搜索结果：Muse

Unified visual tokenization faces a fundamental trade-off between high-fidelity pixel reconstruction (spatial equivariance) and semantic abstraction (conceptual invariance). We attribute this conflict to Manifold Misalignment: naive joint optimization induces opposing gradients, creating a zero-sum game between reconstruction and perception. To address this, we propose MUSE, a framework based on Topological Orthogonality. By treating Structure as an orthogonal bridge, MUSE decouples optimization within Transformers: structural gradients refine attention topology, while semantic gradients update feature values. This turns destructive interference into Mutual Reinforcement. Experiments show that MUSE breaks the trade-off, achieving state-of-the-art generation quality (gFID 3.08) and surpassing its teacher InternViT-300M in linear probing (85.2\% vs. 82.5\%), demonstrating that structurally aligned reconstruction can enhance semantic perception. Code is available at https://github.com/PanqiYang1/MUSE.

Muse Spark is the latest large language model developed by Meta. In this report, we first present evaluations for catastrophic risk domains under Meta's Advanced AI Scaling Framework, along with the evidence that informed our launch decision. We then discuss additional considerations, such as Muse Spark's broader content safety and behavioral profile, that are relevant to overall safety but fall outside the catastrophic risk domains governed by the Framework. Our preparedness results covering Chemical and Biological, Cybersecurity, and Loss of Control risks assess Muse Spark's deployment within Meta AI as presenting acceptable levels of residual risks under our Advanced AI Scaling Framework. We conducted a broad set of evaluations targeting dual-use and high-risk capabilities across these catastrophic risk domains. Those evaluations identified elevated risks prior to mitigations, with Chemical and Biological capabilities assessed as likely reaching the "high risk" category under the Advanced AI Scaling Framework before safeguards were applied. We have implemented a multi-layered set of mitigations that address the identified risks, and Muse Spark demonstrates state-of-the-art refus

This paper introduces an innovative state estimator, MUSE (MUlti-sensor State Estimator), designed to enhance state estimation's accuracy and real-time performance in quadruped robot navigation. The proposed state estimator builds upon our previous work presented in [1]. It integrates data from a range of onboard sensors, including IMUs, encoders, cameras, and LiDARs, to deliver a comprehensive and reliable estimation of the robot's pose and motion, even in slippery scenarios. We tested MUSE on a Unitree Aliengo robot, successfully closing the locomotion control loop in difficult scenarios, including slippery and uneven terrain. Benchmarking against Pronto [2] and VILENS [3] showed 67.6% and 26.7% reductions in translational errors, respectively. Additionally, MUSE outperformed DLIO [4], a LiDAR-inertial odometry system in rotational errors and frequency, while the proprioceptive version of MUSE (P-MUSE) outperformed TSIF [5], with a 45.9% reduction in absolute trajectory error (ATE).

Accurate visual state estimation has been a central topic in robotics with a wide range of applications in robot navigation, autonomous driving, and autonomous flight. Recent advances in robot perception have led to significant improvements in the accuracy and robustness of state estimation, yet a fundamental challenge remains in how to quantify and calibrate its precision, i.e., how confident we are in an estimate and whether failures can be detected. This issue is particularly pronounced in visual-inertial odometry (VIO), where the heteroscedastic and multimodal nature of the problem makes uncertainty quantification especially difficult. This paper introduces MUSE (Multimodal Uncertainty Quantification of State Estimation), a novel real-time learning-based framework that leverages the strong and efficient sequential modeling capacity of Mamba to estimate localization uncertainty from multiple asynchronous sensor streams. Experiments on both public and in-house datasets demonstrate that MUSE achieves superior reliability and robustness compared to existing uncertainty quantification methods, and ablation studies justify the benefits of its key design choices.

We present Muse, a text-to-image Transformer model that achieves state-of-the-art image generation performance while being significantly more efficient than diffusion or autoregressive models. Muse is trained on a masked modeling task in discrete token space: given the text embedding extracted from a pre-trained large language model (LLM), Muse is trained to predict randomly masked image tokens. Compared to pixel-space diffusion models, such as Imagen and DALL-E 2, Muse is significantly more efficient due to the use of discrete tokens and requiring fewer sampling iterations; compared to autoregressive models, such as Parti, Muse is more efficient due to the use of parallel decoding. The use of a pre-trained LLM enables fine-grained language understanding, translating to high-fidelity image generation and the understanding of visual concepts such as objects, their spatial relationships, pose, cardinality etc. Our 900M parameter model achieves a new SOTA on CC3M, with an FID score of 6.06. The Muse 3B parameter model achieves an FID of 7.88 on zero-shot COCO evaluation, along with a CLIP score of 0.32. Muse also directly enables a number of image editing applications without the need

