Search

The holographic correspondence predicts that certain strongly coupled quantum systems describe an emergent, higher-dimensional bulk spacetime in which excitations enjoy local dynamics. We consider a general holographic state dual to an asymptotically AdS bulk spacetime, and study boundary correlation functions of local fields integrated against wavepackets. We derive a factorization formula showing that when the wavepackets suitably meet at a common bulk point, the boundary correlators develop sharp features controlled by flat-space-like bulk scattering processes. These features extend along boundary hyperboloids whose shape naturally reveals the bulk geometry. We discuss different choices of operator ordering, which lead to inclusive and out-of-time-ordered amplitudes, as well as fields of various spins and masses., Comment: 61 pages, 10 figures

This paper reviews object detection methods for finding small objects from remote sensing imagery and provides an empirical evaluation of four state-of-the-art methods to gain insights into method performance and technical challenges. In particular, we use car detection from urban satellite images and bee box detection from satellite images of agricultural lands as application scenarios. Drawing from the existing surveys and literature, we identify several top-performing methods for the empirical study. Public, high-resolution satellite image datasets are used in our experiments.

This study addresses the need for accurate and efficient object detection in assistive technologies for visually impaired individuals. We evaluate four real-time object detection algorithms YOLO, SSD, Faster R-CNN, and Mask R-CNN within the context of indoor navigation assistance. Using the Indoor Objects Detection dataset, we analyze detection accuracy, processing speed, and adaptability to indoor environments. Our findings highlight the trade-offs between precision and efficiency, offering insights into selecting optimal algorithms for realtime assistive navigation. This research advances adaptive machine learning applications, enhancing indoor navigation solutions for the visually impaired and promoting accessibility., Comment: 5 pages, 2 figures, 3 tables

Motivated by the holographic correspondence, we study the boundary imprint of bulk lightcones in spacetimes with boundaries. These lightcones can be observed whenever a localized event takes place in the bulk. The associated boundary surfaces (hyperboloids) reveal the bulk conformal metric. We work out a Hamilton-Jacobi description of these surfaces and analyze them in explicit examples. Bulk causality translates into a boundary inclusion property from which the bulk geodesic equation can be derived under some assumptions., Comment: 23 pages, 6 figures

Fix an integer $s \ge 2$. Let $\mathcal{P}$ be a set of $n$ points and let $\mathcal{L}$ be a set of lines in a linear space such that no line in $\mathcal{L}$ contains more than $(n-1)/(s-1)$ points of $\mathcal{P}$. Suppose that for every $s$-set $S$ in $\mathcal{P}$, there is a pair of points in $S$ that lies in a line from $\mathcal{L}$. We prove that $|\mathcal{L}| \ge (n-1)/(s-1)+s-1$ for $n$ large, and this is sharp when $n-1$ is a multiple of $s-1$. This generalizes the de Bruijn-Erd\H os theorem which is the case $s=2$. Our result is proved in the more general setting of linear hypergraphs.

The rapid evolution of communication technologies, compounded by recent geopolitical events such as the Viasat cyberattack in February 2022, has highlighted the urgent need for fast and reliable satellite missions for military and civil security operations. Consequently, this paper examines two Earth observation (EO) missions: one utilizing a single low Earth orbit (LEO) satellite and another through a network of LEO satellites, employing a secure-by-component design strategy. This approach begins by defining the scope of technical security engineering, decomposing the system into components and data flows, and enumerating attack surfaces. Then it proceeds by identifying threats to low-level components, applying secure-by-design principles, redesigning components into secure blocks in alignment with the Space Attack Research & Tactic Analysis (SPARTA) framework, and crafting shall statements to refactor the system design, with a particular focus on improving the security of the link segment.

In recent years, advances in artificial intelligence (AI) have transformed structural biology, particularly protein structure prediction. Though AI-based methods, such as AlphaFold (AF), often predict single conformations of proteins with high accuracy and confidence, predictions of alternative folds are often inaccurate, low-confidence, or simply not predicted at all. Here, we review three blind spots that alternative conformations reveal about AF-based protein structure prediction. First, proteins that assume conformations distinct from their training-set homologs can be mispredicted. Second, AF overrelies on its training set to predict alternative conformations. Third, degeneracies in pairwise representations can lead to high-confidence predictions inconsistent with experiment. These weaknesses suggest approaches to predict alternative folds more reliably.

