October 2023 arXiv papers — page 123

Mohammed Ayman Shalaby, Charles Champagne Cossette, Jerome Le Ny, James Richard Forbes

This document is in supplement to the paper titled "Multi-Robot Relative Pose Estimation and IMU Preintegration Using Passive UWB Transceivers", available at [1]. The purpose of this document is to show how IMU biases can be incorporated into the framework presented in [1], while maintaining the differential Sylvester equation form of the process model.

Haote Li, Yu Shee, Brandon Allen, Federica Maschietto

We introduce the Kernel-Elastic Autoencoder (KAE), a self-supervised generative model based on the transformer architecture with enhanced performance for molecular design. KAE is formulated based on two novel loss functions: modified maximum mean discrepancy and weighted reconstruction. KAE addresses the long-standing challenge of achieving valid generation

Jiaxing Zhao, Joerg Aichelin, Pol Bernard Gossiaux, Klaus Werner

The creation of a quark-gluon plasma (QGP) is expected in heavy ion collisions. It came as a surprise that proton-proton collisions at ultrarelativistic energies show as well a ``QGP-like'' behavior and signs of the creation of a fluid, although the corresponding system size is not more than a few cubic femtometers. Even more surprisingly, also heavy flavor

Christian A. Schiller

Reinforcement Learning (RL) has achieved significant milestones in the gaming domain, most notably Google DeepMind's AlphaGo defeating human Go champion Ken Jie. This victory was also made possible through the Atari Learning Environment (ALE): The ALE has been foundational in RL research, facilitating significant RL algorithm developments such as AlphaGo and

Thomas Aird, Duarte Ribeiro

We study the equational theories and bases of meets and joins of several varieties of plactic-like monoids. Using those results, we construct sublattices of the lattice of varieties of monoids, generated by said varieties. We calculate the axiomatic ranks of their elements, obtain plactic-like congruences whose corresponding factor monoids generate varieties

Erfan Darzi, Nanna M. Sijtsema, P. M. A van Ooijen

This paper explores the security aspects of federated learning applications in medical image analysis. Current robustness-oriented methods like adversarial training, secure aggregation, and homomorphic encryption often risk privacy compromises. The central aim is to defend the network against potential privacy breaches while maintaining model robustness agai

Milad Farsi, Shuhao Bian, Nasser L. Azad, Xiaobing Shi

We propose a novel framework for designing a resilient Model Predictive Control (MPC) targeting uncertain linear systems under cyber attack. Assuming a periodic attack scenario, we model the system under Denial of Service (DoS) attack, also with measurement noise, as an uncertain linear system with parametric and additive uncertainty. To detect anomalies, we

Ethan Callanan, Amarachi Mbakwe, Antony Papadimitriou, Yulong Pei

Large Language Models (LLMs) have demonstrated remarkable performance on a wide range of Natural Language Processing (NLP) tasks, often matching or even beating state-of-the-art task-specific models. This study aims at assessing the financial reasoning capabilities of LLMs. We leverage mock exam questions of the Chartered Financial Analyst (CFA) Program to c

Deyu Zou, Shikun Liu, Siqi Miao, Victor Fung

Geometric deep learning (GDL) has gained significant attention in scientific fields, for its proficiency in modeling data with intricate geometric structures. However, very few works have delved into its capability of tackling the distribution shift problem, a prevalent challenge in many applications. To bridge this gap, we propose GeSS, a comprehensive benc

Saleh Rezaeiravesh, Christian Gscheidle, Adam Peplinski, Jochen Garcke

The statistics obtained from turbulent flow simulations are generally uncertain due to finite time averaging. The techniques available in the literature to accurately estimate these uncertainties typically only work in an offline mode, that is, they require access to all available samples of a time series at once. In addition to the impossibility of online m

Jing Li, E. Ya. Sherman, Andreas Ruschhaupt

We explore the dynamics of a hybrid classical-quantum system consisting of a classical piston and a self-interacting pseudospin 1/2 Bose-Einstein condensate with a time-dependent Rabi coupling. We investigate the mechanical work produced by the piston moving as a result of the quantum pressure of the condensate. The time-dependent Rabi field redistributes th

Sean J. Wang, Honghao Zhu, Aaron M. Johnson

Autonomous off-road driving is challenging as risky actions taken by the robot may lead to catastrophic damage. As such, developing controllers in simulation is often desirable as it provides a safer and more economical alternative. However, accurately modeling robot dynamics is difficult due to the complex robot dynamics and terrain interactions in unstruct

