of success rate and path weighted success rate. The model secured the top position in the DialFRED Challenge, which took place at the CVPR 2023 Embodied AI workshop., Comment: Accepted for presentation at Fourth Annual Embodied AI Workshop at CVPR

Image captioning studies heavily rely on automatic evaluation metrics such as BLEU and METEOR. However, such n-gram-based metrics have been shown to correlate poorly with human evaluation, leading to the proposal of alternative metrics such as SPICE for English; however, no equivalent metrics have been established for other languages. Therefore, in this study, we propose an automatic evaluation metric called JaSPICE, which evaluates Japanese captions based on scene graphs. The proposed method generates a scene graph from dependencies and the predicate-argument structure, and extends the graph using synonyms. We conducted experiments employing 10 image captioning models trained on STAIR Captions and PFN-PIC and constructed the Shichimi dataset, which contains 103,170 human evaluations. The results showed that our metric outperformed the baseline metrics for the correlation coefficient with the human evaluation., Comment: Accepted at CoNLL 2023. Project page: https://yuiga.dev/jaspice/en

We experimentally test the error-disturbance uncertainty relation (EDR) in generalized, variable strength measurements of superconducting qubits on a NISQ processor. Making use of sequential weak measurements that keeps the initial signal state practically unchanged prior to the main measurement, we demonstrate that the Heisenberg EDR is violated, yet the Ozawa and Branciard EDRs are valid throughout the range of measurement strengths from no measurement to projection measurement. Our results verify that universal EDRs are valid even in a noisy quantum processor and will stimulate research on measurement-based quantum information and communication protocols using a NISQ processor., Comment: 6 pages, 4 figures. arXiv admin note: text overlap with arXiv:1308.5868

We introduce a neural-preconditioned iterative solver for Poisson equations with mixed boundary conditions. Typical Poisson discretizations yield large, ill-conditioned linear systems. Iterative solvers can be effective for these problems, but only when equipped with powerful preconditioners. Unfortunately, effective preconditioners like multigrid require costly setup phases that must be re-executed every time domain shapes or boundary conditions change, forming a severe bottleneck for problems with evolving boundaries. In contrast, we present a neural preconditioner trained to efficiently approximate the inverse of the discrete Laplacian in the presence of such changes. Our approach generalizes to domain shapes, boundary conditions, and grid sizes outside the training set. The key to our preconditioner's success is a novel, lightweight neural network architecture featuring spatially varying convolution kernels and supporting fast inference. We demonstrate that our solver outperforms state-of-the-art methods like algebraic multigrid as well as recently proposed neural preconditioners on challenging test cases arising from incompressible fluid simulations.

The carbon footprint associated with large language models (LLMs) is a significant concern, encompassing emissions from their training, inference, experimentation, and storage processes, including operational and embodied carbon emissions. An essential aspect is accurately estimating the carbon impact of emerging LLMs even before their training, which heavily relies on GPU usage. Existing studies have reported the carbon footprint of LLM training, but only one tool, mlco2, can predict the carbon footprint of new neural networks prior to physical training. However, mlco2 has several serious limitations. It cannot extend its estimation to dense or mixture-of-experts (MoE) LLMs, disregards critical architectural parameters, focuses solely on GPUs, and cannot model embodied carbon footprints. Addressing these gaps, we introduce \textit{\carb}, an end-to-end carbon footprint projection model designed for both dense and MoE LLMs. Compared to mlco2, \carb~significantly enhances the accuracy of carbon footprint estimations for various LLMs. The source code is released at \url{https://github.com/SotaroKaneda/MLCarbon}., Comment: 15 pages, 8 figures

We examine spatial variations of the C$^0$/CO abundance ratio ($X_{\mathrm{C/CO}}$) in the vicinity of the $\gamma$-ray supernova remnant W51C, based on [CI] ($^3P_1$-$^3P_0$), $^{12}$CO(1-0), and $^{13}$CO(1-0) observations with the ASTE and Nobeyama 45-m telescopes. We find that $X_{\mathrm{C/CO}}$ varies in a range of 0.02-0.16 (0.05 in median) inside the molecular clouds of $A_V>$100 mag, where photodissociation of CO by the interstellar UV is negligible. Furthermore, $X_{\mathrm{C/CO}}$ is locally enhanced up to by a factor of four near the W51C center, depending on the projected distance from the W51C center. In high-$A_V$ molecular clouds, $X_{\mathrm{C/CO}}$ is determined by the ratio of the cosmic-ray (CR) ionization rate to the H$_2$ density, and we find no clear spatial variation of the H$_2$ density against the projected distance. Hence, the high CR ionization rate may locally enhance $X_{\mathrm{C/CO}}$ near the W51C center. We also find that the observed spatial extent of the enhanced $X_{\mathrm{C/CO}}$ ($\sim$17 pc) is consistent with the diffusion distance of CRs with the energy of 100 MeV. The fact suggests that the low-energy CRs accelerated in W51C enhance $X_{\mathrm{C/CO}}$. The CR ionization rate at the $X_{\mathrm{C/CO}}$-enhanced cloud is estimated to be 3$\times$10$^{-16}$ s$^{-1}$ on the basis of time-dependent PDR simulations of $X_{\mathrm{C/CO}}$, the value of which is 30 times higher than that in the standard Galactic environment. These results demonstrate that [CI] is a powerful probe to investigate the interaction between CRs and the interstellar medium for a wide area in the vicinity of supernova remnants., Comment: 17 pages, 8 figures, accepted for publication in PASJ

We perform a systematic study of evolutionary stages and stellar masses of young stellar objects (YSOs) in the Large Magellanic Cloud (LMC) to investigate properties of star formation of the galaxy. There are 4825 sources in our YSO sample, which are constructed by combining the previous studies identifying YSOs in the LMC. Spectral energy distributions of the YSOs from optical to infrared wavelengths were fitted with a model consisting of stellar, polycyclic aromatic hydrocarbon and dust emissions. We utilize the stellar-to-dust luminosity ratios thus derived to study the evolutionary stages of the sources; younger YSOs are expected to show lower stellar-to-dust luminosity ratios. We find that most of the YSOs are associated with the interstellar gas across the galaxy, which are younger with more gases, suggesting that more recent star formation is associated with larger amounts of the interstellar medium (ISM). N157 shows a hint of higher stellar-to-dust luminosity ratios between active star-forming regions in the LMC, suggesting that recent star formation in N157 is possibly in later evolutionary stages. We also find that the stellar mass function tends to be bottom-heavy in supergiant shells (SGSs), indicating that gas compression by SGSs may be ineffective in compressing the ISM enough to trigger massive star formation. There is no significant difference in the stellar mass function between YSOs likely associated with the interface between colliding SGSs and those with a single SGS, suggesting that gas compression by collisions between SGSs may also be ineffective for massive star formation., Comment: 26 pages, 16 figures, accepted for publication in ApJ