May 2024 arXiv papers — page 38

CMS Collaboration

The LHC has provided an unprecedented amount of proton-proton collision data, bringing forth exciting opportunities to address fundamental open questions in particle physics. These questions can potentially be answered by performing searches for very rare processes predicted by models that attempt to extend the standard model of particle physics. The data co

Klaudia Bałazy, Mohammadreza Banaei, Karl Aberer, Jacek Tabor

The growth of large language models underscores the need for parameter-efficient fine-tuning. Despite its popularity, LoRA encounters storage and computational challenges when deploying multiple task- or user-specific modules. To address this, we introduce LoRA-XS, a novel fine-tuning method backed by a theoretical derivation. LoRA-XS drastically reduces tra

Yash Kulkarni, Susheela Sharma, Jordan P. Amadio, Farshid Alambeigi

Vital for spinal fracture treatment, pedicle screw fixation is the gold standard for spinal fixation procedures. Nevertheless, due to the screw pullout and loosening issues, this surgery often fails to be effective for patients suffering from osteoporosis (i.e., having low bone mineral density). These failures can be attributed to the rigidity of existing dr

Yu Wang, Nedim Lipka, Ruiyi Zhang, Alexa Siu

Despite the impressive advancements of Large Language Models (LLMs) in generating text, they are often limited by the knowledge contained in the input and prone to producing inaccurate or hallucinated content. To tackle these issues, Retrieval-augmented Generation (RAG) is employed as an effective strategy to enhance the available knowledge base and anchor t

Ngoc Mai Monica Huynh, Luca Franco Pavarino, Simone Scacchi

The aim of the present work is to design, analyze theoretically, and test numerically, a generalized Dryja-Smith-Widlund (GDSW) preconditioner for composite Discontinuous Galerkin discretizations of multicompartment parabolic reaction-diffusion equations, where the solution can exhibit natural discontinuities across the domain. We prove that the resulting pr

Susheela Sharma, Yash Kulkarni, Sarah Go, Jeff Bonyun

Spinal fixation procedures are currently limited by the rigidity of the existing instruments and pedicle screws leading to fixation failures and rigid pedicle screw pull out. Leveraging our recently developed Concentric Tube Steerable Drilling Robot (CT-SDR) in integration with a robotic manipulator, to address the aforementioned issue, here we introduce the

Heeseung Bang, Aditya Dave, Filippos N. Tzortzoglou, Andreas A. Malikopoulos

In this paper, we introduce a metric to evaluate the equity in mobility and a routing framework to enhance the metric within multi-modal intelligent transportation systems. The mobility equity metric (MEM) simultaneously accounts for service accessibility and transportation costs to quantify the equity and fairness in a transportation network. Finally, we de

Cheikh Lo, Abdoul Karim Sane

We show that a bijection $f:\mathbb{H}^2\rightarrow\mathbb{H}^2$ of the hyperbolic plane that sends horocycles to horocycles (respectively hypercycles to hypercycles) is an isometry. This extends a previous result of J. Jeffers on geodesics to all curves with constant curvature in $\mathbb{H}^2$. We go beyond by showing that every abstract automorphism of th

Dimitri P. Voytan, Sriram Ravula, Alexandru Ardel, Elad Liebman

Seismic images often contain both coherent and random artifacts which complicate their interpretation. To mitigate these artifacts, we introduce a novel unsupervised deep-learning method based on Deep Image Prior (DIP) which uses convolutional neural networks. Our approach optimizes the network weights to refine the migration velocity model, rather than the

Yansong Qu, Shaohui Dai, Xinyang Li, Jianghang Lin

3D open-vocabulary scene understanding, crucial for advancing augmented reality and robotic applications, involves interpreting and locating specific regions within a 3D space as directed by natural language instructions. To this end, we introduce GOI, a framework that integrates semantic features from 2D vision-language foundation models into 3D Gaussian Sp

Victor Blanchi, Hugo Paquet

An "element-free" probability distribution is what remains of a probability distribution after we forget the elements to which the probabilities were assigned. These objects naturally arise in Bayesian statistics, in situations where elements are used as labels and their specific identity is not important. This paper develops the structural theory of element

Anni Li, Christos G. Cassandras

We consider a mixed-traffic environment in transportation systems, where Connected and Automated Vehicles (CAVs) coexist with potentially non-cooperative Human-Driven Vehicles (HDVs). We develop a cooperation compliance control framework to incentivize HDVs to align their behavior with socially optimal objectives using a ``refundable toll'' scheme so as to a

