Search

While the sources of the diffuse astrophysical neutrino flux detected by the IceCube Neutrino Observatory are still largely unknown, one of the promising methods used towards understanding this is investigating the potential temporal and spatial correlations between neutrino alerts and the electromagnetic radiation from blazars. We report on the multiwavelength target-of-opportunity observations of the blazar B3 2247+381, taken in response to an IceCube multiplet alert for a cluster of muon neutrino events compatible with the source location between May 20, 2022 and November 10, 2022. B3 2247+381 was not detected with VERITAS during this time period. The source was found to be in a low-flux state in the optical, ultraviolet and gamma-ray bands for the time interval corresponding to the neutrino event, but was detected in the hard X-ray band with NuSTAR during this period. We find the multiwavelength spectral energy distribution is well described using a simple one-zone leptonic synchrotron self-Compton radiation model. Moreover, assuming the neutrinos originate from hadronic processes within the jet, the neutrino flux would be accompanied by a photon flux from the cascade emission, and the integrated photon flux required in such a case would significantly exceed the total multiwavelength fluxes and the VERITAS upper limits presented here. The lack of flaring activity observed with VERITAS, combined with the low multiwavelength flux levels, and given the significance of the neutrino excess is at 3$\sigma$ level (uncorrected for trials), makes B3 2247+381 an unlikely source of the IceCube multiplet. We conclude that the neutrino excess is likely a background fluctuation., Comment: 26 pages, 5 figures. Accepted for publication in the Astrophysical Journal (ApJ)

We present the first asteroseismic analysis of the K3\,V planet host HD~219134, based on four consecutive nights of radial velocities collected with the Keck Planet Finder. We applied Gold deconvolution to the power spectrum to disentangle modes from sidelobes in the spectral window, and extracted 25 mode frequencies with spherical degrees $0\leq\ell\leq3$. We derive the fundamental properties using five different evolutionary-modeling pipelines and report a mass of 0.763 $\pm$ 0.020 (stat) $\pm$ 0.007 (sys) M$_\odot$, a radius of 0.748 $\pm$ 0.007 (stat) $\pm$ 0.002 (sys) R$_\odot$, and an age of 10.151 $\pm$ 1.520 (stat) $\pm$ 0.810 (sys) Gyr. Compared to the interferometric radius 0.783 $\pm$ 0.005~R$_\odot$, the asteroseismic radius is 4\% smaller at the 4-$\sigma$ level -- a discrepancy not easily explained by known interferometric systematics, modeling assumptions on atmospheric boundary conditions and mixing lengths, magnetic fields, or tidal heating. HD~219134 is the first main-sequence star cooler than 5000~K with an asteroseismic age estimate and will serve as a critical calibration point for stellar spin-down relations. We show that existing calibrated prescriptions for angular momentum loss, incorporating weakened magnetic braking with asteroseismically constrained stellar parameters, accurately reproduce the observed rotation period. Additionally, we revised the masses and radii of the super-Earths in the system, which support their having Earth-like compositions. Finally, we confirm that the oscillation amplitude in radial velocity scales as $(L/M)^{1.5}$ in K dwarfs, in contrast to the $(L/M)^{0.7}$ relation observed in G dwarfs. These findings provide significant insights into the structure and angular momentum loss of K-type stars., Comment: 23 pages, 11 figures. submitted to AAS Journals. Comments welcome

Guided diffusion-model generation is a promising direction for customizing the generation process of a pre-trained diffusion-model to address the specific downstream tasks. Existing guided diffusion models either rely on training of the guidance model with pre-collected datasets or require the objective functions to be differentiable. However, for most real-world tasks, the offline datasets are often unavailable, and their objective functions are often not differentiable, such as image generation with human preferences, molecular generation for drug discovery, and material design. Thus, we need an $\textbf{online}$ algorithm capable of collecting data during runtime and supporting a $\textbf{black-box}$ objective function. Moreover, the $\textbf{query efficiency}$ of the algorithm is also critical because the objective evaluation of the query is often expensive in the real-world scenarios. In this work, we propose a novel and simple algorithm, $\textbf{Fast Direct}$, for query-efficient online black-box target generation. Our Fast Direct builds a pseudo-target on the data manifold to update the noise sequence of the diffusion model with a universal direction, which is promising to perform query-efficient guided generation. Extensive experiments on twelve high-resolution ($\small {1024 \times 1024}$) image target generation tasks and six 3D-molecule target generation tasks show $\textbf{6}\times$ up to $\textbf{10}\times$ query efficiency improvement and $\textbf{11}\times$ up to $\textbf{44}\times$ query efficiency improvement, respectively. Our implementation is publicly available at: https://github.com/kimyong95/guide-stable-diffusion/tree/fast-direct

Acoustic guidelines for neonatal intensive care units (NICUs) aim to protect vulnerable neonates from noise-induced physiological harm. However, the lack of recognised international standards for measuring neonatal soundscapes has led to inconsistencies in instrumentation and microphone placement in existing literature, raising concerns about the relevance and effectiveness of these guidelines. This study addresses these gaps through long-term acoustic measurements in an operational NICU and a high-dependency ward. We investigate the influence of microphone positioning, bed placement, and ward layout on the assessment of NICU soundscapes. Beyond traditional A-weighted decibel metrics, this study evaluates C-weighted metrics for low-frequency noise, the occurrence of tonal sounds (e.g., alarms), and transient loud events known to disrupt neonates' sleep. Using linear mixed-effects models with aligned ranks transformation ANOVA (LME-ART-ANOVA), our results reveal significant differences in measured noise levels based on microphone placement, highlighting the importance of capturing sound as perceived directly at the neonate's ears. Additionally, bed position and ward layout significantly impact noise exposure, with a NICU bed position consistently exhibiting the highest sound levels across all (psycho)acoustic metrics. These findings support the adoption of binaural measurements along with the integration of additional (psycho)acoustic metrics, such as tonality and transient event occurrence rates, to reliably characterise the neonatal auditory experience., Comment: Accepted manuscript submitted to Building and Environment

