Search

The circumgalactic medium (CGM) around massive galaxies plays a crucial role in regulating star formation and feedback. Using the CAMELS simulation suite, we develop emulators for the X-ray surface brightness profile and the X-ray luminosity--stellar mass scaling relation to investigate how stellar and AGN feedback shape the X-ray properties of the hot CGM. Our analysis shows that at CGM scales ($10^{12} \lesssim M_{\rm halo}/M_\odot \lesssim 10^{13}$, $10\lesssim r/{\rm kpc} \lesssim 400$), stellar feedback more significantly impacts the X-ray properties than AGN feedback within the parameters studied. Comparing the emulators to recent eROSITA All-Sky Survey observations, it was found that stronger feedback than currently implemented in the IllustrisTNG, SIMBA, and Astrid simulations is required to match observed CGM properties. However, adopting these enhanced feedback parameters causes deviations in the stellar-mass-halo-mass relations from observational constraints below the group mass scale. This tension suggests possible unaccounted systematics in X-ray CGM observations or inadequacies in the feedback models of cosmological simulations., Comment: 13 pages, 6 figures, submitted to ApJ

We propose a new approach to visual perception for relative localization of agents within large-scale swarms of UAVs. Inspired by biological perception utilized by schools of sardines, swarms of bees, and other large groups of animals capable of moving in a decentralized yet coherent manner, our method does not rely on detecting individual neighbors by each agent and estimating their relative position, but rather we propose to regress a neighbor density over distance. This allows for a more accurate distance estimation as well as better scalability with respect to the number of neighbors. Additionally, a novel swarm control algorithm is proposed to make it compatible with the new relative localization method. We provide a thorough evaluation of the presented methods and demonstrate that the regressing approach to distance estimation is more robust to varying relative pose of the targets and that it is suitable to be used as the main source of relative localization for swarm stabilization.

Solving systems of partial differential equations (PDEs) is a fundamental task in computational science, traditionally addressed by numerical solvers. Recent advancements have introduced neural operators and physics-informed neural networks (PINNs) to tackle PDEs, achieving reduced computational costs at the expense of solution quality and accuracy. Gaussian processes (GPs) have also been applied to linear PDEs, with the advantage of always yielding precise solutions. In this work, we propose Boundary Ehrenpreis-Palamodov Gaussian Processes (B-EPGPs), a novel framework for constructing GP priors that satisfy both general systems of linear PDEs with constant coefficients and linear boundary conditions. We explicitly construct GP priors for representative PDE systems with practical boundary conditions. Formal proofs of correctness are provided and empirical results demonstrating significant accuracy improvements over state-of-the-art neural operator approaches., Comment: 25 pages, 19 figures. Code available at $\href{https://github.com/Jimmy000207/Boundary-EPGP}{\text{this https URL}}$. The paper and all ancillary files are released under CC-BY

The growing demand for customized visual content has led to the rise of personalized text-to-image (T2I) diffusion models. Despite their remarkable potential, they pose significant privacy risk when misused for malicious purposes. In this paper, we propose a novel and efficient adversarial attack method, Concept Protection by Selective Attention Manipulation (CoPSAM) which targets only the cross-attention layers of a T2I diffusion model. For this purpose, we carefully construct an imperceptible noise to be added to clean samples to get their adversarial counterparts. This is obtained during the fine-tuning process by maximizing the discrepancy between the corresponding cross-attention maps of the user-specific token and the class-specific token, respectively. Experimental validation on a subset of CelebA-HQ face images dataset demonstrates that our approach outperforms existing methods. Besides this, our method presents two important advantages derived from the qualitative evaluation: (i) we obtain better protection results for lower noise levels than our competitors; and (ii) we protect the content from unauthorized use thereby protecting the individual's identity from potential misuse., Comment: Accepted at Safe Generative AI Workshop (NeurIPS 2024)

This paper introduces a novel Laplacian matrix aiming to enable the construction of spectral convolutional networks and to extend the signal processing applications for directed graphs. Our proposal is inspired by a Haar-like transformation and produces a Hermitian matrix which is not only in one-to-one relation with the adjacency matrix, preserving both direction and weight information, but also enjoys desirable additional properties like scaling robustness, sensitivity, continuity, and directionality. We take a theoretical standpoint and support the conformity of our approach with the spectral graph theory. Then, we address two use-cases: graph learning (by introducing HaarNet, a spectral graph convolutional network built with our Haar-Laplacian) and graph signal processing. We show that our approach gives better results in applications like weight prediction and denoising on directed graphs., Comment: 30 pages, 9 figures, 4 tables

In this work, we study the alignment (BrainScore) of large language models (LLMs) fine-tuned for moral reasoning on behavioral data and/or brain data of humans performing the same task. We also explore if fine-tuning several LLMs on the fMRI data of humans performing moral reasoning can improve the BrainScore. We fine-tune several LLMs (BERT, RoBERTa, DeBERTa) on moral reasoning behavioral data from the ETHICS benchmark [Hendrycks et al., 2020], on the moral reasoning fMRI data from Koster-Hale et al. [2013], or on both. We study both the accuracy on the ETHICS benchmark and the BrainScores between model activations and fMRI data. While larger models generally performed better on both metrics, BrainScores did not significantly improve after fine-tuning., Comment: Accepted to the 2nd Workshop on Unifying Representations in Neural Models (UniReps) at NeurIPS 2024