Jianghui Ji, Dongjie Tan, Chunhui Bao, Xiumin Huang

The Standards of Fundamental Astronomy (SOFA) is a service provided by the International Astronomical Union (IAU) that offers algorithms and software for astronomical calculations, which was released in two versions by FORTRAN 77 and ANSI C, respectively. In this work, we implement the python package PyMsOfa for SOFA service by three ways: (1) a python wrapp

Susan Athey, Niall Keleher, Jann Spiess

In many settings, interventions may be more effective for some individuals than others, so that targeting interventions may be beneficial. We analyze the value of targeting in the context of a large-scale field experiment with over 53,000 college students, where the goal was to use "nudges" to encourage students to renew their financial-aid applications befo

Foivos I. Diakogiannis, Suzanne Furby, Peter Caccetta, Xiaoliang Wu

State-of-the-art models in semantic segmentation primarily operate on single, static images, generating corresponding segmentation masks. This one-shot approach leaves little room for error correction, as the models lack the capability to integrate multiple observations for enhanced accuracy. Inspired by work on semantic change detection, we address this lim

Hanhan Zhou, Tian Lan, Guru Venkataramani, Wenbo Ding

Cross-device Federated Learning (FL) faces significant challenges where low-end clients that could potentially make unique contributions are excluded from training large models due to their resource bottlenecks. Recent research efforts have focused on model-heterogeneous FL, by extracting reduced-size models from the global model and applying them to local c

Yao-Hung Hubert Tsai, Vansh Dhar, Jialu Li, Bowen Zhang

Recent efforts to enable visual navigation using large language models have mainly focused on developing complex prompt systems. These systems incorporate instructions, observations, and history into massive text prompts, which are then combined with pre-trained large language models to facilitate visual navigation. In contrast, our approach aims to fine-tun

C. M. Wilson, R. Ganesh, D. A. Crandles

The recently-discovered high-entropy oxides offer a paradoxical combination of crystalline arrangement and high disorder. They differ qualitatively from established paradigms for disordered solids such as glasses and alloys. In these latter systems, it is well known that disorder induces localized vibrational excitations. In this article, we explore the poss

Andrea Calcinari, Steffen Gielen

We define generalised Gaussian states for quantum cosmological models based on the $\mathfrak{su(1,1)}$ algebra, with particular emphasis on its realisation in group field theory for a single field mode, and study their semiclassical properties. These states are generalisations of coherent, squeezed and thermal states considered previously. As two possible c

Daniel Galvin, Roberto Ladu

We construct the first examples of non-smoothable self-homeomorphisms of smooth $4$-manifolds with boundary that fix the boundary and act trivially on homology. As a corollary, we construct self-diffeomorphisms of $4$-manifolds with boundary that fix the boundary and act trivially on homology but cannot be isotoped to any self-diffeomorphism supported in a c

Margaret L. Duffy, Brian Medeiros, Andrew Gettelman, Trude Eidhammer

The sensitivity of cloud feedbacks to atmospheric model parameters is evaluated using a CAM6 perturbed parameter ensemble (PPE). The CAM6 PPE perturbs 45 parameters across 262 simulations, 206 of which are used here. The spread in total cloud feedback and its six components across the CAM6 PPE are comparable to the spread across the CMIP6 and AMIP ensembles,

Ross G. Pinsky

A permutation is \it separable \rm if it can be obtained from the singleton permutation by iterating direct sums and skew sums. Equivalently, it is separable if and only it avoids the patterns 2413 and 3142. Under the uniform probability on separable permutations of $[n]$, let the random variable $A_n$ denote the length of the longest alternating subsequence

Jack Dippel, Adrian Vetta

In the famous network creation game of Fabrikant et al. a set of agents play a game to build a connected graph. The $n$ agents form the vertex set $V$ of the graph and each vertex $v\in V$ buys a set $E_v$ of edges inducing a graph $G=(V,\bigcup\limits_{v\in V} E_v)$. The private objective of each vertex is to minimize the sum of its building cost (the cost

James Berneburg, Daigo Shishika, Cameron Nowzari

This paper theoretically investigates the closed-loop performance of active disturbance rejection control (ADRC) on a third-order linear plant with relative degree 3, subject to a class of exogenous disturbances. While PID control cannot be guaranteed to be capable of stabilizing such plants, ADRC offers a model-free alternative. However, many existing works

Simon Ouellette, Rolf Pfister, Hansueli Jud

Algorithmic generalization in machine learning refers to the ability to learn the underlying algorithm that generates data in a way that generalizes out-of-distribution. This is generally considered a difficult task for most machine learning algorithms. Here, we analyze algorithmic generalization when counting is required, either implicitly or explicitly. We