The Teledyne COSMOS-66 is a next-generation CMOS camera designed for astronomical imaging, featuring a large-format sensor ($8120 \times 8120$ pixels, each $10 \mu m$), high quantum efficiency, high frame rates, and a correlated multi-sampling mode that achieves low read noise. We performed a suite of bench-top and on-sky tests to characterize this sensor and analyze its suitability for use in astronomical instruments. This paper presents measurements of linearity, conversion gain, read noise, dark current, quantum efficiency, image lag, and crosstalk. We found that the sensor exhibits nonlinear response below 5% of saturation. This nonlinearity is plausibly attributable to the trapping of electrons in each pixel. We developed and implemented a pixel-by-pixel nonlinearity correction, enabling accurate photometric measurements across the dynamic range. After implementing this correction, operating in the correlated multi-sampling mode, the sensor achieved an effective read noise of $2.9 e^-$ and dark current of $0.12 e^-/pix/s$ at $-25^\circ C$. The quantum efficiency exceeded 50% from 250 nm to 800 nm, peaking at 89% at 600 nm. We observed significant optical crosstalk between the pixels, likely caused by photoelectron diffusion. To demonstrate the sensor's astronomical performance, we mounted it on the WINTER 1m telescope at Palomar Observatory. These tests confirmed that the linearity calibration enables accurate stellar photometry and validated our measured noise levels. Overall, the COSMOS-66 delivers similar noise performance to large-format CCDs, with higher frame rates and relaxed cooling requirements. If pixel design improvements are made to mitigate the nonlinearity and crosstalk, then the camera may combine the advantages of low-noise CMOS image sensors with the integration simplicity of large-format CCDs, broadening its utility to a host of astronomical science cases., Comment: 45 pages, 16 figures, submitted to JATIS

Large language models (LLMs) are vulnerable to universal jailbreaks-prompting strategies that systematically bypass model safeguards and enable users to carry out harmful processes that require many model interactions, like manufacturing illegal substances at scale. To defend against these attacks, we introduce Constitutional Classifiers: safeguards trained on synthetic data, generated by prompting LLMs with natural language rules (i.e., a constitution) specifying permitted and restricted content. In over 3,000 estimated hours of red teaming, no red teamer found a universal jailbreak that could extract information from an early classifier-guarded LLM at a similar level of detail to an unguarded model across most target queries. On automated evaluations, enhanced classifiers demonstrated robust defense against held-out domain-specific jailbreaks. These classifiers also maintain deployment viability, with an absolute 0.38% increase in production-traffic refusals and a 23.7% inference overhead. Our work demonstrates that defending against universal jailbreaks while maintaining practical deployment viability is tractable.

Reinforcement Learning algorithms aim to learn optimal control strategies through iterative interactions with an environment. A critical element in this process is the experience replay buffer, which stores past experiences, allowing the algorithm to learn from a diverse range of interactions rather than just the most recent ones. This buffer is especially essential in dynamic environments with limited experiences. However, efficiently selecting high-value experiences to accelerate training remains a challenge. Drawing inspiration from the role of reward prediction errors (RPEs) in biological systems, where they are essential for adaptive behaviour and learning, we introduce Reward Predictive Error Prioritised Experience Replay (RPE-PER). This novel approach prioritises experiences in the buffer based on RPEs. Our method employs a critic network, EMCN, that predicts rewards in addition to the Q-values produced by standard critic networks. The discrepancy between these predicted and actual rewards is computed as RPE and utilised as a signal for experience prioritisation. Experimental evaluations across various continuous control tasks demonstrate RPE-PER's effectiveness in enhancing the learning speed and performance of off-policy actor-critic algorithms compared to baseline approaches., Comment: This paper was accepted for presentation at the 2024 Australasian Conference on Robotics and Automation (ACRA 2024). It consists of 10 pages, including four figures and two tables

Systematic literature review is essential for evidence-based medicine, requiring comprehensive analysis of clinical trial publications. However, the application of artificial intelligence (AI) models for medical literature mining has been limited by insufficient training and evaluation across broad therapeutic areas and diverse tasks. Here, we present LEADS, an AI foundation model for study search, screening, and data extraction from medical literature. The model is trained on 633,759 instruction data points in LEADSInstruct, curated from 21,335 systematic reviews, 453,625 clinical trial publications, and 27,015 clinical trial registries. We showed that LEADS demonstrates consistent improvements over four cutting-edge generic large language models (LLMs) on six tasks. Furthermore, LEADS enhances expert workflows by providing supportive references following expert requests, streamlining processes while maintaining high-quality results. A study with 16 clinicians and medical researchers from 14 different institutions revealed that experts collaborating with LEADS achieved a recall of 0.81 compared to 0.77 experts working alone in study selection, with a time savings of 22.6%. In data extraction tasks, experts using LEADS achieved an accuracy of 0.85 versus 0.80 without using LEADS, alongside a 26.9% time savings. These findings highlight the potential of specialized medical literature foundation models to outperform generic models, delivering significant quality and efficiency benefits when integrated into expert workflows for medical literature mining.

Large language models (LLMs) have proven to be very capable, but access to the best models currently rely on inference providers which introduces trust challenges -- how can we be sure that the provider is using the model configuration they claim? We propose TOPLOC, a novel method for verifiable inference that addresses this problem. TOPLOC leverages a compact locality sensitive hashing mechanism for intermediate activations which can detect unauthorized modifications to models, prompts, or precision with 100% accuracy, achieving no false positives or negatives in our empirical evaluations. Our approach is robust across diverse hardware configurations, GPU types, and algebraic reorderings, which allows for validation speeds significantly faster than the original inference. By introducing a polynomial encoding scheme, TOPLOC minimizes memory overhead of the generated commits by $1000\times$, requiring only 258 bytes of storage per 32 new tokens compared to the 262KB requirement of storing the token embeddings directly for Llama-3.1-8B-Instruct. Our method empowers users to verify LLM inference computations efficiently, fostering greater trust and transparency in open ecosystems and lays a foundation for decentralized and verifiable AI services., Comment: 18 pages, 13 tables, 5 figures

The existence of a quantum critical point (QCP) and fluctuations around it are believed to be important for understanding the phase diagram in unconventional superconductors such as cuprates, iron pnictides, and heavy fermion superconductors. However, the QCP is usually buried deep within the superconducting dome and is difficult to investigate. The connection between quantum critical fluctuations and superconductivity remains an outstanding problem in condensed matter. Here combining both electrical transport and Nernst experiments, we explicitly demonstrate the onset of superconductivity at an unconventional QCP in gate-tuned monolayer tungsten ditelluride (WTe2), with features incompatible with the conventional Bardeen-Cooper-Schrieffer (BCS) scenario. The results lead to a novel superconducting phase diagram that is distinguished from other known superconductors. Two distinct gate-tuned quantum phase transitions are observed at the ends of the superconducting dome. We find that quantum fluctuations around the QCP of the underdoped regime are essential for understanding how the monolayer superconductivity is established. The unconventional phase diagram we report here illustrates a previously unknown relation between superconductivity and QCP.