We introduce MUSE, a watermarking algorithm for tabular generative models. Previous approaches typically leverage DDIM invertibility to watermark tabular diffusion models, but tabular diffusion models exhibit significantly poorer invertibility compared to other modalities, compromising performance. Simultaneously, tabular diffusion models require substantially less computation than other modalities, enabling a multi-sample selection approach to tabular generative model watermarking. MUSE embeds watermarks by generating multiple candidate samples and selecting one based on a specialized scoring function, without relying on model invertibility. Our theoretical analysis establishes the relationship between watermark detectability, candidate count, and dataset size, allowing precise calibration of watermarking strength. Extensive experiments demonstrate that MUSE achieves state-of-the-art watermark detectability and robustness against various attacks while maintaining data quality, and remains compatible with any tabular generative model supporting repeated sampling, effectively addressing key challenges in tabular data watermarking. Specifically, it reduces the distortion rates on fid

We present aMUSEd, an open-source, lightweight masked image model (MIM) for text-to-image generation based on MUSE. With 10 percent of MUSE's parameters, aMUSEd is focused on fast image generation. We believe MIM is under-explored compared to latent diffusion, the prevailing approach for text-to-image generation. Compared to latent diffusion, MIM requires fewer inference steps and is more interpretable. Additionally, MIM can be fine-tuned to learn additional styles with only a single image. We hope to encourage further exploration of MIM by demonstrating its effectiveness on large-scale text-to-image generation and releasing reproducible training code. We also release checkpoints for two models which directly produce images at 256x256 and 512x512 resolutions.

Current conversational recommendation systems focus predominantly on text. However, real-world recommendation settings are generally multimodal, causing a significant gap between existing research and practical applications. To address this issue, we propose Muse, the first multimodal conversational recommendation dataset. Muse comprises 83,148 utterances from 7,000 conversations centered around the Clothing domain. Each conversation contains comprehensive multimodal interactions, rich elements, and natural dialogues. Data in Muse are automatically synthesized by a multi-agent framework powered by multimodal large language models (MLLMs). It innovatively derives user profiles from real-world scenarios rather than depending on manual design and history data for better scalability, and then it fulfills conversation simulation and optimization. Both human and LLM evaluations demonstrate the high quality of conversations in Muse. Additionally, fine-tuning experiments on three MLLMs demonstrate Muse's learnable patterns for recommendations and responses, confirming its value for multimodal conversational recommendation. Our dataset and codes are available at https://anonymous.4open.scie

The MUon Scattering Experiment (MUSE) was motivated by the proton radius puzzle arising from the discrepancy between muonic hydrogen spectroscopy and electron-proton measurements. The MUSE physics goals also include testing lepton universality, precisely measuring two-photon exchange contribution, and testing radiative corrections. MUSE addresses these physics goals through simultaneous measurement of high precision cross sections for electron-proton and muon-proton scattering using a mixed-species beam. The experiment will run at both positive and negative beam polarities. Measuring precise cross sections requires understanding both the incident beam energy and the radiative corrections. For this purpose, a lead-glass calorimeter was installed at the end of the beam line in the MUSE detector system. In this article we discuss the detector specifications, calibration and performance. We demonstrate that the detector performance is well reproduced by simulation, and meets experimental requirements.

The MuSe 2023 is a set of shared tasks addressing three different contemporary multimodal affect and sentiment analysis problems: In the Mimicked Emotions Sub-Challenge (MuSe-Mimic), participants predict three continuous emotion targets. This sub-challenge utilises the Hume-Vidmimic dataset comprising of user-generated videos. For the Cross-Cultural Humour Detection Sub-Challenge (MuSe-Humour), an extension of the Passau Spontaneous Football Coach Humour (Passau-SFCH) dataset is provided. Participants predict the presence of spontaneous humour in a cross-cultural setting. The Personalisation Sub-Challenge (MuSe-Personalisation) is based on the Ulm-Trier Social Stress Test (Ulm-TSST) dataset, featuring recordings of subjects in a stressed situation. Here, arousal and valence signals are to be predicted, whereas parts of the test labels are made available in order to facilitate personalisation. MuSe 2023 seeks to bring together a broad audience from different research communities such as audio-visual emotion recognition, natural language processing, signal processing, and health informatics. In this baseline paper, we introduce the datasets, sub-challenges, and provided feature sets.