Heasung Kim, Sravan Kumar Ankireddy

In this work, we consider the problem of network parameter optimization for rate maximization. We frame this as a joint optimization problem of power control, beam forming, and interference cancellation. We consider the setting where multiple Base Stations (BSs) communicate with multiple user equipment (UEs). Because of the exponential computational complexi

Yixiao Li, Yifan Yu, Chen Liang, Pengcheng He

Quantization is an indispensable technique for serving Large Language Models (LLMs) and has recently found its way into LoRA fine-tuning. In this work we focus on the scenario where quantization and LoRA fine-tuning are applied together on a pre-trained model. In such cases it is common to observe a consistent gap in the performance on downstream tasks betwe

Rosa Wallace Everson, Tenley Hutchinson-Smith, Alejandro Vigna-Gómez, Enrico Ramirez-Ruiz

Thorne-\.{Z}ytkow objects (T\.{Z}Os), hypothetical merger products in which a neutron star is embedded in a stellar core, are traditionally considered steady-state configurations. Their assembly, especially through dynamical channels, is not well-understood. The predominant focus in the literature has been on the observational signatures related to the evolu

Antonio Alfieri, Alberto Cavallo

The scope of the paper is threefold. First, we build on recent work by Hayden to compute Hedden's tau-invariant $\tau_{\xi}(L)$ in the case when $\xi$ is a Stein fillable contact structure on a rational homology sphere, and $L$ is a transverse link arising as the boundary of a pseudo-holomorphic curve. This leads to a new proof of the relative Thom conjectur

Niloofar Bahadori, Yoshitomo Matsubara, Marco Levorato, Francesco Restuccia

Modern IEEE 802.11 (Wi-Fi) networks extensively rely on multiple-input multiple-output (MIMO) to significantly improve throughput. To correctly beamform MIMO transmissions, the access point needs to frequently acquire a beamforming matrix (BM) from each connected station. However, the size of the matrix grows with the number of antennas and subcarriers, resu

YuanTso Li, Wei Yu, Hamid Sarlak

This study presents a comprehensive numerical analysis of a full-scale horizontal-axis Floating Offshore Wind Turbine (FOWT) subjected to harmonic surging motions under both laminar and turbulent inflow conditions. Utilizing high-fidelity Computational Fluid Dynamics (CFD) simulations, namely Large-Eddy Simulation (LES) with Actuator Line Model (ALM), this r

Evi M. C. Huijben, Sina Amirrajab, Josien P. W. Pluim

Out-of-distribution (OOD) detection is crucial for the safety and reliability of artificial intelligence algorithms, especially in the medical domain. In the context of the Medical OOD (MOOD) detection challenge 2023, we propose a pipeline that combines a histogram-based method and a diffusion-based method. The histogram-based method is designed to accuratel

Hikmatullah Mohammadi, Ziaullah Momand, Parwin Habibi, Nazifa Ramaki

Afghanistan has witnessed many armed conflicts throughout history, especially in the past 20 years; these events have had a significant impact on human lives, including military and civilians, with potential fatalities. In this research, we aim to leverage state-of-the-art machine learning techniques to classify the outcomes of Afghanistan armed conflicts to

Joseph M. Shunia

We establish a novel connection between the central binomial coefficients $\binom{2n}{n}$ and Gould's sequence through the construction of a specialized multivariate polynomial quotient ring. Our ring structure is characterized by ideals generated from elements defined by polynomial recurrence relations, and we prove the conditions under which the set of pol

David Urbanik, Ziquan Yang

We introduce a new method to study mixed characteristic deformation of line bundles. In particular, for sufficiently large smooth projective families $f : \mathscr{X} \to \mathscr{S}$ defined over the ring of $N$-integers $\mathscr{O}_{L}[1/N]$ of a number field $L$, we produce a proper closed subscheme $\mathscr{E} \subsetneq \mathscr{S}$ outside of which a

Kyle E. Marlantes, Kevin J. Maki

A hybrid data-driven method, which combines low-fidelity physics with machine learning (ML) to model nonlinear forces and moments at a reduced computational cost, is applied to predict the roll motions of an appended ONR Tumblehome (ONRT) hull in waves. The method is trained using CFD data of unforced roll decay time series--a common data set used in paramet

Alexander Most, Maksim Eren, Nigel Lawrence, Boian Alexandrov

Supervisory Control and Data Acquisition (SCADA) systems often serve as the nervous system for substations within power grids. These systems facilitate real-time monitoring, data acquisition, control of equipment, and ensure smooth and efficient operation of the substation and its connected devices. Previous work has shown that dimensionality reduction-based