Evaluating corporate sustainability performance is essential to drive sustainable business practices, amid the need for a more sustainable economy. However, this is hindered by the complexity and volume of corporate sustainability data (i.e. sustainability disclosures), not least by the effectiveness of the NLP tools used to analyse them. To this end, we identify three primary challenges - immateriality, complexity, and subjectivity, that exacerbate the difficulty of extracting insights from sustainability disclosures. To address these issues, we introduce ESGSenticNet, a publicly available knowledge base for sustainability analysis. ESGSenticNet is constructed from a neurosymbolic framework that integrates specialised concept parsing, GPT-4o inference, and semi-supervised label propagation, together with a hierarchical taxonomy. This approach culminates in a structured knowledge base of 44k knowledge triplets - ('halve carbon emission', supports, 'emissions control'), for effective sustainability analysis. Experiments indicate that ESGSenticNet, when deployed as a lexical method, more effectively captures relevant and actionable sustainability information from sustainability disclosures compared to state of the art baselines. Besides capturing a high number of unique ESG topic terms, ESGSenticNet outperforms baselines on the ESG relatedness and ESG action orientation of these terms by 26% and 31% respectively. These metrics describe the extent to which topic terms are related to ESG, and depict an action toward ESG. Moreover, when deployed as a lexical method, ESGSenticNet does not require any training, possessing a key advantage in its simplicity for non-technical stakeholders.

Many real-world problems, such as airfoil design, involve optimizing a black-box expensive objective function over complex structured input space (e.g., discrete space or non-Euclidean space). By mapping the complex structured input space into a latent space of dozens of variables, a two-stage procedure labeled as generative model based optimization (GMO) in this paper, shows promise in solving such problems. However, the latent dimension of GMO is hard to determine, which may trigger the conflicting issue between desirable solution accuracy and convergence rate. To address the above issue, we propose a multi-form GMO approach, namely generative multi-form optimization (GMFoO), which conducts optimization over multiple latent spaces simultaneously to complement each other. More specifically, we devise a generative model which promotes positive correlation between latent spaces to facilitate effective knowledge transfer in GMFoO. And further, by using Bayesian optimization (BO) as the optimizer, we propose two strategies to exchange information between these latent spaces continuously. Experimental results are presented on airfoil and corbel design problems and an area maximization problem as well to demonstrate that our proposed GMFoO converges to better designs on a limited computational budget.

Bayesian optimization (BO) is well known to be sample-efficient for solving black-box problems. However, the BO algorithms can sometimes get stuck in suboptimal solutions even with plenty of samples. Intrinsically, such suboptimal problem of BO can attribute to the poor surrogate accuracy of the trained Gaussian process (GP), particularly that in the regions where the optimal solutions locate. Hence, we propose to build multiple GP models instead of a single GP surrogate to complement each other and thus resolving the suboptimal problem of BO. Nevertheless, according to the bias-variance tradeoff equation, the individual prediction errors can increase when increasing the diversity of models, which may lead to even worse overall surrogate accuracy. On the other hand, based on the theory of Rademacher complexity, it has been proved that exploiting the agreement of models on unlabeled information can help to reduce the complexity of the hypothesis space, and therefore achieving the required surrogate accuracy with fewer samples. Such value of model agreement has been extensively demonstrated for co-training style algorithms to boost model accuracy with a small portion of samples. Inspired by the above, we propose a novel BO algorithm labeled as co-learning BO (CLBO), which exploits both model diversity and agreement on unlabeled information to improve the overall surrogate accuracy with limited samples, and therefore achieving more efficient global optimization. Through tests on five numerical toy problems and three engineering benchmarks, the effectiveness of proposed CLBO has been well demonstrated.

Assuming Galactic cosmic rays originate in supernovae and the winds of massive stars, starburst galaxies should produce very-high-energy (VHE; E$>$100 GeV) gamma-ray emission via the interaction of their copious quantities of cosmic rays with the large reservoirs of dense gas within the galaxies. Such VHE emission was detected by VERITAS from the starburst galaxy M 82 in 2008-09. An extensive, multi-year campaign followed these initial observations, yielding a total of 254 h of good quality VERITAS data on M 82. Leveraging modern analysis techniques and the larger exposure, these VERITAS data show a more statistically significant VHE signal ($\sim$6.5 standard deviations ($\sigma$)). The corresponding photon spectrum is well fit by a power law ($\Gamma = 2.3 \pm 0.3_{stat} \pm0.2_{sys}$) and the observed integral flux is F($>$450 GeV) = $(3.2 \pm0.6_{stat} \pm 0.6_{sys}) \times 10^{-13}~\mathrm{cm^{-2}~s}^{-1}$, or $\sim$0.4\% of the Crab Nebula flux above the same energy threshold. The improved VERITAS measurements, when combined with various multi-wavelength data, enable modeling of the underlying emission and transport processes. A purely leptonic scenario is found to be a poor representation of the gamma-ray spectral energy distribution (SED). A lepto-hadronic scenario with cosmic rays following a power-law spectrum in momentum (index $s\simeq 2.25$), and with significant bremsstrahlung below $1$~GeV, provides a good match to the observed SED. The synchrotron emission from the secondary electrons indicates that efficient non-radiative losses of cosmic-ray electrons may be related to advective escape from the starburst core., Comment: 15 pages, 7 figures; Accepted for the publication in The Astrophysical Journal (ApJ)

This paper explores the development of a multimodal sentiment analysis model that integrates text, audio, and visual data to enhance sentiment classification. The goal is to improve emotion detection by capturing the complex interactions between these modalities, thereby enabling more accurate and nuanced sentiment interpretation. The study evaluates three feature fusion strategies -- late stage fusion, early stage fusion, and multi-headed attention -- within a transformer-based architecture. Experiments were conducted using the CMU-MOSEI dataset, which includes synchronized text, audio, and visual inputs labeled with sentiment scores. Results show that early stage fusion significantly outperforms late stage fusion, achieving an accuracy of 71.87\%, while the multi-headed attention approach offers marginal improvement, reaching 72.39\%. The findings suggest that integrating modalities early in the process enhances sentiment classification, while attention mechanisms may have limited impact within the current framework. Future work will focus on refining feature fusion techniques, incorporating temporal data, and exploring dynamic feature weighting to further improve model performance.