Predicting future disease progression risk from medical images is challenging due to patient heterogeneity, and subtle or unknown imaging biomarkers. Moreover, deep learning (DL) methods for survival analysis are susceptible to image domain shifts across scanners. We tackle these issues in the task of predicting late dry Age-related Macular Degeneration (dAMD) onset from retinal OCT scans. We propose a novel DL method for survival prediction to jointly predict from the current scan a risk score, inversely related to time-to-conversion, and the probability of conversion within a time interval $t$. It uses a family of parallel hyperplanes generated by parameterizing the bias term as a function of $t$. In addition, we develop unsupervised losses based on intra-subject image pairs to ensure that risk scores increase over time and that future conversion predictions are consistent with AMD stage prediction using actual scans of future visits. Such losses enable data-efficient fine-tuning of the trained model on new unlabeled datasets acquired with a different scanner. Extensive evaluation on two large datasets acquired with different scanners resulted in a mean AUROCs of 0.82 for Dataset-1 and 0.83 for Dataset-2, across prediction intervals of 6,12 and 24 months., Comment: accepted in MICCAI 2024

Resource sharing is a crucial part of a multi-robot system. We propose a Boolean satisfiability based approach to resource sharing. Our key contributions are an algorithm for converting any constrained assignment to a weighted-SAT based optimization. We propose a theorem that allows optimal resource assignment problems to be solved via repeated application of a SAT solver. Additionally we show a way to encode continuous time ordering constraints using Conjunctive Normal Form (CNF). We benchmark our new algorithms and show that they can be used in an ad-hoc setting. We test our algorithms on a fleet of simulated and real world robots and show that the algorithms are able to handle real world situations. Our algorithms and test harnesses are opensource and build on Open-RMFs fleet management system., Comment: IROS 2024

The double-cone geometry is a saddle of the gravitational path integral, which explains the chaotic statistics of the spectrum of black hole microstates. This geometry is the usual AdS-Schwarzschild black hole, but with a periodic identification of the time coordinate; the resulting singularity at the black hole horizon is regulated by making the geometry slightly complex. Here, we consider generalizations of the double-cone geometry which include the Lorentzian cosmology that sits between the event horizon and the black hole singularity. We analyze this in two and three dimensions, where the cosmology has compact spatial sections and big bang/crunch singularities. These singularities are regulated in the same way by slightly complexifying the metric. We show that this is possible while satisfying the Kontsevich-Segal criterion, implying that these geometries can be interpreted as perturbatively stable saddle points in general relativity. This procedure leads to a novel description of the cosmology in terms of standard observables in the dual boundary CFT. In three dimensions, the cosmological solution gives a new contribution to the two-point function of the density of states in the boundary CFT. Unlike the usual double cone, it describes correlations between black hole microstates with different masses, and in a limit describes correlations between the statistics of heavy states and states near the BTZ threshold., Comment: 27 pages, 8 figures; v2: minor corrections; v3: additional minor corrections, major revisions in the appendix

Vertical Federated Learning (VFL) is a machine learning paradigm for learning from vertically partitioned data (i.e. features for each input are distributed across multiple "guest" clients and an aggregating "host" server owns labels) without communicating raw data. Traditionally, VFL involves an "entity resolution" phase where the host identifies and serializes the unique entities known to all guests. This is followed by private set intersection to find common entities, and an "entity alignment" step to ensure all guests are always processing the same entity's data. However, using only data of entities from the intersection means guests discard potentially useful data. Besides, the effect on privacy is dubious and these operations are computationally expensive. We propose a novel approach that eliminates the need for set intersection and entity alignment in categorical tasks. Our Entity Augmentation technique generates meaningful labels for activations sent to the host, regardless of their originating entity, enabling efficient VFL without explicit entity alignment. With limited overlap between training data, this approach performs substantially better (e.g. with 5% overlap, 48.1% vs 69.48% test accuracy on CIFAR-10). In fact, thanks to the regularizing effect, our model performs marginally better even with 100% overlap., Comment: GLOW @ IJCAI 2024 (12 pages + 2 page bibliography. 15 figures.)

Understanding how local distortions determine the functional properties of high entropy materials, containing five or more elements at a crystallographic site, is an open challenge. We address this for a compositionally complex spinel oxide (Mn$_{0.2}$Co$_{0.2}$Ni$_{0.2}$Cu$_{0.2}$Zn$_{0.2}$)Cr$_2$O$_4$ ($A^5$Cr$_2$O$_4$). By comparatively examining extended X-ray absorption fine structure on $A^5$Cr$_2$O$_4$ and its parent counterparts $A$Cr$_2$O$_4$ along with density functional theory calculations for multiple configurations, we find that the element-specific distortions go beyond the first neighbor. Specifically, the strong Jahn-Teller distortion present in CuCr$_2$O$_4$ is found to be completely suppressed in $A^5$Cr$_2$O$_4$. Instead, there is a broad distribution of Cu-O and Cu-Cr bond distances while other $A$-O distances acquire certain specific values. This study demonstrates the additional flexibility of a cationic sublattice in maintaining a uniform long-range structure, in contrast to previous reports showing only the accommodative anionic sublattice. Remarkably, despite the presence of multiple magnetic ions and variable bond lengths, the mean field magnetic interactions of $A^5$Cr$_2$O$_4$ exhibit a striking resemblance to those of NiCr$_2$O$_4$. This compelling observation originates from the comparability of bond lengths around Cr in both materials. Our study paves the way for a deeper understanding of the impact of local structural distortions in compositionally complex quantum materials, enabling the targeted design with tailored properties.