Scott Harper, Martin W. Liebeck

Feit and Tits (1978) proved that a nontrivial projective representation of minimal dimension of a finite extension of a finite nonabelian simple group $G$ factors through a projective representation of $G$, except for some groups of Lie type in characteristic 2; the exact exceptions for $G$ were determined by Kleidman and Liebeck (1989). We generalise this r

Amalia Tzartinoglou, Christos G. Tsagas

Bianchi cosmologies are ``natural'' anisotropic extensions of the Friedmann universes and they have long been used to investigate the cosmological implications of anisotropy. The latter introduces new ingredients to the standard scenarios, although there are physical processes and effects that maintain their basic Friedmann features when extended to Bianchi

Yijiao Zhang, Yubai Yuan, Yuexia Zhang, Zhongyi Zhu

Mediation analysis plays a crucial role in causal inference as it can investigate the pathways through which treatment influences outcome. Most existing mediation analysis assumes that mediation effects are static and homogeneous within populations. However, mediation effects usually change over time and exhibit significant heterogeneity among individuals in

Vidush Singhal, Chaitanya Koparkar, Joseph Zullo, Artem Pelenitsyn

While programmers know that the low-level memory representation of data structures can have significant effects on performance, compiler support to optimize the layout of those structures is an under-explored field. Prior work has optimized the layout of individual, non-recursive structures without considering how collections of those objects in linked or re

Yuna Hattori

Inflow control is essential for most fluid mechanics experiments. Although vertically falling soap film flows have been extensively used in the last four decades to study two-dimensional flows, its inflow stability has not yet been discussed in detail. In this article, aiming to improve the inflow stability of the system, we discuss how flow driving systems

Sujatha Vijayakrishnan, Z. Berkson-Korenberg, J. Mainville, L. W. Engel

The concept of fluidic viscosity is ubiquitous in our everyday life and for it to arise the fluidic medium must necessarily form a continuum where macroscopic properties can emerge. While a powerful concept for tangible liquids, hydrodynamic manifestation of collective flow in electronic systems such as two-dimensional electron gases (2DEGs) has only been sh

Emile Contal, Garrin McGoldrick

Large Language Models (LLM) hold immense promise for real-world applications, but their generic knowledge often falls short of domain-specific needs. Fine-tuning, a common approach, can suffer from catastrophic forgetting and hinder generalizability. In-Context Learning (ICL) offers an alternative, which can leverage Retrieval-Augmented Generation (RAG) to p

Patrick Erik Bradley

A parametrised diffusion operator on the regular domain $\Omega$ of a $p$-adic Schottky group is constructed. It is defined as an integral operator on the complex-valued functions on $\Omega$ which are invariant under the Schottky group $\Gamma$, where integration is against the measure defined by an invariant regular differential 1-form $\omega$. It is prov

A. Slaoui, M. El Kirdi, R. Ahl Laamara, M. Alabdulhafith

Cyclic quantum teleportation schemes requires at least the existence of three collaborators acting all as senders and receivers of quantum information, each one of them has an information to be transmitted to the next neighbour in a circular manner. Here, new cyclic quantum teleportation scheme is proposed for perfectly transmitting cyclically three arbitrar

A. A. Shiryaev

The eventual presence of the diamond carbon allotrope in space is discussed in numerous theoretical and experimental studies. The review summarizes the principal mechanisms of nanodiamond formation and experimental results of spectroscopic and structural investigations of nano- and microdiamonds from meteorites. The size dependence of diamond spectroscopic p

Meng Ding, Kaiyi Ji, Di Wang, Jinhui Xu

Continual learning, focused on sequentially learning multiple tasks, has gained significant attention recently. Despite the tremendous progress made in the past, the theoretical understanding, especially factors contributing to catastrophic forgetting, remains relatively unexplored. In this paper, we provide a general theoretical analysis of forgetting in th

Nguyen Phuc Tran, Duy Thanh Tran, Thi Thuy Nga Duong

In recent years, a study by environmental organizations in the world and Vietnam shows that weather change is quite complex. global warming has become a serious problem in the modern world, which is a concern for scientists. last century, it was difficult to forecast the weather due to missing weather monitoring stations and technological limitations. this m

Zachary Davis, Jesús M. Rueda-Becerril, Dimitrios Giannios

A wide range of astrophysical sources exhibit extreme and rapidly varying electromagnetic emission indicative of efficient non-thermal particle acceleration. Understanding these sources often involves comparing data with a broad range of theoretical scenarios. To this end, it is beneficial to have tools that enable not only fast and efficient parametric inve