Deep learning models trained on finite data lack a complete understanding of the physical world. On the other hand, physics-informed neural networks (PINNs) are infused with such knowledge through the incorporation of mathematically expressible laws of nature into their training loss function. By complying with physical laws, PINNs provide advantages over purely data-driven models in limited-data regimes. This feature has propelled them to the forefront of scientific machine learning, a domain characterized by scarce and costly data. However, the vision of accurate physics-informed learning comes with significant challenges. This review examines PINNs for the first time in terms of model optimization and generalization, shedding light on the need for new algorithmic advances to overcome issues pertaining to the training speed, precision, and generalizability of today's PINN models. Of particular interest are the gradient-free methods of neuroevolution for optimizing the uniquely complex loss landscapes arising in PINN training. Methods synergizing gradient descent and neuroevolution for discovering bespoke neural architectures and balancing multiple conflicting terms in physics-informed learning objectives are positioned as important avenues for future research. Yet another exciting track is to cast neuroevolution as a meta-learner of generalizable PINN models., Comment: 20 pages, 8 figures, 1 table

Ultrafast magnetization dynamics driven by ultrashort pump lasers is typically explained by changes in electronic populations and scattering pathways of excited conduction electrons. This conventional approach overlooks the fundamental role of quantum mechanical selection rules, governing transitions from core states to the conduction band, that forms the key method of the probing step in these experiments. By employing fully ab initio time-dependent density functional theory, we reveal that these selection rules profoundly influence the interpretation of ultrafast spin dynamics at specific probe energies. Our analysis for hcp Co and fcc Ni at the M edge demonstrates that the transient dynamics, as revealed in pump-probe experiments, arise from a complex interplay of optical excitations of the M shell. Taking into account the selection rules and conduction electron spin flips, this leads to highly energy-dependent dynamics. These findings address longstanding discrepancies in experimental TMOKE measurements and show that only through meticulous consideration of matrix elements at the probe stage, can one ensure that magnetization dynamics is revealed in its true nature, instead of being muddled by artifacts arising from the choice of probe energy.

On the main sequence, the asteroseismic small frequency separation $\delta\nu_{02}$ between radial and quadrupole p-modes is customarily interpreted to be a direct diagnostic of internal structure. Such an interpretation is based on a well-known integral estimator relating $\delta\nu_{02}$ to a radially-averaged sound-speed gradient. However, this estimator fails, catastrophically, when evaluated on structural models of red giants: their small separations must therefore be interpreted differently. We derive a single expression which both reduces to the classical estimator when applied to main-sequence stellar models, yet reproduces the qualitative features of the small separation for stellar models of very evolved red giants. This expression indicates that the small separations of red giants scale primarily with their global seismic properties as $\delta\nu_{02} \propto \Delta\nu^2/\nu_\mathrm{max}$, rather than being in any way sensitive to their internal structure. Departures from this asymptotic behaviour, during the transition from the main-sequence to red giant regimes, have been recently reported in open-cluster Christensen-Dalsgaard (C-D) diagrams from K2 mission data. Investigating them in detail, we demonstrate that they occur when the convective envelope boundary passes a specific acoustic distance -- roughly a third of a wavelength at $\nu_\mathrm{max}$ -- from the centre of the star, at which point radial modes become maximally sensitive to the position of the boundary. The shape of the corresponding features on $\epsilon_p$ and C-D (or $r_{02}$) diagrams may be useful in constraining the nature of convective boundary mixing, in the context of undershooting beneath a convective envelope., Comment: 14 pages, 8 figures. Accepted for publication in The Astrophysical Journal

As AI systems become more capable, widely deployed, and increasingly autonomous in critical areas such as cybersecurity, biological research, and healthcare, ensuring their safety and alignment with human values is paramount. Machine unlearning -- the ability to selectively forget or suppress specific types of knowledge -- has shown promise for privacy and data removal tasks, which has been the primary focus of existing research. More recently, its potential application to AI safety has gained attention. In this paper, we identify key limitations that prevent unlearning from serving as a comprehensive solution for AI safety, particularly in managing dual-use knowledge in sensitive domains like cybersecurity and chemical, biological, radiological, and nuclear (CBRN) safety. In these contexts, information can be both beneficial and harmful, and models may combine seemingly harmless information for harmful purposes -- unlearning this information could strongly affect beneficial uses. We provide an overview of inherent constraints and open problems, including the broader side effects of unlearning dangerous knowledge, as well as previously unexplored tensions between unlearning and existing safety mechanisms. Finally, we investigate challenges related to evaluation, robustness, and the preservation of safety features during unlearning. By mapping these limitations and open challenges, we aim to guide future research toward realistic applications of unlearning within a broader AI safety framework, acknowledging its limitations and highlighting areas where alternative approaches may be required.

Many models are proposed in the literature on biomedical event extraction(BEE). Some of them use the shortest dependency path(SDP) information to represent the argument classification task. There is an issue with this representation since even missing one word from the dependency parsing graph may totally change the final prediction. To this end, the full adjacency matrix of the dependency graph is used to embed individual tokens using a graph convolutional network(GCN). An ablation study is also done to show the effect of the dependency graph on the overall performance. The results show a significant improvement when dependency graph information is used. The proposed model slightly outperforms state-of-the-art models on BEE over different datasets., Comment: 6 figures, 4 tables

Existing asteroseismic rotational measurements assume that stars rotate around a single axis. However, tidal torques from misaligned companions, or their possible engulfment, may bring the rotational axis of a star's envelope out of alignment with its core, breaking azimuthal symmetry. I derive perturbative expressions for asteroseismic signatures of such hitherto unexamined rotational configurations, under the ``shellular approximation'' of constant rotation rates on radially stratified mass shells. In the aligned case, the distribution of power between multiplet components is determined by the inclination of the rotational axis; radial differential misalignment causes this to vary from multiplet to multiplet. I examine in particular detail the phenomenology of gravitoacoustic mixed modes as seen in evolved sub- and red giants, where near-resonance avoided crossings may break geometrical degeneracies. Upon applying the revised asteroseismic observational methodology that results from this theoretical discussion to revisit Kepler-56 -- a red giant with a misaligned planetary system -- I find that its core and envelope rotate around different rotational axes. While the rotational axis of its core is indeed misaligned from the orbit normal of its transiting planets (consistently with earlier studies), its envelope's rotational axis is close to lying in the sky plane, and may well be aligned with them. More detailed asteroseismic modelling, and spectroscopic follow-up, will be required to fully elucidate the full spin-orbit geometry of the Kepler-56 system, and potentially discriminate between hypotheses for how it formed., Comment: 19 pages, 7 figures, 2 tables. Accepted for publication in The Astrophysical Journal