Contrastive pretraining provides robust representations by ensuring their invariance to different image transformations while simultaneously preventing representational collapse. Equivariant contrastive learning, on the other hand, provides representations sensitive to specific image transformations while remaining invariant to others. By introducing equivariance to time-induced transformations, such as disease-related anatomical changes in longitudinal imaging, the model can effectively capture such changes in the representation space. In this work, we propose a Time-equivariant Contrastive Learning (TC) method. First, an encoder embeds two unlabeled scans from different time points of the same patient into the representation space. Next, a temporal equivariance module is trained to predict the representation of a later visit based on the representation from one of the previous visits and the corresponding time interval with a novel regularization loss term while preserving the invariance property to irrelevant image transformations. On a large longitudinal dataset, our model clearly outperforms existing equivariant contrastive methods in predicting progression from intermediate age-related macular degeneration (AMD) to advanced wet-AMD within a specified time-window., Comment: Accepted at MICCAI 2024 (early accept, top 11%)

Recent transformer-based ASR models have achieved word-error rates (WER) below 4%, surpassing human annotator accuracy, yet they demand extensive server resources, contributing to significant carbon footprints. The traditional server-based architecture of ASR also presents privacy concerns, alongside reliability and latency issues due to network dependencies. In contrast, on-device (edge) ASR enhances privacy, boosts performance, and promotes sustainability by effectively balancing energy use and accuracy for specific applications. This study examines the effects of quantization, memory demands, and energy consumption on the performance of various ASR model inference on the NVIDIA Jetson Orin Nano. By analyzing WER and transcription speed across models using FP32, FP16, and INT8 quantization on clean and noisy datasets, we highlight the crucial trade-offs between accuracy, speeds, quantization, energy efficiency, and memory needs. We found that changing precision from fp32 to fp16 halves the energy consumption for audio transcription across different models, with minimal performance degradation. A larger model size and number of parameters neither guarantees better resilience to noise, nor predicts the energy consumption for a given transcription load. These, along with several other findings offer novel insights for optimizing ASR systems within energy- and memory-limited environments, crucial for the development of efficient on-device ASR solutions. The code and input data needed to reproduce the results in this article are open sourced are available on [https://github.com/zzadiues3338/ASR-energy-jetson].

We propose precise effective field theory criteria to obtain a four-dimensional de Sitter space within M-theory. To this effect, starting with the state space described by the action of metric perturbations, fluxes etc over the supersymmetric Minkowski vacuum in eleven-dimensions, we discuss the most general low energy effective action in terms of the eleven-dimensional fields including non-perturbative and non-local terms. Given this, our criteria to obtain a valid four-dimensional de Sitter solution at far IR involve satisfying the Schwinger-Dyson equations of the associated path integral, as well as obeying positivity constraints on the dual IIA string coupling and its time derivative. For excited states, the Schwinger-Dyson equations imply an effective emergent potential different from the original potential. We show that while vacuum solutions and arbitrary coherent states fail to satisfy these criteria, a specific class of excited states called the Glauber-Sudarshan states obey them. Using the resurgent structure of observables computed using the path integral over the Glauber-Sudarshan states, four-dimensional de Sitter in the flat slicing can be constructed using a Glauber-Sudarshan state in M-theory. Among other novel results, we discuss the smallness of the positive cosmological constant, including the curious case where the cosmological constant is very slowly varying with time. We also discuss the resolution of identity with the Glauber-Sudarshan states, generation and the convergence properties of the non-perturbative and the non-local effects, the problems with the static patch and other related topics. We analyze briefly the issues related to the compatibility of the Wilsonian effective action with Borel resummations and discuss how they influence the effective field theory description in a four-dimensional de Sitter space., Comment: 145 pages, 16 figures; a fairly non-technical summary of our work appears in section 8.1; v2: Typos corrected and references updated; v3: Typos corrected, text elaborated at various places and references updated. Final version to appear in JHEP

We analyze a random greedy process to construct $q$-uniform linear hypergraphs using the differential equation method. We show for $q=o(\sqrt{\log n})$, that this process yields a hypergraph with $\frac{n(n-1)}{q(q-1)}(1-o(1))$ edges. We also give some bounds for maximal linear hypergraphs.