Accurate lower-limb joint kinematic estimation is critical for applications such as patient monitoring, rehabilitation, and exoskeleton control. While previous studies have employed wearable sensor-based deep learning (DL) models for estimating joint kinematics, these methods often require extensive new datasets to adapt to unseen gait patterns. Meanwhile, researchers in computer vision have advanced human pose estimation models, which are easy to deploy and capable of real-time inference. However, such models are infeasible in scenarios where cameras cannot be used. To address these limitations, we propose a computer vision-based DL adaptation framework for real-time joint kinematic estimation. This framework requires only a small dataset (i.e., 1-2 gait cycles) and does not depend on professional motion capture setups. Using transfer learning, we adapted our temporal convolutional network (TCN) to stiff knee gait data, allowing the model to further reduce root mean square error by 9.7% and 19.9% compared to a TCN trained on only able-bodied and stiff knee datasets, respectively. Our framework demonstrates a potential for smartphone camera-trained DL models to estimate real-time joint kinematics across novel users in clinical populations with applications in wearable robots.

Patches of catalyst imprinted on supporting walls induce motion of the fluid around them once they are supplied with the chemical species (``fuel'') that are converted by the catalytic chemical reaction. While the functioning of such chemically active micropumps is conceptually well understood, an in-depth characterization of the induced hydrodynamic flow, and in particular of its possible dependences on parameters such as material properties of the patch and the wall or the geometry of the experimental cell, remains elusive. By using a simple model for the chemical activity of a patch imprinted on a planar wall, we determine analytically the induced hydrodynamic flow in a Newtonian solution that occupies the half space above the wall supporting the patch. This can be seen as an approximation for an experimental-cell geometry with a height much larger than its diameter, the latter, in turn, being much larger than the size of the patch.) The general flow is a linear superposition of a surface-driven and a bulk-driven component; they have different topologies and, generically, each one dominates in distinct regions, with the surface-driven flow being most relevant at small heights above the wall. The surface-driven flows exhibit a somewhat unexpectedly rich behavior, including qualitative changes in the topology of the flow, as a function of the contrast in surface-chemistry (osmotic slip coefficient) between the patch and the support wall. The results are expected to provide guidance for the interpretation of the drift of tracers by the ambient flow, which is the observable usually studied in experimental investigations of chemically active micropumps.

Many practical prediction algorithms represent inputs in Euclidean space and replace the discrete 0/1 classification loss with a real-valued surrogate loss, effectively reducing classification tasks to stochastic optimization. In this paper, we investigate the expressivity of such reductions in terms of key resources, including dimension and the role of randomness. We establish bounds on the minimum Euclidean dimension $D$ needed to reduce a concept class with VC dimension $d$ to a Stochastic Convex Optimization (SCO) problem in $\mathbb{R}^D$, formally addressing the intuitive interpretation of the VC dimension as the number of parameters needed to learn the class. To achieve this, we develop a generalization of the Borsuk-Ulam Theorem that combines the classical topological approach with convexity considerations. Perhaps surprisingly, we show that, in some cases, the number of parameters $D$ must be exponentially larger than the VC dimension $d$, even if the reduction is only slightly non-trivial. We also present natural classification tasks that can be represented in much smaller dimensions by leveraging randomness, as seen in techniques like random initialization. This result resolves an open question posed by Kamath, Montasser, and Srebro (COLT 2020). Our findings introduce new variants of \emph{dimension complexity} (also known as \emph{sign-rank}), a well-studied parameter in learning and complexity theory. Specifically, we define an approximate version of sign-rank and another variant that captures the minimum dimension required for a reduction to SCO. We also propose several open questions and directions for future research.

While working within the spatial domain can pose problems associated with ill-conditioned scores caused by power-law decay, recent advances in diffusion-based generative models have shown that transitioning to the wavelet domain offers a promising alternative. However, within the wavelet domain, we encounter unique challenges, especially the sparse representation of high-frequency coefficients, which deviates significantly from the Gaussian assumptions in the diffusion process. To this end, we propose a multi-scale generative modeling in the wavelet domain that employs distinct strategies for handling low and high-frequency bands. In the wavelet domain, we apply score-based generative modeling with well-conditioned scores for low-frequency bands, while utilizing a multi-scale generative adversarial learning for high-frequency bands. As supported by the theoretical analysis and experimental results, our model significantly improve performance and reduce the number of trainable parameters, sampling steps, and time.

The paper presents a Graph Attention Convolutional Network (GACN) for flow reconstruction from very sparse data in time-varying geometries. The model incorporates a feature propagation algorithm as a preprocessing step to handle extremely sparse inputs, leveraging information from neighboring nodes to initialize missing features. In addition, a binary indicator is introduced as a validity mask to distinguish between the original and propagated data points, enabling more effective learning from sparse inputs. Trained on a unique data set of Direct Numerical Simulations (DNS) of a motored engine at a technically relevant operating condition, the GACN shows robust performance across different resolutions and domain sizes and can effectively handle unstructured data and variable input sizes. The model is tested on previously unseen DNS data as well as on an experimental data set from Particle Image Velocimetry (PIV) measurements that were not considered during training. A comparative analysis shows that the GACN consistently outperforms both a conventional Convolutional Neural Network (CNN) and cubic interpolation methods on the DNS and PIV test sets by achieving lower reconstruction errors and better capturing fine-scale turbulent structures. In particular, the GACN effectively reconstructs flow fields from domains up to 14 times larger than those observed during training, with the performance advantage increasing for larger domains.