The aim of this study is to better understand the connection between the Lyman $α$ rest-frame equivalent width (EW$_0$) and spectral properties as well as ultraviolet (UV) continuum morphology by obtaining reliable EW$_0$ histograms for a statistical sample of galaxies and by assessing the fraction of objects with large equivalent widths. We used integral field spectroscopy from MUSE combined with broad-band data from the Hubble Space Telescope (HST) to measure EW$_0$. We analysed the emission lines of $1920$ Lyman $α$ emitters (LAEs) detected in the full MUSE-Wide (one hour exposure time) and MUSE-Deep (ten hour exposure time) surveys and found UV continuum counterparts in archival HST data. We fitted the UV continuum photometric images using the Galfit software to gain morphological information on the rest-UV emission and fitted the spectra obtained from MUSE to determine the double peak fraction, asymmetry, full-width at half maximum, and flux of the Lyman $α$ line. The two surveys show different histograms of Lyman $α$ EW$_0$. In MUSE-Wide, $20\%$ of objects have EW$_0 > 240$ Å, while this fraction is only $11\%$ in MUSE-Deep and $\approx 16\%$ for the full sample. This incl

The Multi Unit Spectroscopic Explorer (MUSE) is a second-generation VLT panoramic integral-field spectrograph under preliminary design study. MUSE has a field of 1x1 arcmin**2 sampled at 0.2x0.2 arcsec**2 and is assisted by the VLT ground layer adaptive optics ESO facility using four laser guide stars. The simultaneous spectral range is 465-930 nm, at a resolution of R~3000. MUSE couples the discovery potential of a large imaging device to the measuring capabilities of a high-quality spectrograph, while taking advantage of the increased spatial resolution provided by adaptive optics. This makes MUSE a unique and tremendously powerful instrument for discovering and characterizing objects that lie beyond the reach of even the deepest imaging surveys. MUSE has also a high spatial resolution mode with 7.5x7.5 arcsec**2 field of view sampled at 25 milli-arcsec. In this mode MUSE should be able to obtain diffraction limited data-cubes in the 600-930 nm wavelength range. Although the MUSE design has been optimized for the study of galaxy formation and evolution, it has a wide range of possible applications; e.g. monitoring of outer planets atmosphere, environment of young stellar objects,

Multimodal Sentiment Analysis in Real-life Media (MuSe) 2020 is a Challenge-based Workshop focusing on the tasks of sentiment recognition, as well as emotion-target engagement and trustworthiness detection by means of more comprehensively integrating the audio-visual and language modalities. The purpose of MuSe 2020 is to bring together communities from different disciplines; mainly, the audio-visual emotion recognition community (signal-based), and the sentiment analysis community (symbol-based). We present three distinct sub-challenges: MuSe-Wild, which focuses on continuous emotion (arousal and valence) prediction; MuSe-Topic, in which participants recognise domain-specific topics as the target of 3-class (low, medium, high) emotions; and MuSe-Trust, in which the novel aspect of trustworthiness is to be predicted. In this paper, we provide detailed information on MuSe-CaR, the first of its kind in-the-wild database, which is utilised for the challenge, as well as the state-of-the-art features and modelling approaches applied. For each sub-challenge, a competitive baseline for participants is set; namely, on test we report for MuSe-Wild a combined (valence and arousal) CCC of .25

The Multimodal Sentiment Analysis Challenge (MuSe) 2022 is dedicated to multimodal sentiment and emotion recognition. For this year's challenge, we feature three datasets: (i) the Passau Spontaneous Football Coach Humor (Passau-SFCH) dataset that contains audio-visual recordings of German football coaches, labelled for the presence of humour; (ii) the Hume-Reaction dataset in which reactions of individuals to emotional stimuli have been annotated with respect to seven emotional expression intensities, and (iii) the Ulm-Trier Social Stress Test (Ulm-TSST) dataset comprising of audio-visual data labelled with continuous emotion values (arousal and valence) of people in stressful dispositions. Using the introduced datasets, MuSe 2022 2022 addresses three contemporary affective computing problems: in the Humor Detection Sub-Challenge (MuSe-Humor), spontaneous humour has to be recognised; in the Emotional Reactions Sub-Challenge (MuSe-Reaction), seven fine-grained `in-the-wild' emotions have to be predicted; and in the Emotional Stress Sub-Challenge (MuSe-Stress), a continuous prediction of stressed emotion values is featured. The challenge is designed to attract different research comm

MUSE is a novel slide-free imaging technique for histological examination of tissues that can serve as an alternative to traditional histology. In order to bridge the gap between MUSE and traditional histology, we aim to convert MUSE images to resemble authentic hematoxylin- and eosin-stained (H&E) images. We evaluated four models: a non-machine-learning-based color-mapping unmixing-based tool, CycleGAN, DualGAN, and GANILLA. CycleGAN and GANILLA provided visually compelling results that appropriately transferred H&E style and preserved MUSE content. Based on training an automated critic on real and generated H&E images, we determined that CycleGAN demonstrated the best performance. We have also found that MUSE color inversion may be a necessary step for accurate modality conversion to H&E. We believe that our MUSE-to-H&E model can help improve adoption of novel slide-free methods by bridging a perceptual gap between MUSE imaging and traditional histology.