Large Language Models (LLMs) are emerging as powerful tools in healthcare, particularly for complex, domain-specific tasks. This study describes the development and evaluation of the PErioperative AI CHatbot (PEACH), a secure LLM-based system integrated with local perioperative guidelines to support preoperative clinical decision-making. PEACH was embedded with 35 institutional perioperative protocols in the secure Claude 3.5 Sonet LLM framework within Pair Chat (developed by Singapore Government) and tested in a silent deployment with real-world data. Accuracy, safety, and usability were assessed. Deviations and hallucinations were categorized based on potential harm, and user feedback was evaluated using the Technology Acceptance Model (TAM). Updates were made after the initial silent deployment to amend one protocol. In 240 real-world clinical iterations, PEACH achieved a first-generation accuracy of 97.5% (78/80) and an overall accuracy of 96.7% (232/240) across three iterations. The updated PEACH demonstrated improved accuracy of 97.9% (235/240), with a statistically significant difference from the null hypothesis of 95% accuracy (p = 0.018, 95% CI: 0.952-0.991). Minimal hallucinations and deviations were observed (both 1/240 and 2/240, respectively). Clinicians reported that PEACH expedited decisions in 95% of cases, and inter-rater reliability ranged from kappa 0.772-0.893 within PEACH and 0.610-0.784 among attendings. PEACH is an accurate, adaptable tool that enhances consistency and efficiency in perioperative decision-making. Future research should explore its scalability across specialties and its impact on clinical outcomes., Comment: 21 pages, 3 figures, 1 graphical abstract

In the era of increasing privacy concerns and demand for personalized experiences, traditional Reinforcement Learning with Human Feedback (RLHF) frameworks face significant challenges due to their reliance on centralized data. We introduce Federated Reinforcement Learning with Human Feedback (FedRLHF), a novel framework that decentralizes the RLHF process. FedRLHF enables collaborative policy learning across multiple clients without necessitating the sharing of raw data or human feedback, thereby ensuring robust privacy preservation. Leveraging federated reinforcement learning, each client integrates human feedback locally into their reward functions and updates their policies through personalized RLHF processes. We establish rigorous theoretical foundations for FedRLHF, providing convergence guarantees, and deriving sample complexity bounds that scale efficiently with the number of clients. Empirical evaluations on the MovieLens and IMDb datasets demonstrate that FedRLHF not only preserves user privacy but also achieves performance on par with centralized RLHF, while enhancing personalization across diverse client environments., Comment: Accepted to AAMAS 2025. This preprint represents the full version of the paper, including all proofs, experimental details, and additional discussions

Incorporating multi-modal features as side information has recently become a trend in recommender systems. To elucidate user-item preferences, recent studies focus on fusing modalities via concatenation, element-wise sum, or attention mechanisms. Despite having notable success, existing approaches do not account for the modality-specific noise encapsulated within each modality. As a result, direct fusion of modalities will lead to the amplification of cross-modality noise. Moreover, the variation of noise that is unique within each modality results in noise alleviation and fusion being more challenging. In this work, we propose a new Spectrum-based Modality Representation (SMORE) fusion graph recommender that aims to capture both uni-modal and fusion preferences while simultaneously suppressing modality noise. Specifically, SMORE projects the multi-modal features into the frequency domain and leverages the spectral space for fusion. To reduce dynamic contamination that is unique to each modality, we introduce a filter to attenuate and suppress the modality noise adaptively while capturing the universal modality patterns effectively. Furthermore, we explore the item latent structures by designing a new multi-modal graph learning module to capture associative semantic correlations and universal fusion patterns among similar items. Finally, we formulate a new modality-aware preference module, which infuses behavioral features and balances the uni- and multi-modal features for precise preference modeling. This empowers SMORE with the ability to infer both user modality-specific and fusion preferences more accurately. Experiments on three real-world datasets show the efficacy of our proposed model. The source code for this work has been made publicly available at https://github.com/kennethorq/SMORE., Comment: Accepted to ACM Web Search and Data Mining (WSDM) 2025

As AI systems are integrated into social networks, there are AI safety concerns that AI-generated content may dominate the web, e.g. in popularity or impact on beliefs. To understand such questions, this paper proposes the Digital Ecosystem of Beliefs (Digico), the first evolutionary framework for controlled experimentation with multi-population interactions in simulated social networks. The framework models a population of agents which change their messaging strategies due to evolutionary updates following a Universal Darwinism approach, interact via messages, influence each other's beliefs through dynamics based on a contagion model, and maintain their beliefs through cognitive Lamarckian inheritance. Initial experiments with an abstract implementation of Digico show that: a) when AIs have faster messaging, evolution, and more influence in the recommendation algorithm, they get 80% to 95% of the views, depending on the size of the influence benefit; b) AIs designed for propaganda can typically convince 50% of humans to adopt extreme beliefs, and up to 85% when agents believe only a limited number of channels; c) a penalty for content that violates agents' beliefs reduces propaganda effectiveness by up to 8%. We further discuss implications for control (e.g. legislation) and Digico as a means of studying evolutionary principles.

We develop the framework of Indirect Query Bayesian Optimization (IQBO), a new class of Bayesian optimization problems where the integrated feedback is given via a conditional expectation of the unknown function $f$ to be optimized. The underlying conditional distribution can be unknown and learned from data. The goal is to find the global optimum of $f$ by adaptively querying and observing in the space transformed by the conditional distribution. This is motivated by real-world applications where one cannot access direct feedback due to privacy, hardware or computational constraints. We propose the Conditional Max-Value Entropy Search (CMES) acquisition function to address this novel setting, and propose a hierarchical search algorithm to address the multi-resolution setting and improve the computational efficiency. We show regret bounds for our proposed methods and demonstrate the effectiveness of our approaches on simulated optimization tasks., Comment: Preliminary work. Under review

We introduce a version of probabilistic Kleene algebra with angelic nondeterminism and a corresponding class of automata. Our approach implements semantics via distributions over multisets in order to overcome theoretical barriers arising from the lack of a distributive law between the powerset and Giry monads. We produce a full Kleene theorem and a coalgebraic theory, as well as both operational and denotational semantics and equational reasoning principles.