We construct a strong Markov process corresponding to the Dirichlet form of Servadei and Valdinoci and use the process to solve the corresponding Neumann boundary problem for the fractional Laplacian and the half-line., Comment: 85 pages

The Digital Twin Oriented Complex Networked System (DT-CNS) aims to build and extend a Complex Networked System (CNS) model with progressively increasing dynamics complexity towards an accurate reflection of reality -- a Digital Twin of reality. Our previous work proposed evolutionary DT-CNSs to model the long-term adaptive network changes in an epidemic outbreak. This study extends this framework by proposeing the temporal DT-CNS model, where reinforcement learning-driven nodes make decisions on temporal directed interactions in an epidemic outbreak. We consider cooperative nodes, as well as egocentric and ignorant "free-riders" in the cooperation. We describe this epidemic spreading process with the Susceptible-Infected-Recovered ($SIR$) model and investigate the impact of epidemic severity on the epidemic resilience for different types of nodes. Our experimental results show that (i) the full cooperation leads to a higher reward and lower infection number than a cooperation with egocentric or ignorant "free-riders"; (ii) an increasing number of "free-riders" in a cooperation leads to a smaller reward, while an increasing number of egocentric "free-riders" further escalate the infection numbers and (iii) higher infection rates and a slower recovery weakens networks' resilience to severe epidemic outbreaks. These findings also indicate that promoting cooperation and reducing "free-riders" can improve public health during epidemics.

In a real-world social network, weak ties (reflecting low-intensity, infrequent interactions) act as bridges and connect people to different social circles, giving them access to diverse information and opportunities that are not available within one's immediate, close-knit vicinity. Weak ties can be crucial for creativity and innovation, as it introduces new ideas and approaches that people can then combine in novel ways, leading to innovative solutions and creative breakthroughs. Do weak ties facilitate creativity in software in similar ways? In this paper, we show that the answer is ``yes.'' Concretely, we study the correlation between developers' knowledge acquisition through three distinct interaction networks on GitHub and the innovativeness of the projects they develop, across over 38,000 Python projects hosted on GitHub. Our findings suggest that the diversity of projects in which developers engage correlates positively with the innovativeness of their future project developments, whereas the volume of interactions exerts minimal influence. Notably, acquiring knowledge through weak interactions (e.g., starring) as opposed to strong ones (e.g., committing) emerges as a stronger predictor of future novelty., Comment: 11 pages

We introduce FrontierMath, a benchmark of hundreds of original, exceptionally challenging mathematics problems crafted and vetted by expert mathematicians. The questions cover most major branches of modern mathematics -- from computationally intensive problems in number theory and real analysis to abstract questions in algebraic geometry and category theory. Solving a typical problem requires multiple hours of effort from a researcher in the relevant branch of mathematics, and for the upper end questions, multiple days. FrontierMath uses new, unpublished problems and automated verification to reliably evaluate models while minimizing risk of data contamination. Current state-of-the-art AI models solve under 2% of problems, revealing a vast gap between AI capabilities and the prowess of the mathematical community. As AI systems advance toward expert-level mathematical abilities, FrontierMath offers a rigorous testbed that quantifies their progress.

Scalar particles that couple to two up quarks may be produced at the LHC even if they are ultraheavy, in the $8-10$ TeV mass range. A renormalizable theory that includes a diquark particle of this type ($S_{uu}$), two vectorlike quarks ($\chi_1, \chi_2$), and a gauge-singlet pseudoscalar, predicts LHC signals involving four or more jets of very high $p_T$. Two remarkable events observed by the CMS experiment, each involving four high $p_T$-jets, may be due to an $S_{uu}$ of mass near 8.5 TeV, and a $\chi_2$ mass of 2.1 TeV. A separate excess reported by CMS in the nonresonant dijet pair search is consistent with a $\chi_1$ mass of $0.95$ TeV. This hypothesis may be tested through CMS and ATLAS searches for signals with a pair of dijet resonances of masses clustered around 1 TeV, and separately around 2 TeV, which have $4j$ invariant masses in the $5-8$ TeV range. These additional signals would arise from cascade decays of $S_{uu} \to \chi_2 \chi_2 $, which lead to $5j$ and $6j$ events with invariant masses around 8 TeV. Depending on the couplings of the heavy colored particles, additional signals are possible, involving for example highly-boosted top quarks., Comment: 25 pages

This document is comprised of a collection of updated preliminary parameters for the key parts of the muon collider. The updated preliminary parameters follow on from the October 2023 Tentative Parameters Report. Particular attention has been given to regions of the facility that are believed to hold greater technical uncertainty in their design and that have a strong impact on the cost and power consumption of the facility. The data is collected from a collaborative spreadsheet and transferred to overleaf.