Mark C. Messner, Tianchen Hu, Tianju Chen

Stiff systems of ordinary differential equations (ODEs) and sparse training data are common in scientific problems. This paper describes efficient, implicit, vectorized methods for integrating stiff systems of ordinary differential equations through time and calculating parameter gradients with the adjoint method. The main innovation is to vectorize the prob

Dionysios Anninos, Damián A. Galante, Chawakorn Maneerat

We consider four-dimensional general relativity with vanishing cosmological constant defined on a manifold with a boundary. In Lorentzian signature, the timelike boundary is of the form $\boldsymbol{\sigma} \times \mathbb{R}$, with $\boldsymbol{\sigma}$ a spatial two-manifold that we take to be either flat or $S^2$. In Euclidean signature, we take the bounda

Xin Qian

In this paper, we are concerned with noncollapsed Riemannian manifolds $(M^{n},g)$ with integral curvature bounds, as well as their Gromov-Hausdorff limits $(M^{n}_{i},g_{i})\xrightarrow{GH}(X,d)$. Our main result generalizes Cheeger's Hausdorff dimension estimate for the singular set in [7] and improve it into Minkowski dimension estimate in the spirit of C

Ergun Akleman, Shubham Agarwall, Donald H. House, Tolga Talha Yildiz

In color spaces where the chromatic term is given in polar coordinates, the shortest distance between colors of the same value is circular. By converting such a space into a complex polar form with a real-valued value axis, a color algebra for combining colors is immediately available. In this work, we introduce two complex space operations utilizing this ob

Md Manjurul Ahsan, Shivakumar Raman, Zahed Siddique

Additive manufacturing (AM) is gaining attention across various industries like healthcare, aerospace, and automotive. However, identifying defects early in the AM process can reduce production costs and improve productivity - a key challenge. This study explored the effectiveness of machine learning (ML) approaches, specifically transfer learning (TL) model

Yuan-Heng Wang, Hoshin V. Gupta

Although decades of effort have been devoted to building Physical-Conceptual (PC) models for predicting the time-series evolution of geoscientific systems, recent work shows that Machine Learning (ML) based Gated Recurrent Neural Network technology can be used to develop models that are much more accurate. However, the difficulty of extracting physical under

Federica Maria Surace, Pierre Fromholz, Francesco Scazza, Marcello Dalmonte

We propose a protocol for the scalable quantum simulation of SU($N$)$\times$U(1) lattice gauge theories with alkaline-earth like atoms in optical lattices in both one- and two-dimensional systems. The protocol exploits the combination of naturally occurring SU($N$) pseudo-spin symmetry and strong inter-orbital interactions that is unique to such atomic speci

Panagiotis Kotetes, Yun-Peng Huang

We propose an alternative route to stabilize magnetic skyrmions which does not require Dzyaloshinkii-Moriya interactions, magnetic anisotropy, or an external Zeeman field. Our so-called magnetic skyrmion catalysis (MSC) solely relies on the emergence of flux in the system's ground state. We review scenarios that allow for a nonzero flux and summarize the mag

Kenta Taguchi, Keiichi Maeda, Hiroyuki Maehara, Akito Tajitsu

The lowest possible mass of ONeMg white dwarfs (WDs) has not been clarified despite its importance in the formation and evolution of WDs. We tackle this issue by studying the properties of V1405 Cas (Nova Cassiopeiae 2021), which is an outlier given a combination of its very slow light-curve evolution and the recently reported neon-nova identification. We re

Charlie Duclut, Stefano Bo, Ruben Lier, Jay Armas

Odd viscoelastic materials are constrained by fewer symmetries than their even counterparts. The breaking of these symmetries allow these materials to exhibit different features, which have attracted considerable attention in recent years. Immersing a bead in such complex fluids allows for probing their physical properties, highlighting signatures of their o

Shengqi Sang, Yijian Zou, Timothy H. Hsieh

Open system quantum dynamics can generate a variety of long-range entangled mixed states, yet it has been unclear in what sense they constitute phases of matter. To establish that two mixed states are in the same phase, as defined by their two-way connectivity via local quantum channels, we use the renormalization group (RG) and decoders of quantum error cor

Alexandre M. Pombo, Daniela D. Doneva

It was recently found that in certain flavours of scalar-Gauss-Bonnet gravity linearly stable bald black holes can co-exist with stable scalarized solutions. The transition between both can be ignited by a large nonlinear perturbation, thus the process was dubbed non-linear scalarization, and it happens with a jump that leads to interesting astrophysical imp