Zhenfeng Tu, Santiago Aranguri, Arthur Jacot

The training dynamics of linear networks are well studied in two distinct setups: the lazy regime and balanced/active regime, depending on the initialization and width of the network. We provide a surprisingly simple unifying formula for the evolution of the learned matrix that contains as special cases both lazy and balanced regimes but also a mixed regime

Jing Cheng, Yasser G. Alqaham, Zhenyu Gan

This study explores the dynamics of asymmetrical bounding gaits in quadrupedal robots, focusing on the integration of torso pitching and hip motion to enhance speed and stability. Traditional control strategies often enforce a fixed posture, minimizing natural body movements to simplify the control problem. However, this approach may overlook the inherent dy

Rainer Klages

This brief `New & Notable' (perspectives-type) article contains a mini-review on stochastic modelling of cell migration before elaborating on the article by Klimek et al., arXiv:2311.16753 [Biophys. J. 123, 1173-1183 (2024)].

Zhongshan An, Lan-Hsuan Huang

We study local structure of the moduli space of compact Einstein metrics with respect to the boundary conformal metric and mean curvature. In dimension three, we confirm M. Anderson's conjecture in a strong sense, showing that the map from Einstein metrics to such boundary data is generically a local diffeomorphism. In dimensions greater than three, we obtai

Daiki Ikuma, Shunnosuke Ikeda, Noriyoshi Sukegawa, Yuichi Takano

This paper is concerned with the container pre-marshalling problem, which involves relocating containers in the storage area so that they can be efficiently loaded onto ships without reshuffles. In reality, however, ship arrival times are affected by various external factors, which can cause the order of container retrieval to be different from the initial p

Florent Forest, Katharina Rombach, Olga Fink

Deep learning approaches have recently been extensively explored for the prognostics of industrial assets. However, they still suffer from a lack of interpretability, which hinders their adoption in safety-critical applications. To improve their trustworthiness, explainable AI (XAI) techniques have been applied in prognostics, primarily to quantify the impor

Mayara Antunes, Bernardo Carvalho, Welington Cordeiro, José Cueto

We prove that local stable/unstable sets of homeomorphisms of an infinite compact metric space satisfying the gluing-orbit property always contain compact and perfect subsets of the space. As a consequence, we prove that if a positively countably expansive homeomorphism satisfies the gluing-orbit property, then the space is a single periodic orbit. We also p

Arthur Jacot, Alexandre Kaiser

We study Leaky ResNets, which interpolate between ResNets and Fully-Connected nets depending on an 'effective depth' hyper-parameter $\tilde{L}$. In the infinite depth limit, we study 'representation geodesics' $A_{p}$: continuous paths in representation space (similar to NeuralODEs) from input $p=0$ to output $p=1$ that minimize the parameter norm of the ne

Ruth Charney, Matthew Cordes, Antoine Goldsborough, Alessandro Sisto

We show that the Morse boundary exhibits interesting examples of both the existence and non-existence of Cannon-Thurston maps for normal subgroups, in contrast with the hyperbolic case.

Dorian Quelle, Alexandre Bovet

Bluesky is a nascent Twitter-like and decentralized social media network with novel features and unprecedented data access. This paper provides a characterization of its interaction network, studying the political leaning, polarization, network structure, and algorithmic curation mechanisms of five million users. The dataset spans from the website's first re

Ben Freivogel, Tianyi Li

We analyze the size of quantum gravity effects near black hole horizons. By considering black holes in asymptotically AdS spacetime, we can make use of the "quantum deviation" to estimate the size of quantum gravity corrections to the semiclassical analysis. We find that, in a typical pure state, corrections to correlation functions are typically of order ex

Marcelo Matheus Gauy, Larissa Cristina Berti, Arnaldo Cândido, Augusto Camargo Neto

This work investigates Artificial Intelligence (AI) systems that detect respiratory insufficiency (RI) by analyzing speech audios, thus treating speech as a RI biomarker. Previous works collected RI data (P1) from COVID-19 patients during the first phase of the pandemic and trained modern AI models, such as CNNs and Transformers, which achieved $96.5\%$ accu

Quan Liu, Brandon T. Swartz, Ivan Kravchenko, Jason G. Valentine

Deep neural networks (DNNs) have heavily relied on traditional computational units like CPUs and GPUs. However, this conventional approach brings significant computational burdens, latency issues, and high power consumption, limiting their effectiveness. This has sparked the need for lightweight networks like ExtremeC3Net. On the other hand, there have been