We present KinetiX, a software toolkit to generate computationally efficient fuel-specific routines for the chemical source term, thermodynamic and mixture-averaged transport properties for use in combustion simulation codes. The C++ routines are designed for high-performance execution on both CPU and GPU architectures. On CPUs, chemical kinetics computations are optimized by eliminating redundant operations and using data alignment and loops with trivial access patterns that enable auto-vectorization, reducing the latency of complex mathematical operations. On GPUs, performance is improved by loop unrolling, reducing the number of costly exponential evaluations and limiting the number of live variables for better register usage. The accuracy of the generated routines is checked against reference values computed using Cantera and the maximum relative errors are below $10^{-5}\%$. We evaluate the performance of the kernels on some of the latest CPU and GPU architectures from AMD and NVIDIA, i.e., AMD EPYC 9653, AMD MI250X, and NVIDIA H100. The routines generated by KinetiX outperform the general-purpose Cantera library, achieving speedups of up to 2.4x for species production rates and 3.2x for mixture-averaged transport properties on CPUs. Compared to the routines generated by PelePhysics (CEPTR), KinetiX achieves speedups of up to 2.6x on CPUs and 1.7x on GPUs for the species production rates kernel on a single-threaded basis.

This paper addresses the challenges of predicting bioprocess performance, particularly in monoclonal antibody (mAb) production, where conventional statistical methods often fall short due to time-series data's complexity and high dimensionality. We propose a novel Hyperbox Mixture Regression (HMR) model which employs hyperbox-based input space partitioning to enhance predictive accuracy while managing uncertainty inherent in bioprocess data. The HMR model is designed to dynamically generate hyperboxes for input samples in a single-pass process, thereby improving learning speed and reducing computational complexity. Our experimental study utilizes a dataset that contains 106 bioreactors. This study evaluates the model's performance in predicting critical quality attributes in monoclonal antibody manufacturing over a 15-day cultivation period. The results demonstrate that the HMR model outperforms comparable approximators in accuracy and learning speed and maintains interpretability and robustness under uncertain conditions. These findings underscore the potential of HMR as a powerful tool for enhancing predictive analytics in bioprocessing applications.

We compute from density functional theory calculations the dynamical magnetic effective charges of the Jahn-Teller ground state of KCoF$_3$ and we unexpectedly obtain the largest up-to-date values reported in real materials (close to 200$\times 10^{-2}\mu_{\text{B}}/\text{\r{A}}$). We determine that those large values are due to both spin and orbital contributions, as expected from the presence of unquenched Co-d orbital magnetic moment, even though we show that the Co dynamical magnetic effective charges values are surprisingly zero by symmetry. Hence, we conclude that even if KCoF$_3$ is not multiferroic, it is antimagnetoelectric with a compensating but large individual magnetic sublattice magnetoelectric response of 210 ps/m per spin-channel. Our study shows that the dynamical magnetic effective charges might be an important property of magnetic materials beyond the magnetoelectric ones.

Artificial intelligence may assist healthcare systems in meeting increasing demand for pathology services while maintaining diagnostic quality and reducing turnaround time and costs. We aimed to investigate the performance of an institutionally developed system for prostate cancer detection, grading, and workflow optimization and to contrast this with commercial alternatives. From August 2021 to March 2023, we scanned 21,396 slides from 1,147 patients with positive biopsies. We developed models for cancer detection, grading, and screening of equivocal cases for IHC ordering. We compared a task-specific model trained using the PANDA dataset of prostate cancer biopsies with one built using features extracted by the general-purpose histology foundation model, UNI and compare their performance in an unfiltered prospectively collected dataset that reflects our patient population (1737 slides,95 patients). We evaluated the contributions of a bespoke model designed to improve sensitivity in detecting small cancer foci and scoring of broader patterns observed at lower resolution. We found high concordance between the developed systems and pathologist reference in detection (AUC 98.5, sensitivity 95.0, and specificity 97.8), ISUP grading (quadratic Cohen's kappa 0.869), grade group 3 or higher (AUC 97.5, sensitivity 94.9, specificity 96.6) and comparable to published data from commercial systems. Screening could reduce IHC ordering for equivocal cases by 44.5% with an overall error rate of 1.8% (1.4% false positive, 0.4% false negative rates). Institutions like academic medical centers that have high scanning volumes and report abstraction capabilities can develop accurate computational pathology models for internal use. These models have the potential to aid in quality control role and to improve workflow in the pathology lab to help meet future challenges in prostate cancer diagnosis.

The recent tension in the value of the cosmological parameter $S_8 \equiv \sigma_8(\Omega_M/0.3)^{1/2}$, which represents the amplitude of the matter density fluctuations of the universe, has not been resolved. In this work, we present constraints on $S_8$ with the X-ray angular power spectra of clusters and groups measured with the half-sky map from the eROSITA All Sky Survey data release 1 (eRASS1). Thanks to the extensive sky coverage of eRASS1, its power spectrum achieves unprecedented precision compared to previous measurements. Using a well-calibrated, physical halo gas model that includes astrophysics of feedback and non-thermal pressure support, we obtain $S_8 = 0.80^{+0.02}_{-0.01}$ with 1$\sigma$ uncertainty that is competitive against other cosmological probes. Our derived $S_8$ value is smaller than the primary CMB measurements from Planck, but still consistent to within $1\sigma$. We also obtain constraints on the astrophysics of feedback, non-thermal pressure, equation of state in cluster cores, and gas boundaries in clusters and groups. We discuss how additional X-ray observations, and cosmological surveys in microwave and optical, will further improve the cosmological constraints with the angular power spectrum. Our work demonstrates that the angular power spectrum of clusters and groups is a promising probe of both cosmology and astrophysics., Comment: 12 pages, 4 figures, submitted to ApJ