David Grant, Nikole K. Lewis, Hannah R. Wakeford, Natasha E. Batalha

Clouds are prevalent in many of the exoplanet atmospheres that have been observed to date. For transiting exoplanets, we know if clouds are present because they mute spectral features and cause wavelength-dependent scattering. While the exact composition of these clouds is largely unknown, this information is vital to understanding the chemistry and energy b

Yue Pan, Andrey Kravtsov

We investigate the impact of bursty star formation on several galaxy scaling relations of dwarf galaxies using the $\texttt{GRUMPY}$ galaxy formation model. While this model reproduces the star formation rate (SFR)-stellar mass, stellar mass-gas mass, and stellar mass-metallicity relations, the scatter of these relations in the original model is smaller than

Máté Farkas

Device-independent quantum key distribution allows for proving the security of a shared cryptographic key between two distant parties with potentially untrusted devices. The security proof is based on the measurement outcome statistics (correlation) of a Bell experiment, and security is guaranteed by the laws of quantum theory. While it is known that the obs

ChangHoon Hahn, Francisco Villaescusa-Navarro, Peter Melchior, Romain Teyssier

We present the first cosmological constraints using only the observed photometry of galaxies. Villaescusa-Navarro et al. (2022; arXiv:2201.02202) recently demonstrated that the internal physical properties of a single simulated galaxy contain a significant amount of cosmological information. These physical properties, however, cannot be directly measured fro

Hamsa Padmanabhan, Abraham Loeb

We predict the X-ray background (XRB) expected from the population of quasars detected by the JWST spectroscopic surveys over the redshift range $z \sim 4-7$. We find that the measured UV emissivities, in combination with a best-fitting quasar SED template, imply a $\sim 10$ times higher unresolved X-ray background than constrained by current experiments. We

Jean-François Fortin, Wen-Jie Ma, Sarthak Parikh, Lorenzo Quintavalle

We establish that all of the one- and two-dimensional global conformal blocks are, up to some choice of prefactor, free-particle wavefunctions in tensor products of AdS$_3$ or limits thereof. Our first core observation is that the six-point comb-channel conformal blocks correspond to free-particle wavefunctions on an AdS$_3$ constructed directly in cross-rat

Krista Lynne Smith, Lia Sartori

The sampling strategy of the Transiting Exoplanet Survey Satellite (TESS) make TESS light curves extremely valuable to investigate high cadence optical variability of AGN. However, because the TESS instrument was primarily designed for exoplanet science, the use of the satellite for other applications requires careful treatment of the data. In this paper we

Tommaso Faleo, Eric Brunner, Jonathan W. Webb, Alexander Pickston

Entanglement and interference are both hallmark effects of quantum physics. Particularly rich dynamics arise when multiple (at least partially) indistinguishable particles are subjected to either of these phenomena. By combining both entanglement and many-particle interference, we propose an interferometric setting through which N-particle interference can b

Aaron Hui

Johnson noise thermometry enables direct measurement of the electron temperature, a valuable probe of many-body systems. Practical use of this technique calls for non-equilibrium generalizations of the Johnson-Nyquist theorem. For a hydrodynamic Corbino device, however, a na\"ive use of the Shockley-Ramo theorem alongside the "Corbino paradox" leads to yet a

Emil Mottola, Andrey V. Sadofyev, Andreas Stergiou

An effective field theory (EFT) for dynamical axions in Weyl semimetals (WSMs) is presented. A pseudoscalar axion excitation is predicted in WSMs at sufficiently low temperatures, independently of the strength of the Weyl fermion self-coupling. For strong fermion self-coupling the axion is the gapless Goldstone boson of chiral $U(1)^{\text{ch}}$ spontaneous

Vladislav Borisov

Solid state theory, density functional theory and its generalizations for correlated systems together with numerical simulations on supercomputers allow nowadays to model magnetic systems realistically and in detail and can be even used to predict new materials, paving the way for more rapid material development for applications in energy storage and convers

Mainak Mukhopadhyay, Manibrata Sen

Neutrino propagation through a turbulent medium can be highly non-adiabatic leading to distinct signatures in the survival probabilities. A core-collapse supernova can be host to a number of hydrodynamic instabilities which occur behind the shockfront. Such instabilities between the forward shock and a possible reverse shock can lead to cascades introducing

Thomas Vandermeulen

Gauging a symmetry can be thought of as the insertion of a spacetime-filling defect. Accordingly, we regard each gaugeable symmetry in a theory as defining a $-1$-form symmetry via condensation. The resulting operators, called gauge defects, have a natural fusion product, generally non-invertible, which we explore in a variety of two-dimensional theories.