3D Gaussian Splatting (3DGS) has demonstrated impressive Novel View Synthesis (NVS) results in a real-time rendering manner. During training, it relies heavily on the average magnitude of view-space positional gradients to grow Gaussians to reduce rendering loss. However, this average operation smooths the positional gradients from different viewpoints and rendering errors from different pixels, hindering the growth and optimization of many defective Gaussians. This leads to strong spurious artifacts in some areas. To address this problem, we propose Hard Gaussian Splatting, dubbed HGS, which considers multi-view significant positional gradients and rendering errors to grow hard Gaussians that fill the gaps of classical Gaussian Splatting on 3D scenes, thus achieving superior NVS results. In detail, we present positional gradient driven HGS, which leverages multi-view significant positional gradients to uncover hard Gaussians. Moreover, we propose rendering error guided HGS, which identifies noticeable pixel rendering errors and potentially over-large Gaussians to jointly mine hard Gaussians. By growing and optimizing these hard Gaussians, our method helps to resolve blurring and needle-like artifacts. Experiments on various datasets demonstrate that our method achieves state-of-the-art rendering quality while maintaining real-time efficiency.

This paper presents a novel approach for synthesizing control barrier functions (CBFs) from high relative degree safety constraints: Rectified CBFs (ReCBFs). We begin by discussing the limitations of existing High-Order CBF approaches and how these can be overcome by incorporating an activation function into the CBF construction. We then provide a comparative analysis of our approach with related methods, such as CBF backstepping. Our results are presented first for safety constraints with relative degree two, then for mixed-input relative degree constraints, and finally for higher relative degrees. The theoretical developments are illustrated through simple running examples and an aircraft control problem., Comment: 6 pages, submitted to L-CSS/ACC 2025

Large language models (LLMs) have shown promising potential for next Point-of-Interest (POI) recommendation. However, existing methods only perform direct zero-shot prompting, leading to ineffective extraction of user preferences, insufficient injection of collaborative signals, and a lack of user privacy protection. As such, we propose a novel Multitask Reflective Large Language Model for Privacy-preserving Next POI Recommendation (MRP-LLM), aiming to exploit LLMs for better next POI recommendation while preserving user privacy. Specifically, the Multitask Reflective Preference Extraction Module first utilizes LLMs to distill each user's fine-grained (i.e., categorical, temporal, and spatial) preferences into a knowledge base (KB). The Neighbor Preference Retrieval Module retrieves and summarizes the preferences of similar users from the KB to obtain collaborative signals. Subsequently, aggregating the user's preferences with those of similar users, the Multitask Next POI Recommendation Module generates the next POI recommendations via multitask prompting. Meanwhile, during data collection, a Privacy Transmission Module is specifically devised to preserve sensitive POI data. Extensive experiments on three real-world datasets demonstrate the efficacy of our proposed MRP-LLM in providing more accurate next POI recommendations with user privacy preserved., Comment: 14 pages, 7 figures

In this report, we introduce INTELLECT-1, the first 10 billion parameter language model collaboratively trained across the globe, demonstrating that large-scale model training is no longer confined to large corporations but can be achieved through a distributed, community-driven approach. INTELLECT-1 was trained on 1 trillion tokens using up to 14 concurrent nodes distributed across 3 continents, with contributions from 30 independent compute providers dynamically joining and leaving the training process, while maintaining 83-96% compute utilization and 36.2-41.4% model FLOPS utilization. We leverage PRIME, our scalable distributed training framework designed for fault-tolerant, high-performance training on unreliable, globally distributed nodes. Key innovations in PRIME include the ElasticDeviceMesh, which manages dynamic global process groups for fault-tolerant communication across the internet and local process groups for communication within a node, live checkpoint recovery kernels, and a hybrid DiLoCo-FSDP2 implementation. Using PRIME with DiLoCo and our custom int8 all-reduce, we achieve a 400x reduction in communication bandwidth compared to traditional data-parallel training settings while delivering comparable performance. These results demonstrate the feasibility and promise of training frontier foundation models in a decentralized network of global GPU resources., Comment: 19 pages, 6 figures

It has been almost 40 years since the proposal of the idea that Hawking radiation of black holes does not lead to a complete evaporation but rather a "remnant" state. Though traditionally viewed with great criticisms especially from the high energy physics community, in recent years, various approaches have demonstrated that black hole remnants remain a viable possibility. In this review, which is primarily aimed as an introduction to the subject, we will discuss some possible routes to forming remnants and their respective properties and challenges., Comment: Invited contribution for the book "The Black Hole Information Paradox" (Eds. Ali Akil and Cosimo Bambi, Springer Singapore, expected in 2025); ver.2: improved and expanded some discussions; fixed typos and grammatical errors

The increasing use of Generative Artificial Intelligence (GAI) tools in education highlights the need to understand their influence on individuals' thinking processes and agency. This research explored 20 university students' interaction with GAI during programming. Participants completed surveys, recorded their screens during an hour-long programming session, and reflected on their GAI use. To analyse the data, we developed an AI-augmented thinking coding scheme with four dimensions: Question Formulation, Solution Development, Solution Analysis and Evaluation, and Solution Refinement. Participants were categorised into human-led and AI-led groups based on the time ratio of human-generating source code versus copying source code from GAI. T-tests indicated that the human-led group spent significantly more time on Solution Development and Solution Refinement than the AI-led group. Sequential pattern mining revealed distinct patterns of the two groups: the human-led group often refined GAI outputs, while the AI-led group frequently relied on direct answers from GAI. Correlation analyses found that positive attitudes towards AI, critical thinking, and programming self-efficacy positively correlated with Question Formulation; critical thinking was positively related to Solution Refinement; and programming self-efficacy was negatively associated with Solution Analysis and Evaluation. This study enhances understanding of the thinking process in GAI-supported programming.

The rapid development of AI models has led to a growing emphasis on enhancing their capabilities for complex input data such as videos. While large-scale video datasets have been introduced to support this growth, the unique challenges of reducing redundancies in video \textbf{sets} have not been explored. Compared to image datasets or individual videos, video \textbf{sets} have a two-layer nested structure, where the outer layer is the collection of individual videos, and the inner layer contains the correlations among frame-level data points to provide temporal information. Video \textbf{sets} have two dimensions of redundancies: within-sample and inter-sample redundancies. Existing methods like key frame selection, dataset pruning or dataset distillation are not addressing the unique challenge of video sets since they aimed at reducing redundancies in only one of the dimensions. In this work, we are the first to study Video Set Distillation, which synthesizes optimized video data by jointly addressing within-sample and inter-sample redundancies. Our Information Diversification and Temporal Densification (IDTD) method jointly reduces redundancies across both dimensions. This is achieved through a Feature Pool and Feature Selectors mechanism to preserve inter-sample diversity, alongside a Temporal Fusor that maintains temporal information density within synthesized videos. Our method achieves state-of-the-art results in Video Dataset Distillation, paving the way for more effective redundancy reduction and efficient AI model training on video datasets.