With the advent of large pre-trained vision-language models such as CLIP, prompt learning methods aim to enhance the transferability of the CLIP model. They learn the prompt given few samples from the downstream task given the specific class names as prior knowledge, which we term as semantic-aware classification. However, in many realistic scenarios, we only have access to few samples and knowledge of the class names (e.g., when considering instances of classes). This challenging scenario represents the semantic-agnostic discriminative case. Text-to-Image (T2I) personalization methods aim to adapt T2I models to unseen concepts by learning new tokens and endowing these tokens with the capability of generating the learned concepts. These methods do not require knowledge of class names as a semantic-aware prior. Therefore, in this paper, we first explore Textual Inversion and reveal that the new concept tokens possess both generation and classification capabilities by regarding each category as a single concept. However, learning classifiers from single-concept textual inversion is limited since the learned tokens are suboptimal for the discriminative tasks. To mitigate this issue, we propose Multi-Class textual inversion, which includes a discriminative regularization term for the token updating process. Using this technique, our method MC-TI achieves stronger Semantic-Agnostic Classification while preserving the generation capability of these modifier tokens given only few samples per category. In the experiments, we extensively evaluate MC-TI on 12 datasets covering various scenarios, which demonstrates that MC-TI achieves superior results in terms of both classification and generation outcomes., Comment: Accepted in WACV 2025. Code link: https://github.com/wangkai930418/mc_ti

We propose a novel teacher-student framework to distill knowledge from multiple teachers trained on distinct datasets. Each teacher is first trained from scratch on its own dataset. Then, the teachers are combined into a joint architecture, which fuses the features of all teachers at multiple representation levels. The joint teacher architecture is fine-tuned on samples from all datasets, thus gathering useful generic information from all data samples. Finally, we employ a multi-level feature distillation procedure to transfer the knowledge to a student model for each of the considered datasets. We conduct image classification experiments on seven benchmarks, and action recognition experiments on three benchmarks. To illustrate the power of our feature distillation procedure, the student architectures are chosen to be identical to those of the individual teachers. To demonstrate the flexibility of our approach, we combine teachers with distinct architectures. We show that our novel Multi-Level Feature Distillation (MLFD) can significantly surpass equivalent architectures that are either trained on individual datasets, or jointly trained on all datasets at once. Furthermore, we confirm that each step of the proposed training procedure is well motivated by a comprehensive ablation study. We publicly release our code at https://github.com/AdrianIordache/MLFD., Comment: Accepted at WACV 2025

Transformers and large language models~(LLMs) have seen rapid adoption in all domains. Their sizes have exploded to hundreds of billions of parameters and keep increasing. Under these circumstances, the training of transformers is very expensive and often hits a ``memory wall'', i.e., even when using 3D parallelism (pipeline, tensor, data) and aggregating the memory of many GPUs, it is still not enough to hold the necessary data structures (model parameters, optimizer state, gradients, activations) in GPU memory. To compensate, state-of-the-art approaches offload the optimizer state, at least partially, to the host memory and perform hybrid CPU-GPU computations. However, the management of the combined host-GPU memory is often suboptimal and results in poor overlapping between data movements and computations. This leads to missed opportunities to simultaneously leverage the interconnect bandwidth and computational capabilities of CPUs and GPUs. In this paper, we leverage a key observation that the interleaving of the forward, backward and update phases generate fluctuations in the GPU memory utilization, which can be exploited to dynamically move a part of the optimizer state between the host and the GPU memory at each iteration. To this end, we design and implement \proj, a novel technique to split the LLM into subgroups, whose update phase is scheduled on either the CPU or the GPU based on our proposed performance model that addresses the trade-off between data movement cost, acceleration on the GPUs vs the CPUs, and competition for shared resources. We integrate our approach with DeepSpeed and demonstrate 2.5$\times$ faster iterations over state-of-the-art approaches using extensive experiments.

This paper presents a new look at the neural network (NN) robustness problem, from the point of view of graph theory analysis, specifically graph curvature. Graph curvature (e.g., Ricci curvature) has been used to analyze system dynamics and identify bottlenecks in many domains, including road traffic analysis and internet routing. We define the notion of neural Ricci curvature and use it to identify bottleneck NN edges that are heavily used to ``transport data" to the NN outputs. We provide an evaluation on MNIST that illustrates that such edges indeed occur more frequently for inputs where NNs are less robust. These results will serve as the basis for an alternative method of robust training, by minimizing the number of bottleneck edges.

Track finding can be considered as a complex optimization problem initially introduced in particle physics involving the reconstruction of particle trajectories. A track is typically composed of several consecutive segments (track segments) that resembles a smooth curve without bifurcations. In this paper various modeling approaches are explored in order to assess both their impact and their effectiveness in solving them using quantum and classical methods. We present implementations of three classical models using CPLEX, two quantum models running on actual D-Wave quantum computers, and one quantum model on a D-Wave simulator. To facilitate a fair comparative study and encourage future research in this area, we introduce a new set of benchmark instances, categorized into small, medium, and large scales. Our evaluation of these methods on the benchmark instances indicates that D-Wave methods offer an excellent balance between computation time and result quality, outperforming CPLEX in numerous cases.