Claude Duhr, Albrecht Klemm, Florian Loebbert, Christoph Nega

We analyse the family of Calabi-Yau varieties attached to four-point fishnet integrals in two dimensions. We find that the Picard-Fuchs operators for fishnet integrals are exterior powers of the Picard-Fuchs operators for ladder integrals. This implies that the periods of the Calabi-Yau varieties for fishnet integrals can be written as determinants of period

Griffin Hosseinzadeh, Kerry Paterson, Jillian C. Rastinejad, Manisha Shrestha

We present upgraded infrastructure for Searches after Gravitational Waves Using ARizona Observatories (SAGUARO) during LIGO, Virgo, and KAGRA's fourth gravitational-wave (GW) observing run (O4). These upgrades implement many of the lessons we learned after a comprehensive analysis of potential electromagnetic counterparts to the GWs discovered during the pre

J. Moran-Fraile, F. K. Roepke, R. Pakmor, M. A. Aloy

The merger of a white dwarf (WD) and a neutron star (NS) is a relatively common event that will produce an observable electromagnetic signal. Furthermore, the compactness of these stellar objects makes them an interesting candidate for gravitational wave (GW) astronomy, potentially being in the frequency range of LISA and other missions. To date, three-dimen

Prerak Garg, Desika Narayanan, Ryan L. Sanders, Romeel Davè

The ratios of strong rest-frame optical emission lines are the dominant indicator of metallicities in high-redshift galaxies. Since typical strong-line based metallicity indicators are calibrated on auroral lines at $z=0$, their applicability for galaxies in the distant Universe is unclear. In this paper, we make use of mock emission line data from cosmologi

Bin B. Ren, Myriam Benisty, Christian Ginski, Ryo Tazaki

Diverse protoplanetary disk morphology can result from planet-disk interaction, suggesting planetary presence. To date, most scattered light imaging campaigns have probed polarized light, which is only a fraction of the total light and not very sensitive to planets. To observe and characterize protoplanetary disk systems in the near-infrared in both polarize

Jingkang Yang, Yuhao Dong, Shuai Liu, Bo Li

Large vision-language models (VLMs) have achieved substantial progress in multimodal perception and reasoning. When integrated into an embodied agent, existing embodied VLM works either output detailed action sequences at the manipulation level or only provide plans at an abstract level, leaving a gap between high-level planning and real-world manipulation.

Xiaoming Zhao, Alex Colburn, Fangchang Ma, Miguel Angel Bautista

Rendering scenes observed in a monocular video from novel viewpoints is a challenging problem. For static scenes the community has studied both scene-specific optimization techniques, which optimize on every test scene, and generalized techniques, which only run a deep net forward pass on a test scene. In contrast, for dynamic scenes, scene-specific optimiza

Haoyi Zhu, Honghui Yang, Xiaoyang Wu, Di Huang

In contrast to numerous NLP and 2D vision foundational models, learning a 3D foundational model poses considerably greater challenges. This is primarily due to the inherent data variability and diversity of downstream tasks. In this paper, we introduce a novel universal 3D pre-training framework designed to facilitate the acquisition of efficient 3D represen

Haotong Lin, Sida Peng, Zhen Xu, Tao Xie

This paper aims to tackle the challenge of dynamic view synthesis from multi-view videos. The key observation is that while previous grid-based methods offer consistent rendering, they fall short in capturing appearance details of a complex dynamic scene, a domain where multi-view image-based rendering methods demonstrate the opposite properties. To combine

Shashanka Venkataramanan, Mamshad Nayeem Rizve, João Carreira, Yuki M. Asano

Self-supervised learning has unlocked the potential of scaling up pretraining to billions of images, since annotation is unnecessary. But are we making the best use of data? How more economical can we be? In this work, we attempt to answer this question by making two contributions. First, we investigate first-person videos and introduce a "Walking Tours" dat

Noshaba Cheema, Rui Xu, Nam Hee Kim, Perttu Hämäläinen

Virtual character animation and movement synthesis have advanced rapidly during recent years, especially through a combination of extensive motion capture datasets and machine learning. A remaining challenge is interactively simulating characters that fatigue when performing extended motions, which is indispensable for the realism of generated animations. Ho

Mengkang Hu, Yao Mu, Xinmiao Yu, Mingyu Ding

This paper studies close-loop task planning, which refers to the process of generating a sequence of skills (a plan) to accomplish a specific goal while adapting the plan based on real-time observations. Recently, prompting Large Language Models (LLMs) to generate actions iteratively has become a prevalent paradigm due to its superior performance and user-fr