Andrew Tanggara, Mile Gu, Kishor Bharti

Quantum error-correcting code (QECC) is the central ingredient in fault-tolerant quantum information processing. An emerging paradigm of dynamical QECC shows that one can robustly encode logical quantum information both temporally and spatially in a more resource-efficient manner than traditional QECCs. Nevertheless, an overarching theory of how dynamical QE

Martin Fournier, Jérémy Fensch, Benoît Commerçon

The presence of kiloparsec-sized bubble structures in both sides of the Galactic plan suggests active phases of Sgr A$^\star$, the central supermassive black hole of the Milky-Way in the last 1-6 Myr. The contribution of such event on the cosmic-ray flux measured in the solar neighborhood is investigated with numerical simulations. We evaluate whether the po

David Harvey

Different models of dark matter can alter the distribution of mass in galaxy clusters in a variety of ways. However, so can uncertain astrophysical feedback mechanisms. Here we present a Machine Learning method that ''learns'' how the impact of dark matter self-interactions differs from that of astrophysical feedback in order to break this degeneracy and mak

Valentin Obst, Arne Heimendahl, Tanmay Singal, David Gross

We describe the symmetry group of the stabilizer polytope for any number $n$ of systems and any prime local dimension $d$. In the qubit case, the symmetry group coincides with the linear and anti-linear Clifford operations. In the case of qudits, the structure is somewhat richer: for $n=1$, it is a wreath product of permutations of bases and permutations of

Amit Adhikary, Debasish Borah, Satyabrata Mahapatra, Indrajit Saha

Light dark matter (DM) with mass around the GeV scale faces weaker bounds from direct detection experiments. If DM couples strongly to a light mediator, it is possible to have observable direct detection rate. However, this also leads to a thermally under-abundant DM relic due to efficient annihilation into light mediators. We propose a novel scenario where

Camille Landri, Ondřej Pejcha

Massive stars in the red supergiant (RSG) phase are known to undergo strong mass loss through winds and observations indicate that a substantial part of this mass loss could be driven by localised and episodic outflows. Various mechanisms have been considered to explain this type of mass loss in RSGs, but these models often focus on single-star evolution. Ho

Nathan Benjamin, Jaeha Lee, Sridip Pal, David Simmons-Duffin

We show that thermal effective field theory controls the long-distance expansion of the partition function of a $d$-dimensional QFT, with an insertion of any finite-order spatial isometry. Consequently, the thermal partition function on a sphere displays a fractal-like structure as a function of angular twist, reminiscent of the behavior of a modular form ne

Biplab Pal, Georges Bouzerar

We present the Stagome lattice, a variant of the Kagome lattice, where one can make any of the bands completely flat by tuning an externally controllable magnetic flux. This systematically allows the energy of the flat band to coincide with the Fermi level. We have analytically calculated the compact localized states associated to each of these flat bands ap

Pranav Nagarajan, Kareem El-Badry, Casey Lam, Henrique Reggiani

We present optical follow-up of IGR J16194-2810, a hard X-ray source discovered by the INTEGRAL mission. The optical counterpart is a $\sim500\,L_\odot$ red giant at a distance of $2.1$ kpc. We measured 17 radial velocities (RVs) of the giant over a period of $271$ days. Fitting these RVs with a Keplerian model, we find an orbital period of $P_{\rm orb} = 19

Maria Knysh, Hong Liu, Natalia Pinzani-Fokeeva

We generalize the formulation of horizon symmetries presented in previous literature to include diffeomorphisms that can shift the location of the horizon. In the context of the AdS/CFT duality, we show that horizon symmetries can be interpreted on the boundary as emergent low-energy gauge symmetries. In particular, we identify a new class of horizon symmetr

Ido Keinan, Iair Arcavi

The discovery of a kilonova associated with the GW170817 binary neutron star merger had far-reaching implications for our understanding of several open questions in physics and astrophysics. Unfortunately, since then, only one robust binary neutron star merger was detected through gravitational waves, GW190425, and no electromagnetic counterpart was identifi

David Boetius, Stefan Leue, Tobias Sutter

Probabilistic verification problems of neural networks are concerned with formally analysing the output distribution of a neural network under a probability distribution of the inputs. Examples of probabilistic verification problems include verifying the demographic parity fairness notion or quantifying the safety of a neural network. We present a new algori

James Fullwood, Arthur J. Parzygnat

While quantum correlations between two spacelike-separated systems are fully encoded by the bipartite density operator associated with the joint system, there does not exist an analogous operator representing general quantum correlations across space and time. This is in stark contrast to the case of classical random variables, which make no distinction betw