Parameter-efficient transfer learning (PETL) aims to adapt pre-trained models to new downstream tasks while minimizing the number of fine-tuned parameters. Adapters, a popular approach in PETL, inject additional capacity into existing networks by incorporating low-rank projections, achieving performance comparable to full fine-tuning with significantly fewer parameters. This paper investigates the relationship between the placement of an adapter and its performance. We observe that adapter location within a network significantly impacts its effectiveness, and that the optimal placement is task-dependent. To exploit this observation, we introduce an extended search space of adapter connections, including long-range and recurrent adapters. We demonstrate that even randomly selected adapter placements from this expanded space yield improved results, and that high-performing placements often correlate with high gradient rank. Our findings reveal that a small number of strategically placed adapters can match or exceed the performance of the common baseline of adding adapters in every block, opening a new avenue for research into optimal adapter placement strategies.

Recent advances in diffusion models have significantly enhanced image generation capabilities. However, customizing these models with new classes often leads to unintended consequences that compromise their reliability. We introduce the concept of open-world forgetting to emphasize the vast scope of these unintended alterations, contrasting it with the well-studied closed-world forgetting, which is measurable by evaluating performance on a limited set of classes or skills. Our research presents the first comprehensive investigation into open-world forgetting in diffusion models, focusing on semantic and appearance drift of representations. We utilize zero-shot classification to analyze semantic drift, revealing that even minor model adaptations lead to unpredictable shifts affecting areas far beyond newly introduced concepts, with dramatic drops in zero-shot classification of up to 60%. Additionally, we observe significant changes in texture and color of generated content when analyzing appearance drift. To address these issues, we propose a mitigation strategy based on functional regularization, designed to preserve original capabilities while accommodating new concepts. Our study aims to raise awareness of unintended changes due to model customization and advocates for the analysis of open-world forgetting in future research on model customization and finetuning methods. Furthermore, we provide insights for developing more robust adaptation methodologies., Comment: Project page: https://hecoding.github.io/open-world-forgetting/

Context. Gamma-ray bursts (GRBs), observed at redshifts as high as 9.4, could serve as valuable probes for investigating the distant Universe. However, this necessitates an increase in the number of GRBs with determined redshifts, as currently, only 12% of GRBs have known redshifts due to observational biases. Aims. We aim to address the shortage of GRBs with measured redshifts, enabling us to fully realize their potential as valuable cosmological probes Methods. Following Dainotti et al. (2024c), we have taken a second step to overcome this issue by adding 30 more GRBs to our ensemble supervised machine learning training sample, an increase of 20%, which will help us obtain better redshift estimates. In addition, we have built a freely accessible and user-friendly web app that infers the redshift of long GRBs (LGRBs) with plateau emission using our machine learning model. The web app is the first of its kind for such a study and will allow the community to obtain redshift estimates by entering the GRB parameters in the app. Results. Through our machine learning model, we have successfully estimated redshifts for 276 LGRBs using X-ray afterglow parameters detected by the Neil Gehrels Swift Observatory and increased the sample of LGRBs with known redshifts by 110%. We also perform Monte Carlo simulations to demonstrate the future applicability of this research. Conclusions. The results presented in this research will enable the community to increase the sample of GRBs with known redshift estimates. This can help address many outstanding issues, such as GRB formation rate, luminosity function, and the true nature of low-luminosity GRBs, and enable the application of GRBs as standard candles, Comment: 18 Figures, 3 tables. Submitted for publication in Astronomy and Astrophysics journal

The growing capabilities of AI models are leading to their wider use, including in safety-critical domains. Explainable AI (XAI) aims to make these models safer to use by making their inference process more transparent. However, current explainability methods are seldom evaluated in the way they are intended to be used: by real-world end users. To address this, we conducted a large-scale user study with 85 healthcare practitioners in the context of human-AI collaborative chest X-ray analysis. We evaluated three types of explanations: visual explanations (saliency maps), natural language explanations, and a combination of both modalities. We specifically examined how different explanation types influence users depending on whether the AI advice and explanations are factually correct. We find that text-based explanations lead to significant over-reliance, which is alleviated by combining them with saliency maps. We also observe that the quality of explanations, that is, how much factually correct information they entail, and how much this aligns with AI correctness, significantly impacts the usefulness of the different explanation types., Comment: EMNLP 2024

AI tools, particularly large language modules, have recently proven their effectiveness within learning management systems and online education programmes. As feedback continues to play a crucial role in learning and assessment in schools, educators must carefully customise the use of AI tools in order to optimally support students in their learning journey. Efforts to improve educational feedback systems have seen numerous attempts reflected in the research studies but mostly have been focusing on qualitatively benchmarking AI feedback against human-generated feedback. This paper presents an exploration of an alternative feedback framework which extends the capabilities of ChatGPT by integrating embeddings, enabling a more nuanced understanding of educational materials and facilitating topic-targeted feedback for quiz-based assessments. As part of the study, we proposed and developed a proof of concept solution, achieving an efficacy rate of 90% and 100% for open-ended and multiple-choice questions, respectively. The results showed that our framework not only surpasses expectations but also rivals human narratives, highlighting the potential of AI in revolutionising educational feedback mechanisms., Comment: Submitted to journal