Zichen Zhang, Yunshuang Li, Osbert Bastani, Abhishek Gupta

Real-world robotic tasks stretch over extended horizons and encompass multiple stages. Learning long-horizon manipulation tasks, however, is a long-standing challenge, and demands decomposing the overarching task into several manageable subtasks to facilitate policy learning and generalization to unseen tasks. Prior task decomposition methods require task-sp

Yiming Xie, Varun Jampani, Lei Zhong, Deqing Sun

We present a novel approach named OmniControl for incorporating flexible spatial control signals into a text-conditioned human motion generation model based on the diffusion process. Unlike previous methods that can only control the pelvis trajectory, OmniControl can incorporate flexible spatial control signals over different joints at different times with o

Xian Liu, Jian Ren, Aliaksandr Siarohin, Ivan Skorokhodov

Despite significant advances in large-scale text-to-image models, achieving hyper-realistic human image generation remains a desirable yet unsolved task. Existing models like Stable Diffusion and DALL-E 2 tend to generate human images with incoherent parts or unnatural poses. To tackle these challenges, our key insight is that human image is inherently struc

Hannah Lange, Fabian Döschl, Juan Carrasquilla, Annabelle Bohrdt

Numerically simulating spinful, fermionic systems is of great interest from the perspective of condensed matter physics. However, the exponential growth of the Hilbert space dimension with system size renders an exact parameterization of large quantum systems prohibitively demanding. This is a perfect playground for neural networks, owing to their immense re

Vishaal Udandarao, Max F. Burg, Samuel Albanie, Matthias Bethge

Recent advances in the development of vision-language models (VLMs) are yielding remarkable success in recognizing visual semantic content, including impressive instances of compositional image understanding. Here, we introduce the novel task of Visual Data-Type Identification, a basic perceptual skill with implications for data curation (e.g., noisy data-re

Po-Chen Ko, Jiayuan Mao, Yilun Du, Shao-Hua Sun

In this work, we present an approach to construct a video-based robot policy capable of reliably executing diverse tasks across different robots and environments from few video demonstrations without using any action annotations. Our method leverages images as a task-agnostic representation, encoding both the state and action information, and text as a gener

Philip A. Ernst, Dongzhou Huang

This paper begins with a study of the exact distribution of the empirical correlation of two independent AR(1) processes with Gaussian increments. We proceed to develop rates of convergence for the distribution of the scaled empirical correlation to the standard Gaussian distribution in both Wasserstein distance and Kolmogorov distance. Given $n$ data points

David Chuan-En Lin, Nikolas Martelaro

Recent advancements in AI foundation models have made it possible for them to be utilized off-the-shelf for creative tasks, including ideating design concepts or generating visual prototypes. However, integrating these models into the creative process can be challenging as they often exist as standalone applications tailored to specific tasks. To address thi

Siddhant Haldar, Lerrel Pinto

Unified models capable of solving a wide variety of tasks have gained traction in vision and NLP due to their ability to share regularities and structures across tasks, which improves individual task performance and reduces computational footprint. However, the impact of such models remains limited in embodied learning problems, which present unique challeng

Yu-Min Hu, Yin-Quan Huang, Wen-Tan Xue, Zhong Wang

One of the most pronounced non-Hermitian phenomena is the non-Hermitian skin effect, which refers to the exponential localization of bulk eigenstates near the boundaries of non-Hermitian systems. Whereas non-Bloch band theory has been developed to describe the non-Hermitian skin effect in lattice systems, its counterpart in continuum systems still lacks a qu

Jan Dubiński, Stanisław Pawlak, Franziska Boenisch, Tomasz Trzciński

Machine Learning as a Service (MLaaS) APIs provide ready-to-use and high-utility encoders that generate vector representations for given inputs. Since these encoders are very costly to train, they become lucrative targets for model stealing attacks during which an adversary leverages query access to the API to replicate the encoder locally at a fraction of t

Łukasz Leżaj

We study a $d$-dimensional non-symmetric strictly $\alpha$-stable L\'{e}vy process $\mathbf{X}$, whose spherical density is bounded and bounded away from the origin. First, we give sharp two-sided estimates on the transition density of $\mathbf{X}$ killed when leaving an arbitrary $\kappa$-fat set. We apply these results to get the existence of the Yaglom li

Judah Goldfeder, John Sipple

Modern commercial Heating, Ventilation, and Air Conditioning (HVAC) devices form a complex and interconnected thermodynamic system with the building and outside weather conditions, and current setpoint control policies are not fully optimized for minimizing energy use and carbon emission. Given a suitable training environment, a Reinforcement Learning (RL) m