Hengameh Bagherian, Melissa Joseph, Martin Schmaltz, Eashwar N. Sivarajan

Models of interacting dark radiation have been shown to alleviate the Hubble tension. Extensions incorporating a coupling between dark matter and dark radiation (DM-DR) have been proposed as combined solutions to both the Hubble and $S_8$ tensions. A key feature of these extended models is a break in the matter power spectrum (MPS), suppressing power for mod

Matti Jarvinen, Dorin Weissman

We establish a new tool for studying strongly coupled matter: an effective theory of black holes in gravity, which maps to a hydrodynamic description of field theories via the gauge-gravity duality. Our approach is inspired by previously known effective theories found in the limit of high number of dimensions. We argue that the new approach can accurately de

A. Vietri, M. Berton, E. Järvelä, M. Kunert-Bajraszewska

The term 'active galactic nuclei' (AGN) subtends a huge variety of objects, classified on their properties at different wavelengths. Peaked sources (PS) represent a class of AGN at the first stage of evolution, characterised by a peaked radio spectrum. Among these radio sources, low-luminosity compact (LLC) sources can be identified as PS accreting with a hi

Gino Isidori, Zachary Polonsky, Arianna Tinari

We analyze $B^0 \to K^0 \bar{\ell}\ell$ long-distance contributions induced by the rescattering of a pair of charmed and charmed-strange mesons. We present an explicit estimate of these contributions using an effective description in terms of hadronic degrees of freedom, supplemented by data on the $B^0 \to D^*D_s (D^*_sD)$ transition in order to reproduce t

Yichen Hu, Jing Wang, Biao Lian

We study the probability distribution of the resistance, or equivalently the charge transmission, of a decoherent quantum Hall-superconductor edge, with the decoherence coming from metallic puddles along the edge. Such metallic puddles may originate from magnetic vortex cores or other superconductivity suppressing perturbations. In contrast to the distributi

Kenny C. Y. Ng, Andrew Hillier, Shin'ichiro Ando

The magnetic fields that emerge from beneath the solar surface and permeate the solar atmosphere are the key drivers of space weather and, thus, understanding them is important to human society. Direct observations, used to measure magnetic fields, can only probe the magnetic fields in the photosphere and above, far from the regions the magnetic fields are b

Sowmiya Balan, Csaba Balázs, Torsten Bringmann, Christopher Cappiello

Sub-GeV dark matter (DM) particles produced via thermal freeze-out evade many of the strong constraints on heavier DM candidates but at the same time face a multitude of new constraints from laboratory experiments, astrophysical observations and cosmological data. In this work we combine all of these constraints in order to perform frequentist and Bayesian g

A. Savino, M. Gennaro, A. E. Dolphin, D. R. Weisz

We empirically assess estimates from v3.0 of the JWST NIRCam Exposure Time Calculator (ETC) using observations of resolved stars in Local Group targets taken as part of the Resolved Stellar Populations Early Release Science (ERS) Program. For bright stars, we find that: (i) purely Poissonian estimates of the signal-to-noise ratio (SNR) are in good agreement

Shiyong Guo, Martin Sasieta, Brian Swingle

We study the dynamical generation of randomness in Brownian systems as a function of the degree of locality of the Hamiltonian. We first express the trace distance to a unitary design for these systems in terms of an effective equilibrium thermal partition function, and provide a set of conditions that guarantee a linear time to design. We relate the trace d

Krishna Rajagopal, Bruno Scheihing-Hitschfeld, Rachel Steinhorst

"Attractor" solutions for the pre-hydrodynamic, far-from-equilibrium, evolution of the matter produced in relativistic heavy ion collisions have emerged as crucial descriptors of the rapid hydrodynamization of quark-gluon plasma (QGP). Adiabatic Hydrodynamization (AH) has been proposed as a framework with which to describe, explain, and predict attractor beh

Giovanni De Toni, Nastaran Okati, Suhas Thejaswi, Eleni Straitouri

Decision support systems based on prediction sets have proven to be effective at helping human experts solve classification tasks. Rather than providing single-label predictions, these systems provide sets of label predictions constructed using conformal prediction, namely prediction sets, and ask human experts to predict label values from these sets. In thi

Robert Wille, Lucas Berent, Tobias Forster, Jagatheesan Kunasaikaran

Quantum computers are becoming a reality and numerous quantum computing applications with a near-term perspective (e.g., for finance, chemistry, machine learning, and optimization) and with a long-term perspective (e.g., for cryptography or unstructured search) are currently being investigated. However, designing and realizing potential applications for thes