In this paper, we present a systematic effort to design, evaluate, and implement a realistic conversational recommender system (CRS). The objective of our system is to allow users to input free-form text to request recommendations, and then receive a list of relevant and diverse items. While previous work on synthetic queries augments large language models (LLMs) with 1-3 tools, we argue that a more extensive toolbox is necessary to effectively handle real user requests. As such, we propose a novel approach that equips LLMs with over 10 tools, providing them access to the internal knowledge base and API calls used in production. We evaluate our model on a dataset of real users and show that it generates relevant, novel, and diverse recommendations compared to vanilla LLMs. Furthermore, we conduct ablation studies to demonstrate the effectiveness of using the full range of tools in our toolbox. We share our designs and lessons learned from deploying the system for internal alpha release. Our contribution is the addressing of all four key aspects of a practicable CRS: (1) real user requests, (2) augmenting LLMs with a wide variety of tools, (3) extensive evaluation, and (4) deployment insights.

We introduce TPCNet, a neural network predictor that combines Convolutional and Transformer architectures with Positional encodings, for neutral atomic hydrogen (HI) spectral analysis. Trained on synthetic datasets, our models predict cold neutral gas fraction ($f_\text{CNM}$) and HI opacity correction factor ($R_\text{HI}$) from emission spectra based on the learned relationships between the desired output parameters and observables (optically-thin column density and peak brightness). As a follow-up to Murray et al. (2020)'s shallow Convolutional Neural Network (CNN), we construct deep CNN models and compare them to TPCNet models. TPCNet outperforms deep CNNs, achieving a 10% average increase in testing accuracy, algorithmic (training) stability, and convergence speed. Our findings highlight the robustness of the proposed model with sinusoidal positional encoding applied directly to the spectral input, addressing perturbations in training dataset shuffling and convolutional network weight initializations. Higher spectral resolutions with increased spectral channels offer advantages, albeit with increased training time. Diverse synthetic datasets enhance model performance and generalization, as demonstrated by producing $f_\text{CNM}$ and $R_\text{HI}$ values consistent with evaluation ground truths. Applications of TPCNet to observed emission data reveal strong agreement between the predictions and Gaussian decomposition-based estimates (from emission and absorption surveys), emphasizing its potential in HI spectral analysis., Comment: This paper has 27 pages, 27 figures and two tables. The work has been accepted for publication in Monthly Notices of the Royal Astronomical Society Main Journal

Few-Shot Learning (FSL) is a challenging task, which aims to recognize novel classes with few examples. Pre-training based methods effectively tackle the problem by pre-training a feature extractor and then performing class prediction via a cosine classifier with mean-based prototypes. Nevertheless, due to the data scarcity, the mean-based prototypes are usually biased. In this paper, we attempt to diminish the prototype bias by regarding it as a prototype optimization problem. To this end, we propose a novel prototype optimization framework to rectify prototypes, i.e., introducing a meta-optimizer to optimize prototypes. Although the existing meta-optimizers can also be adapted to our framework, they all overlook a crucial gradient bias issue, i.e., the mean-based gradient estimation is also biased on sparse data. To address this issue, in this paper, we regard the gradient and its flow as meta-knowledge and then propose a novel Neural Ordinary Differential Equation (ODE)-based meta-optimizer to optimize prototypes, called MetaNODE. Although MetaNODE has shown superior performance, it suffers from a huge computational burden. To further improve its computation efficiency, we conduct a detailed analysis on MetaNODE and then design an effective and efficient MetaNODE extension version (called E2MetaNODE). It consists of two novel modules: E2GradNet and E2Solver, which aim to estimate accurate gradient flows and solve optimal prototypes in an effective and efficient manner, respectively. Extensive experiments show that 1) our methods achieve superior performance over previous FSL methods and 2) our E2MetaNODE significantly improves computation efficiency meanwhile without performance degradation., Comment: An extended version of metanode: prototype optimization as a neural ode for few-shot learning. arXiv admin note: text overlap with arXiv:2103.14341

In the evolutionary computing community, the remarkable language-handling capabilities and reasoning power of large language models (LLMs) have significantly enhanced the functionality of evolutionary algorithms (EAs), enabling them to tackle optimization problems involving structured language or program code. Although this field is still in its early stages, its impressive potential has led to the development of various LLM-based EAs. To effectively evaluate the performance and practical applicability of these LLM-based EAs, benchmarks with real-world relevance are essential. In this paper, we focus on LLM-based recommender systems (RSs) and introduce a benchmark problem set, named RSBench, specifically designed to assess the performance of LLM-based EAs in recommendation prompt optimization. RSBench emphasizes session-based recommendations, aiming to discover a set of Pareto optimal prompts that guide the recommendation process, providing accurate, diverse, and fair recommendations. We develop three LLM-based EAs based on established EA frameworks and experimentally evaluate their performance using RSBench. Our study offers valuable insights into the application of EAs in LLM-based RSs. Additionally, we explore key components that may influence the overall performance of the RS, providing meaningful guidance for future research on the development of LLM-based EAs in RSs.

We study the problem of bounding the posterior distribution of discrete probabilistic programs with unbounded support, loops, and conditioning. Loops pose the main difficulty in this setting: even if exact Bayesian inference is possible, the state of the art requires user-provided loop invariant templates. By contrast, we aim to find guaranteed bounds, which sandwich the true distribution. They are fully automated, applicable to more programs and provide more provable guarantees than approximate sampling-based inference. Since lower bounds can be obtained by unrolling loops, the main challenge is upper bounds, and we attack it in two ways. The first is called residual mass semantics, which is a flat bound based on the residual probability mass of a loop. The approach is simple, efficient, and has provable guarantees. The main novelty of our work is the second approach, called geometric bound semantics. It operates on a novel family of distributions, called eventually geometric distributions (EGDs), and can bound the distribution of loops with a new form of loop invariants called contraction invariants. The invariant synthesis problem reduces to a system of polynomial inequality constraints, which is a decidable problem with automated solvers. If a solution exists, it yields an exponentially decreasing bound on the whole distribution, and can therefore bound moments and tail asymptotics as well, not just probabilities as in the first approach. Both semantics enjoy desirable theoretical properties. In particular, we prove soundness and convergence, i.e. the bounds converge to the exact posterior as loops are unrolled further. On the practical side, we describe Diabolo, a fully-automated implementation of both semantics, and evaluate them on a variety of benchmarks from the literature, demonstrating their general applicability and the utility of the resulting bounds., Comment: Full version of the POPL 2025 article, including proofs and other supplementary material