Estimating rigid objects' poses is one of the fundamental problems in computer vision, with a range of applications across automation and augmented reality. Most existing approaches adopt one network per object class strategy, depend heavily on objects' 3D models, depth data, and employ a time-consuming iterative refinement, which could be impractical for some applications. This paper presents a novel approach, CVAM-Pose, for multi-object monocular pose estimation that addresses these limitations. The CVAM-Pose method employs a label-embedded conditional variational autoencoder network, to implicitly abstract regularised representations of multiple objects in a single low-dimensional latent space. This autoencoding process uses only images captured by a projective camera and is robust to objects' occlusion and scene clutter. The classes of objects are one-hot encoded and embedded throughout the network. The proposed label-embedded pose regression strategy interprets the learnt latent space representations utilising continuous pose representations. Ablation tests and systematic evaluations demonstrate the scalability and efficiency of the CVAM-Pose method for multi-object scenarios. The proposed CVAM-Pose outperforms competing latent space approaches. For example, it is respectively 25% and 20% better than AAE and Multi-Path methods, when evaluated using the $\mathrm{AR_{VSD}}$ metric on the Linemod-Occluded dataset. It also achieves results somewhat comparable to methods reliant on 3D models reported in BOP challenges. Code available: https://github.com/JZhao12/CVAM-Pose, Comment: BMVC 2024, oral presentation, the main paper and supplementary materials are included

We prove a conjecture of Bhatt-Hansen that derived pushforwards along proper morphisms of rigid-analytic spaces commute with Verdier duality on Zariski-constructible complexes. In particular, this yields duality statements for the intersection cohomology of proper rigid-analytic spaces. In our argument, we construct cycle classes in analytic geometry as well as trace maps for morphisms that are either smooth or proper or finite flat, with appropriate coefficients. As an application of our methods, we obtain new, significantly simplified proofs of $p$-adic Poincar\'e duality and the preservation of $\mathbf{F}_p$-local systems under smooth proper higher direct images., Comment: 127 pages, comments welcome

This paper presents a method for task allocation and trajectory generation in cooperative inspection missions using a fleet of multirotor drones, with a focus on wind turbine inspection. The approach generates safe, feasible flight paths that adhere to time-sensitive constraints and vehicle limitations by formulating an optimization problem based on Signal Temporal Logic (STL) specifications. An event-triggered replanning mechanism addresses unexpected events and delays, while a generalized robustness scoring method incorporates user preferences and minimizes task conflicts. The approach is validated through simulations in MATLAB and Gazebo, as well as field experiments in a mock-up scenario., Comment: 2 pages, Accepted for discussion at the workshop session "Formal methods techniques in robotics systems: Design and control" at IROS'24 in Abu Dhabi, UAE

Identifying dependency between two random variables is a fundamental problem. Clear interpretability and ability of a procedure to provide information on the form of the possible dependence is particularly important in evaluating dependencies. We introduce a new estimator of the quantile dependence function and pertinent local acceptance regions. This leads to insightful visualization and evaluation of underlying dependence structure. We also propose a test of independence of two random variables, pertinent to this new estimator. Our procedures are based on ranks and we derive a finite-sample theory that guarantees the inferential validity of our solutions at any given sample size. The procedures are simple to implement and computationally efficient. Large-sample consistency of the proposed test is also proved. We show that, in terms of power, new test is one of the best statistics for independence testing when considering a wide range of alternative models. Finally, we demonstrate use of our approach to visualize dependence structure and to detect local departures from independence through analyzing some datasets.

Large pretrained models are showing increasingly better performance in reasoning and planning tasks across different modalities, opening the possibility to leverage them for complex sequential decision making problems. In this paper, we investigate the capabilities of Large Language Models (LLMs) for reinforcement learning (RL) across a diversity of interactive domains. We evaluate their ability to produce decision-making policies, either directly, by generating actions, or indirectly, by first generating reward models to train an agent with RL. Our results show that, even without task-specific fine-tuning, LLMs excel at reward modeling. In particular, crafting rewards through artificial intelligence (AI) feedback yields the most generally applicable approach and can enhance performance by improving credit assignment and exploration. Finally, in environments with unfamiliar dynamics, we explore how fine-tuning LLMs with synthetic data can significantly improve their reward modeling capabilities while mitigating catastrophic forgetting, further broadening their utility in sequential decision-making tasks.

We construct a Hunt process that can be described as an isotropic $\alpha$-stable L\'evy process reflected from the complement of a bounded open Lipschitz set. In fact, we introduce a new analytic method for concatenating Markov processes. It is based on nonlocal Schr\"odinger perturbations of sub-Markovian transition kernels and the construction of two supermedian functions with different growth rates at infinity. We apply this framework to describe the return distribution and the stationary distribution of the process. To handle the strong Markov property at the reflection time, we introduce a novel ladder process, whose transition semigroup encodes not only the position of the process, but also the number of reflections., Comment: 21 pages, no figures

We present a generative AI algorithm for addressing the challenging task of fast, accurate and robust statistical computation of three-dimensional turbulent fluid flows. Our algorithm, termed as GenCFD, is based on a conditional score-based diffusion model. Through extensive numerical experimentation with both incompressible and compressible fluid flows, we demonstrate that GenCFD provides very accurate approximation of statistical quantities of interest such as mean, variance, point pdfs, higher-order moments, while also generating high quality realistic samples of turbulent fluid flows and ensuring excellent spectral resolution. In contrast, ensembles of operator learning baselines which are trained to minimize mean (absolute) square errors regress to the mean flow. We present rigorous theoretical results uncovering the surprising mechanisms through which diffusion models accurately generate fluid flows. These mechanisms are illustrated with solvable toy models that exhibit the relevant features of turbulent fluid flows while being amenable to explicit analytical formulas., Comment: 71 pages, 30 figures