Omar El Housni, Rajan Udwani

Strategic product placement can have a strong influence on customer purchase behavior in physical stores as well as online platforms. Motivated by this, we consider the problem of optimizing the placement of substitutable products in designated display locations to maximize the expected revenue of the seller. We model the customer behavior as a two-stage pro

Anupam Mitra, Tameem Albash, Philip Daniel Blocher, Jun Takahashi

We study the tractability of classically simulating critical phenomena in the quench dynamics of one-dimensional transverse field Ising models (TFIMs) using highly truncated matrix product states (MPS). We focus on two paradigmatic examples: a dynamical quantum phase transition (DQPT) that occurs in nonintegrable long-range TFIMs, and the infinite-time corre

Licong Lin, Yu Bai, Song Mei

Large transformer models pretrained on offline reinforcement learning datasets have demonstrated remarkable in-context reinforcement learning (ICRL) capabilities, where they can make good decisions when prompted with interaction trajectories from unseen environments. However, when and how transformers can be trained to perform ICRL have not been theoreticall

Subash Neupane, Shaswata Mitra, Ivan A. Fernandez, Swayamjit Saha

Robotics and Artificial Intelligence (AI) have been inextricably intertwined since their inception. Today, AI-Robotics systems have become an integral part of our daily lives, from robotic vacuum cleaners to semi-autonomous cars. These systems are built upon three fundamental architectural elements: perception, navigation and planning, and control. However,

Steve Shkoller, Vlad Vicol

We construct a fundamental piece of the boundary of the maximal globally hyperbolic development (MGHD) of Cauchy data for the multi-dimensional compressible Euler equations, which is necessary for the local shock development problem. For an open set of compressive and generic $H^7$ initial data, we construct unique $H^7$ solutions to the Euler equations in t

Deborah Oliveros, Érika Roldán, Pablo Soberón, Antonio J. Torres

Given a finite set of points in $\mathbb{R}^d$, Tverberg's theorem guarantees the existence of partitions of this set into parts whose convex hulls intersect. We introduce a graph structured on the family of Tverberg partitions of a given set of points, whose edges describe closeness between different Tverberg partitions. We prove bounds on the minimum and m

Giacomo Borghi, Lorenzo Pareschi

Genetic Algorithms (GA) are a class of metaheuristic global optimization methods inspired by the process of natural selection among individuals in a population. Despite their widespread use, a comprehensive theoretical analysis of these methods remains challenging due to the complexity of the heuristic mechanisms involved. In this work, relying on the tools

R. Veillet, O. Venot, B. Sirjean, R. Bounaceur

We aimed to build a new and updated C0-C2 chemical network to study the CHON disequilibrium chemistry of warm and hot exoplanet atmospheres that relies on extensively validated and recent state-of-the-art combustion networks. The reliability range of this network was aimed for conditions between 500 - 2500 K and 100 - 10^-6 bar. We compared the predictions o

Charles Packer, Sarah Wooders, Kevin Lin, Vivian Fang

Large language models (LLMs) have revolutionized AI, but are constrained by limited context windows, hindering their utility in tasks like extended conversations and document analysis. To enable using context beyond limited context windows, we propose virtual context management, a technique drawing inspiration from hierarchical memory systems in traditional

Linlu Qiu, Liwei Jiang, Ximing Lu, Melanie Sclar

The ability to derive underlying principles from a handful of observations and then generalize to novel situations -- known as inductive reasoning -- is central to human intelligence. Prior work suggests that language models (LMs) often fall short on inductive reasoning, despite achieving impressive success on research benchmarks. In this work, we conduct a

Max Sobol Mark, Archit Sharma, Fahim Tajwar, Rafael Rafailov

It is desirable for policies to optimistically explore new states and behaviors during online reinforcement learning (RL) or fine-tuning, especially when prior offline data does not provide enough state coverage. However, exploration bonuses can bias the learned policy, and our experiments find that naive, yet standard use of such bonuses can fail to recover

Peter Cox, Tony Gherghetta, Arpon Paul

We identify the QCD axion and right-handed (sterile) neutrinos as bound states of an $SU(5)$ chiral gauge theory with Peccei-Quinn (PQ) symmetry arising as a global symmetry of the strong dynamics. The strong dynamics is assumed to spontaneously break the PQ symmetry, producing a high-quality axion and naturally generating Majorana masses for the right-hande

Serhii Kryhin, Qiantan Hong, Leonid Levitov

Linear temperature dependence of transport coefficients in metals is often ascribed to non-Fermi-liquid physics. Here we demonstrate the $T$-linear behavior of nonlocal conductivity in a clean 2D electron fluid, where carrier collisions assist conduction and lead to hydrodynamic transport with conductance rather than resistance growing with temperature. The