Experiments on quantum materials have uncovered many interesting quantum phases ranging from superconductivity to a variety of topological quantum matter including the recently observed fractional quantum anomalous Hall insulators. The findings have come in parallel with the development of approaches to probe the rich excitations inherent in such systems. In contrast to observing electrically charged excitations, the detection of charge-neutral electronic excitations in condensed matter remains difficult, though they are essential to understanding a large class of strongly correlated phases. Low-energy neutral excitations are especially important in characterizing unconventional phases featuring electron fractionalization, such as quantum spin liquids, spin ices, and insulators with neutral Fermi surfaces. In this perspective, we discuss searches for neutral fermionic, bosonic, or anyonic excitations in unconventional insulators, highlighting theoretical and experimental progress in probing excitonic insulators, new quantum spin liquid candidates and emergent correlated insulators based on two-dimensional layered crystals and moir\'e materials. We outline the promises and challenges in probing and utilizing quantum insulators, and discuss exciting new opportunities for future advancements offered by ideas rooted in next-generation quantum materials, devices, and experimental schemes., Comment: 23 pages, 4 figures; A Perspective/Review Article published in Nature

Background: Type I X-Ray bursts (XRBs) are energetic stellar explosions that occur on the surface of a neutron star in an accreting binary system with a low-mass H/He-rich companion. The rate of the $^{34}$Ar($\alpha,p$)$^{37}$K reaction may influence features of the light curve that results from the underlying thermonuclear runaway, as shown in recent XRB stellar modelling studies. Purpose: In order to reduce the uncertainty of the rate of this reaction, properties of resonances in the compound nucleus $^{38}$Ca, such as resonance energies, spins, and particle widths, must be well constrained. Method: This work discusses a study of resonances in the $^{38}$Ca compound nucleus produced in the $^{34}$Ar($\alpha,p$) reaction. The experiment was performed at the National Superconducting Cyclotron Laboratory, with the ReA3 facility by measuring proton scattering using an unstable $^{37}$K beam. The kinematics were designed specifically to identify and characterize resonances in the Gamow energy window for the temperature regime relevant to XRBs. Results: The spins and proton widths of newly identified and previously known states in $^{38}$Ca in the energy region of interest for the $^{34}$Ar($\alpha,p$)$^{37}$K reaction have been constrained through an R-Matrix analysis of the scattering data. Conclusions: Using these constraints, a newly estimated rate is applied to an XRB model built using Modules for Experiments in Stellar Astrophysics (MESA), to examine its impact on observables, including the light curve. It is found that the newly determined reaction rate does not substantially affect the features of the light curve.

Molecular docking is a crucial step in drug development, which enables the virtual screening of compound libraries to identify potential ligands that target proteins of interest. However, the computational complexity of traditional docking models increases as the size of the compound library increases. Recently, deep learning algorithms can provide data-driven research and development models to increase the speed of the docking process. Unfortunately, few models can achieve superior screening performance compared to that of traditional models. Therefore, a novel deep learning-based docking approach named Dockformer is introduced in this study. Dockformer leverages multimodal information to capture the geometric topology and structural knowledge of molecules and can directly generate binding conformations with the corresponding confidence measures in an end-to-end manner. The experimental results show that Dockformer achieves success rates of 90.53% and 82.71% on the PDBbind core set and PoseBusters benchmarks, respectively, and more than a 100-fold increase in the inference process speed, outperforming almost all state-of-the-art docking methods. In addition, the ability of Dockformer to identify the main protease inhibitors of coronaviruses is demonstrated in a real-world virtual screening scenario. Considering its high docking accuracy and screening efficiency, Dockformer can be regarded as a powerful and robust tool in the field of drug design., Comment: 15 pages, 10 figures

Despite remarkable achievements in deep learning across various domains, its inherent vulnerability to adversarial examples still remains a critical concern for practical deployment. Adversarial training has emerged as one of the most effective defensive techniques for improving model robustness against such malicious inputs. However, existing adversarial training schemes often lead to limited generalization ability against underlying adversaries with diversity due to their overreliance on a point-by-point augmentation strategy by mapping each clean example to its adversarial counterpart during training. In addition, adversarial examples can induce significant disruptions in the statistical information w.r.t. the target model, thereby introducing substantial uncertainty and challenges to modeling the distribution of adversarial examples. To circumvent these issues, in this paper, we propose a novel uncertainty-aware distributional adversarial training method, which enforces adversary modeling by leveraging both the statistical information of adversarial examples and its corresponding uncertainty estimation, with the goal of augmenting the diversity of adversaries. Considering the potentially negative impact induced by aligning adversaries to misclassified clean examples, we also refine the alignment reference based on the statistical proximity to clean examples during adversarial training, thereby reframing adversarial training within a distribution-to-distribution matching framework interacted between the clean and adversarial domains. Furthermore, we design an introspective gradient alignment approach via matching input gradients between these domains without introducing external models. Extensive experiments across four benchmark datasets and various network architectures demonstrate that our approach achieves state-of-the-art adversarial robustness and maintains natural performance.

Due to privacy and security concerns, recent advancements in group fairness advocate for model training regardless of demographic information. However, most methods still require prior knowledge of demographics. In this study, we explore the potential for achieving fairness without compromising its utility when no prior demographics are provided to the training set, namely \emph{harmless Rawlsian fairness}. We ascertain that such a fairness requirement with no prior demographic information essential promotes training losses to exhibit a Dirac delta distribution. To this end, we propose a simple but effective method named VFair to minimize the variance of training losses inside the optimal set of empirical losses. This problem is then optimized by a tailored dynamic update approach that operates in both loss and gradient dimensions, directing the model towards relatively fairer solutions while preserving its intact utility. Our experimental findings indicate that regression tasks, which are relatively unexplored from literature, can achieve significant fairness improvement through VFair regardless of any prior, whereas classification tasks usually do not because of their quantized utility measurements. The implementation of our method is publicly available at \url{https://github.com/wxqpxw/VFair}., Comment: Accepted as a Poster in Neurips 2024