Leo T is the lowest mass galaxy known to contain neutral gas and to show signs of recent star formation, which makes it a valuable laboratory for studying the nature of gas and star formation at the limits of where galaxies are found to have rejuvenating episodes of star formation. Here we discuss a novel study of Leo T that uses data from the MUSE integral field spectrograph and photometric data from HST. The high sensitivity of MUSE allowed us to increase the number of Leo T stars observed spectroscopically from 19 to 75. We studied the age and metallicity of these stars and identified two populations, all consistent with similar metallicity of [Fe/H] $\sim$ -1.5 dex, suggesting that a large fraction of metals were ejected. Within the young population, we discovered three emission line Be stars, supporting the conclusion that rapidly rotating massive stars are common in metal-poor environments. We find differences in the dynamics of young and old stars, with the young population having a velocity dispersion consistent with the kinematics of the cold component of the neutral gas. This finding directly links the recent star formation in Leo T with the cold component of the neutral

Here we describe a new study of the SNRs and SNR candidates in nearby face-on spiral galaxy M83, based primarily on MUSE integral field spectroscopy. Our revised catalog of SNR candidates in M83 has 366 objects, 81 of which are reported here for the first time. Of these, 229 lie within the MUSE observation region, 160 of which have spectra with [SII]:Halpha ratios exceeding 0.4, the value generally accepted as confirmation that an emission nebula is shock-heated. Combined with 51 SNR candidates outside the MUSE region with high [SII]:Halpha ratios, there are 211 spectroscopically-confirmed SNRs in M83, the largest number of confirmed SNRs in any external galaxy. MUSE's combination of relatively high spectral resolution and broad wavelength coverage has allowed us to explore two other properties of SNRs that could serve as the basis of future SNR searches. Specifically, most of the objects identified as SNRs on the basis of [SII]:Halpha ratios exhibit more velocity broadening and lower ratios of [SIII]:[SII] emission than HII regions. A search for nebulae with the very broad emission lines expected from young, rapidly expanding remnants revealed none, except for the previously ident

We present a novel approach to deriving stellar labels for stars observed in MUSE fields making use of data-driven machine learning methods. Taking advantage of the comparable spectral properties (resolution, wavelength coverage) of the LAMOST and MUSE instruments, we adopt the Data-Driven Payne (DD-Payne) model used on LAMOST observations and apply it to stars observed in MUSE fields. Remarkably, in spite of instrumental differences, according to the cross-validation of 27 LAMOST-MUSE common stars, we are able to determine stellar labels with precision better than 75K in $T_{\rm eff}$, 0.15 dex in $\log g$, and 0.1 dex in abundances of [Fe/H], [Mg/Fe], [Si/Fe], [Ti/Fe], [C/Fe], [Ni/Fe] and [Cr/Fe] for current MUSE observations over a parameter range of 3800<$T_{\rm eff}$<7000 K, -1.5<[Fe/H]<0.5 dex. To date, MUSE has been used to target 13,000 fields across the southern sky since it was first commissioned six years ago and it is unique in its ability to study dense star fields such as globular clusters or the Milky Way bulge. Our method will enable the automated determination of stellar parameters for all stars in these fields. Additionally, it opens the door for appli

Protoplanetary disks contain structures such as gaps, rings, and spirals, which are thought to be produced by the interaction between the disk and embedded protoplanets. However, only a few planet candidates are found orbiting within protoplanetary disks, and most of them are being challenged as having been confused with disk features. We aim to discover more proto-planetary candidates with MUSE, with a secondary aim of improving the high-resolution spectral differential imaging (HRSDI) technique by analyzing the instrumental residuals of MUSE. We analyzed MUSE observations of five young stars and applied the HRSDI technique to perform high-contrast imaging. With a 30 min integration time, MUSE can reach 5$σ$ detection limits in apparent H$α$ line flux down to 10$^{-14}$ and 10$^{-15}$ erg s$^{-1}$ cm$^{-2}$ at 0.075" and 0.25", respectively. In addition to PDS 70 b and c, we did not detect any clear accretion signatures in PDS 70, J1850-3147, and V1094 Sco down to 0.1". MUSE avoids the small sample statistics problem by measuring the noise characteristics in the spatial direction at multiple wavelengths. We detected two asymmetric atomic jets in HD 163296. The HRSDI technique when