Camille Eloy, Gabriel Larios

We construct new continuous families of ${\rm AdS}_3\times S^3\times {\rm T}^4$ and ${\rm AdS}_3\times S^3\times S^3\times S^1$ solutions in heterotic and type II supergravities. These families are found in three-dimensional consistent truncations and controlled by 17 parameters, which include TsT $\beta$ deformations and encompass several supersymmetric sub

Dominik S. Kufel, Jack Kemp, DinhDuy Vu, Simon M. Linsel

We propose and analyze a family of approximately-symmetric neural networks for quantum spin liquid problems. These tailored architectures are parameter-efficient, scalable, and significantly outperform existing symmetry-unaware neural network architectures. Utilizing the mixed-field toric code and PXP Rydberg Hamiltonian models, we demonstrate that our appro

Simone Caletti, Aude Gehrmann-De Ridder, Alexander Huss, Adrian Rodriguez Garcia

The identification of a hadron in the final state of hadron-collider events that feature a leptonically decaying vector boson can provide essential information on the parton content of the colliding protons. Moreover, the study of hadrons inside jets can provide deeper insights into the fragmentation dynamics. We provide theoretical predictions for specific

Zhaoyu Han, Steven Kivelson

By explicit microscopic construction involving a mapping to a quantum vertex model subject to the `ice rule,' we show that an electronically `trivial' band insulator with suitable vibrational (phonon) degrees of freedom can host a ``resonating valence-bond'' state - a quantum phase with emergent gauge fields. This novel type of band insulator is identifiable

Jessica N. Howard, Marc S. Klinger, Anindita Maiti, Alexander G. Stapleton

The Neural Network Field Theory correspondence (NNFT) is a mapping from neural network (NN) architectures into the space of statistical field theories (SFTs). The Bayesian renormalization group (BRG) is an information-theoretic coarse graining scheme that generalizes the principles of the exact renormalization group (ERG) to arbitrarily parameterized probabi

Ryohei Kobayashi, Yuxuan Zhang, Naren Manjunath, Maissam Barkeshli

In the presence of crystalline symmetry, topologically ordered states can acquire a host of symmetry-protected invariants. These determine the patterns of crystalline symmetry fractionalization of the anyons in addition to fractionally quantized responses to lattice defects. Here we show how ground state expectation values of partial rotations centered at hi

Mu Cai, Jianwei Yang, Jianfeng Gao, Yong Jae Lee

Large Multimodal Models (LMMs) such as LLaVA have shown strong performance in visual-linguistic reasoning. These models first embed images into a fixed large number of visual tokens and then feed them into a Large Language Model (LLM). However, this design causes an excessive number of tokens for dense visual scenarios such as high-resolution images and vide

Yuanhui Huang, Wenzhao Zheng, Yunpeng Zhang, Jie Zhou

3D semantic occupancy prediction aims to obtain 3D fine-grained geometry and semantics of the surrounding scene and is an important task for the robustness of vision-centric autonomous driving. Most existing methods employ dense grids such as voxels as scene representations, which ignore the sparsity of occupancy and the diversity of object scales and thus l

Chankyu Lee, Rajarshi Roy, Mengyao Xu, Jonathan Raiman

Decoder-only LLM-based embedding models are beginning to outperform BERT or T5-based embedding models in general-purpose text embedding tasks, including dense vector-based retrieval. In this work, we introduce NV-Embed, incorporating architectural designs, training procedures, and curated datasets to significantly enhance the performance of LLM as a versatil

Ruochen Jiao, Shaoyuan Xie, Justin Yue, Takami Sato

Large Language Models (LLMs) have shown significant promise in real-world decision-making tasks for embodied artificial intelligence, especially when fine-tuned to leverage their inherent common sense and reasoning abilities while being tailored to specific applications. However, this fine-tuning process introduces considerable safety and security vulnerabil

Kuan-Chih Huang, Xiangtai Li, Lu Qi, Shuicheng Yan

Recent advancements in multimodal large language models (LLMs) have demonstrated significant potential across various domains, particularly in concept reasoning. However, their applications in understanding 3D environments remain limited, primarily offering textual or numerical outputs without generating dense, informative segmentation masks. This paper intr

Shaoyuan Xie, Lingdong Kong, Wenwei Zhang, Jiawei Ren

Recent advancements in bird's eye view (BEV) representations have shown remarkable promise for in-vehicle 3D perception. However, while these methods have achieved impressive results on standard benchmarks, their robustness in varied conditions remains insufficiently assessed. In this study, we present RoboBEV, an extensive benchmark suite designed to evalua