In the spirit of the action of the symmetric group on the ring of polynomials in $n$ variables, we consider the actions of the seven frieze groups on rings of formal infinite linear combinations of monomials of restricted degree. For each group we describe the respective subring of invariants. We discuss also the structure of those rings as modules over each frieze group., Comment: 16 pages, 7 figures

Image compression is a critical tool in decreasing the cost of storage and improving the speed of transmission over the internet. While deep learning applications for natural images widely adopts the usage of lossy compression techniques, it is not widespread for 3D medical images. Using three CT datasets (17 tasks) and one MRI dataset (3 tasks) we demonstrate that lossy compression up to 20 times have no negative impact on segmentation quality with deep neural networks (DNN). In addition, we demonstrate the ability of DNN models trained on compressed data to predict on uncompressed data and vice versa with no quality deterioration., Comment: 12 pages, 5 figures, 2 tables; accepted on MICCAI Workshop on Advancing Data Solutions in Medical Imaging AI

In this paper, we record some foundational results on adic geometry that seem to be missing in the existing literature. Namely, we develop the Proj construction and a theory of lci closed immersions in the context of locally noetherian analytic adic spaces. In the context of rigid-analytic spaces, these topics have previously been considered in [GL21] and [Con07]. We also develop an etale six functor formalism in the analytic geometry and give a categorical description of lisse and constructible sheaves. All results of this paper are probably well-known to the experts., Comment: Comments are very welcome. A previous version of the paper appeared before here https://bogdanzavyalov.com/refs/notes/adic_notes.pdf

The extraction of blood vessels has recently experienced a widespread interest in medical image analysis. Automatic vessel segmentation is highly desirable to guide clinicians in computer-assisted diagnosis, therapy or surgical planning. Despite a good ability to extract large anatomical structures, the capacity of U-Net inspired architectures to automatically delineate vascular systems remains a major issue, especially given the scarcity of existing datasets. In this paper, we present a novel approach that integrates into deep segmentation shape priors from a Semi-Overcomplete Convolutional Auto-Encoder (S-OCAE) embedding. Compared to standard Convolutional Auto-Encoders (CAE), it exploits an over-complete branch that projects data onto higher dimensions to better characterize tiny structures. Experiments on retinal and liver vessel extraction, respectively performed on publicly-available DRIVE and 3D-IRCADb datasets, highlight the effectiveness of our method compared to U-Net trained without and with shape priors from a traditional CAE., Comment: 5 pages, 4 figures, conference

The paper addresses task assignment and trajectory generation for collaborative inspection missions using a fleet of multi-rotors, focusing on the wind turbine inspection scenario. The proposed solution enables safe and feasible trajectories while accommodating heterogeneous time-bound constraints and vehicle physical limits. An optimization problem is formulated to meet mission objectives and temporal requirements encoded as Signal Temporal Logic (STL) specifications. Additionally, an event-triggered replanner is introduced to address unforeseen events and compensate for lost time. Furthermore, a generalized robustness scoring method is employed to reflect user preferences and mitigate task conflicts. The effectiveness of the proposed approach is demonstrated through MATLAB and Gazebo simulations, as well as field multi-robot experiments in a mock-up scenario., Comment: This work has been submitted to "Robotics and Autonomous Systems" for possible publication

The precise delineation of blood vessels in medical images is critical for many clinical applications, including pathology detection and surgical planning. However, fully-automated vascular segmentation is challenging because of the variability in shape, size, and topology. Manual segmentation remains the gold standard but is time-consuming, subjective, and impractical for large-scale studies. Hence, there is a need for automatic and reliable segmentation methods that can accurately detect blood vessels from medical images. The integration of shape and topological priors into vessel segmentation models has been shown to improve segmentation accuracy by offering contextual information about the shape of the blood vessels and their spatial relationships within the vascular tree. To further improve anatomical consistency, we propose a new joint prior encoding mechanism which incorporates both shape and topology in a single latent space. The effectiveness of our method is demonstrated on the publicly available 3D-IRCADb dataset. More globally, the proposed approach holds promise in overcoming the challenges associated with automatic vessel delineation and has the potential to advance the field of deep priors encoding., Comment: 5 pages, 3 figures, conference

Extracting hepatic vessels from abdominal images is of high interest for clinicians since it allows to divide the liver into functionally-independent Couinaud segments. In this respect, an automated liver blood vessel extraction is widely summoned. Despite the significant growth in performance of semantic segmentation methodologies, preserving the complex multi-scale geometry of main vessels and ramifications remains a major challenge. This paper provides a new deep supervised approach for vessel segmentation, with a strong focus on representations arising from the different scales inherent to the vascular tree geometry. In particular, we propose a new clustering technique to decompose the tree into various scale levels, from tiny to large vessels. Then, we extend standard 3D UNet to multi-task learning by incorporating scale-specific auxiliary tasks and contrastive learning to encourage the discrimination between scales in the shared representation. Promising results, depicted in several evaluation metrics, are revealed on the public 3D-IRCADb dataset., Comment: 5 pages, 5 figures, conference