Ouail Kitouni, Niklas Nolte, Víctor Samuel Pérez-Díaz, Sokratis Trifinopoulos

Mechanistic Interpretability (MI) promises a path toward fully understanding how neural networks make their predictions. Prior work demonstrates that even when trained to perform simple arithmetic, models can implement a variety of algorithms (sometimes concurrently) depending on initialization and hyperparameters. Does this mean neuron-level interpretabilit

Zhuoling Li, Xiaogang Xu, Zhenhua Xu, SerNam Lim

Recent embodied agents are primarily built based on reinforcement learning (RL) or large language models (LLMs). Among them, RL agents are efficient for deployment but only perform very few tasks. By contrast, giant LLM agents (often more than 1000B parameters) present strong generalization while demanding enormous computing resources. In this work, we combi

Shuai Zeng, Wenzhao Zheng, Jiwen Lu, Haibin Yan

3D object detection aims to recover the 3D information of concerning objects and serves as the fundamental task of autonomous driving perception. Its performance greatly depends on the scale of labeled training data, yet it is costly to obtain high-quality annotations for point cloud data. While conventional methods focus on generating pseudo-labels for unla

Jiahui Lei, Yijia Weng, Adam Harley, Leonidas Guibas

We introduce 4D Motion Scaffolds (MoSca), a modern 4D reconstruction system designed to reconstruct and synthesize novel views of dynamic scenes from monocular videos captured casually in the wild. To address such a challenging and ill-posed inverse problem, we leverage prior knowledge from foundational vision models and lift the video data to a novel Motion

Xiaoman Delores Ding, Zifan Carl Guo, Eric J. Michaud, Ziming Liu

When a neural network can learn multiple distinct algorithms to solve a task, how does it "choose" between them during training? To approach this question, we take inspiration from ecology: when multiple species coexist, they eventually reach an equilibrium where some survive while others die out. Analogously, we suggest that a neural network at initializati

Hao Dong, Yue Zhao, Eleni Chatzi, Olga Fink

Detecting out-of-distribution (OOD) samples is important for deploying machine learning models in safety-critical applications such as autonomous driving and robot-assisted surgery. Existing research has mainly focused on unimodal scenarios on image data. However, real-world applications are inherently multimodal, which makes it essential to leverage informa

Chenxuan Li, Jiaming Liu, Guanqun Wang, Xiaoqi Li

Recently, some studies have integrated Multimodal Large Language Models into robotic manipulation, constructing vision-language-action models (VLAs) to interpret multimodal information and predict SE(3) poses. While VLAs have shown promising progress, they may suffer from failures when faced with novel and complex tasks. To emulate human-like reasoning for m

Alexander Drewitz, Alexis Prévost, Pierre-François Rodriguez

We investigate the bond percolation model on transient weighted graphs ${G}$ induced by the excursion sets of the Gaussian free field on the corresponding metric graph. Under the sole assumption that its sign clusters do not percolate, we derive an extension of Lupu's formula for the two-point function at criticality. We then focus on the low-dimensional cas

Abdulaziz Almuzairee, Nicklas Hansen, Henrik I. Christensen

Q-learning algorithms are appealing for real-world applications due to their data-efficiency, but they are very prone to overfitting and training instabilities when trained from visual observations. Prior work, namely SVEA, finds that selective application of data augmentation can improve the visual generalization of RL agents without destabilizing training.

Zhengfei Kuang, Shengqu Cai, Hao He, Yinghao Xu

Research on video generation has recently made tremendous progress, enabling high-quality videos to be generated from text prompts or images. Adding control to the video generation process is an important goal moving forward and recent approaches that condition video generation models on camera trajectories make strides towards it. Yet, it remains challengin

ZeMing Gong, Austin T. Wang, Xiaoliang Huo, Joakim Bruslund Haurum

Measuring biodiversity is crucial for understanding ecosystem health. While prior works have developed machine learning models for taxonomic classification of photographic images and DNA separately, in this work, we introduce a multimodal approach combining both, using CLIP-style contrastive learning to align images, barcode DNA, and text-based representatio

Rob Burns

We investigate the running sums of some well-known automatic sequences to determine whether they are synchronised.

Navin Kamuni, Dheerendra Panwar

The aim of this study is to teach an algorithm how to recognize different types of music. Users will submit songs for analysis. Since the algorithm hasn't heard these songs before, it needs to figure out what makes each song unique. It does this by breaking down the songs into different parts and studying things like rhythm, melody, and tone via supervised l

Aditya Ravuri, Neil D. Lawrence

This paper shows that dimensionality reduction methods such as UMAP and t-SNE, can be approximately recast as MAP inference methods corresponding to a model introduced in Ravuri et al. (2023), that describes the graph Laplacian (an estimate of the data precision matrix) using a Wishart distribution, with a mean given by a non-linear covariance function evalu

Ema Dimastrogiovanni, Matteo Fasiello, Martino Michelotti, Ogan Özsoy

Inflationary models equipped with Chern-Simons coupling between their axion and gauge sectors exhibit an array of interesting signals including a testable chiral gravitational wave spectrum. The energy injection in the gauge sector triggered by the rolling axion leads to a well-studied enhancement of gauge field fluctuations. These may in turn affect observa

Daniel Hickey, Daniel M. T. Fessler, Kristina Lerman, Keith Burghardt

Many online hate groups exist to disparage others based on race, gender identity, sex, or other characteristics. The accessibility of these communities allows users to join multiple types of hate groups (e.g., a racist community and a misogynistic community), raising the question of whether users who join additional types of hate communities could be further

B. Theodore Zhang, Kohta Murase, Nick Ekanger, Mukul Bhattacharya

We investigate the propagation of ultraheavy (UH) nuclei as ultrahigh-energy cosmic rays (UHECRs). We show that their energy loss lengths at $\lesssim300$ EeV are significantly longer than those of protons and intermediate-mass nuclei, and that the highest-energy cosmic rays with energies beyond $\sim100$ EeV, including the Amaterasu particle, may be UH-UHEC

Lin Xi, Yuan Feng

High-resolution observations with GRAVITY-VLTI instrument have provided abundant information about the flares in Sgr A*, the supermassive black hole in our Galactic center, including the time-dependent location of the centroid (a "hot spot"), the light curve, and polarization. Yuan et al. (2009) proposed a "coronal mass ejection" model to explain the flares

Baiying Liu, Chi-Heng Lo, Freydoon Shahidi

In this paper, following Arthur's ideas, we rework the process of constructing the anti-tempered local Arthur packets for quasi-split classical groups and their pure inner forms. In particular, we present explicit examples illustrating certain gap in a consequential lemma of Arthur and provide a uniform modification, based on the work of Moeglin, Waldspurger

Rui Guo, Greg Farnan, Niall McLaughlin, Barry Devereux

The BioNLP ACL'24 Shared Task on Streamlining Discharge Documentation aims to reduce the administrative burden on clinicians by automating the creation of critical sections of patient discharge letters. This paper presents our approach using the Llama3 8B quantized model to generate the "Brief Hospital Course" and "Discharge Instructions" sections. We employ

H. L. Dao

In this work, we report the results of applying deep learning based on hybrid convolutional-recurrent and purely recurrent neural network architectures to the dataset of almost one million complete intersection Calabi-Yau four-folds (CICY4) to machine-learn their four Hodge numbers $h^{1,1}, h^{2,1}, h^{3,1}, h^{2,2}$. In particular, we explored and experime

Mucong Ding, Yuancheng Xu, Tahseen Rabbani, Xiaoyu Liu

Dataset condensation can be used to reduce the computational cost of training multiple models on a large dataset by condensing the training dataset into a small synthetic set. State-of-the-art approaches rely on matching the model gradients between the real and synthetic data. However, there is no theoretical guarantee of the generalizability of the condense

Tao Li, Chin-Yi Cheng, Amber Xie, Gang Li

Layout design, such as user interface or graphical layout in general, is fundamentally an iterative revision process. Through revising a design repeatedly, the designer converges on an ideal layout. In this paper, we investigate how revision edits from human designer can benefit a multimodal generative model. To do so, we curate an expert dataset that traces

Biqing Qi, Junqi Gao, Kaiyan Zhang, Dong Li

Despite the promising performance of state space models (SSMs) in long sequence modeling, limitations still exist. Advanced SSMs like S5 and S6 (Mamba) in addressing non-uniform sampling, their recursive structures impede efficient SSM computation via convolution. To overcome compatibility limitations in parallel convolutional computation, this paper propose

Ruizhi Shao, Youxin Pang, Zerong Zheng, Jingxiang Sun

We present a novel approach for generating 360-degree high-quality, spatio-temporally coherent human videos from a single image. Our framework combines the strengths of diffusion transformers for capturing global correlations across viewpoints and time, and CNNs for accurate condition injection. The core is a hierarchical 4D transformer architecture that fac