Search

Data scientists often seek to identify the most important features in high-dimensional datasets. This can be done through $L_1$-regularized regression, but this can become inefficient for very high-dimensional datasets. Additionally, high-dimensional regression can leak information about individual datapoints in a dataset. In this paper, we empirically evaluate the established baseline method for feature selection with differential privacy, the two-stage selection technique, and show that it is not stable under sparsity. This makes it perform poorly on real-world datasets, so we consider a different approach to private feature selection. We employ a correlations-based order statistic to choose important features from a dataset and privatize them to ensure that the results do not leak information about individual datapoints. We find that our method significantly outperforms the established baseline for private feature selection on many datasets., Comment: To appear in Proceedings of the 17th ACM Workshop on Artificial Intelligence and Security, 2024

Monolayer graphene under a strong magnetic field near charge neutrality manifests the integer and fractional quantum Hall effects. Since only some of the four spin/valley flavors available to the electrons in each Landau level manifold are filled, they also exhibit spontaneous symmetry breaking the in spin/valley sector, a phenomenon known as quantum Hall ferromagnetism. In this work, we study quantum Hall ferromagnets in the higher Landau level manifolds of monolayer graphene and show that there is an even richer set of symmetry-broken phases than in the lowest Landau level manifold. Specifically, both valley polarized and valley equatorial (where the occupied Landau levels are in an equal superposition of both valleys) ferromagnets, antiferromagnets, and canted antiferromagnets are found. Several types of spin valley entangled phases are found, all of which manifest the simultaneous spontaneous symmetry breaking of both magnetic and lattice symmetries., Comment: 27 pages, 18 figures

gwNRHME is a recently developed framework that seamlessly converts a multi-modal (i.e with several spherical harmonic modes) quasi-circular waveform into multi-modal eccentric waveform if the quadrupolar eccentric waveform is known. Here, we employ the gwNRHME framework to combine a multi-modal quasi-circular waveform model NRHybSur3dq8 and quadrupolar non-spinning eccentric waveform model EccentricIMR to construct multi-modal non-spinning eccentric model NRHybSur3dq8-gwNRHME. Using a total of 35 eccentric numerical relativity (NR) simulations obtained from the SXS and RIT catalogs, we demonstrate that NRHybSur3dq8-gwNRHME model predictions agree well with NR (with typical relative $L_2$ errors of ~0.01 for the dominant quadrupolar mode) for mass ratios $ 1 \leq q \leq 4$ and eccentricities up to ~0.2 measured about 10 cycles before the merger. To demonstrate the modularity of the gwNRHME framework, we further combine EccentricIMR with BHPTNRSur1dq1e4 model and develop a non-spinning eccentric models named BHPTNRSur1dq1e4-gwNRHME. Finally, we develop a different variant of these models by replacing EccentricIMR with EccentricTD. Both the gwNRHME framework and associated models are available through the gwModels package., Comment: 14 pages, 11 figures

We analyse the three-dimensional structure and kinematics of two samples of young stars in the Galactic disc, containing respectively young giants ($\sim$16000 stars out to heliocentric distances of $\sim$7 kpc) and classical Cepheids ($\sim$3400 stars out to heliocentric distances of $\sim$15 kpc). Both samples show evidence of a large-scale vertical corrugation on top of the warp of the Milky Way, which has a vertical height of 150-200 pc, a radial width of about 3 kpc, and a total length of at least 10 kpc, possibly reaching 20 kpc with the Cepheid sample. The stars in the corrugation exhibit both radial and vertical systematic motions, with Galactocentric radial velocities towards the outer disc of about 10-15 km/s. In the vertical motions, once the warp signature is subtracted, the residuals show a large-scale feature of systematically positive vertical velocities, which is located radially outwards with respect to the corrugation, and whose line of maxima approximately coincides with the line of null vertical displacement, consistent with a vertical wave propagating towards the outer parts of the Galactic disc., Comment: 18 pages, 19 figures. Submitted to A&A. We welcome comments, questions, suggestions for missing references, etc

We present BHPTNRSur2dq1e3, a reduced order surrogate model of gravitational waves emitted from binary black hole (BBH) systems in the comparable to large mass ratio regime with aligned spin ($\chi_1$) on the heavier mass ($m_1$). We trained this model on waveform data generated from point particle black hole perturbation theory (ppBHPT) with mass ratios varying from $3 \leq q \leq 1000$ and spins from $-0.8 \leq \chi_1 \leq 0.8$. The waveforms are $13,500 \ m_1$ long and include all spin-weighted spherical harmonic modes up to $\ell = 4$ except the $(4,1)$ and $m = 0$ modes. We find that for binaries with $\chi_1 \lesssim -0.5$, retrograde quasi-normal modes are significantly excited, thereby complicating the modeling process. To overcome this issue, we introduce a domain decomposition approach to model the inspiral and merger-ringdown portion of the signal separately. The resulting model can faithfully reproduce ppBHPT waveforms with a median time-domain mismatch error of $8 \times 10^{-5}$. We then calibrate our model with numerical relativity (NR) data in the comparable mass regime $(3 \leq q \leq 10)$. By comparing with spin-aligned BBH NR simulations at $q = 15$, we find that the dominant quadrupolar (subdominant) modes agree to better than $\approx 10^{-3} \ (\approx 10^{-2})$ when using a time-domain mismatch error, where the largest source of calibration error comes from the transition-to-plunge and ringdown approximations of perturbation theory. Mismatch errors are below $\approx 10^{-2}$ for systems with mass ratios between $6 \leq q \leq 15$ and typically get smaller at larger mass ratio. Our two models - both the ppBHPT waveform model and the NR-calibrated ppBHPT model - will be publicly available through gwsurrogate and the Black Hole Perturbation Toolkit packages., Comment: 20 pages, 17 figures

Using adiabatic point-particle black hole perturbation theory, we simulate plausible gravitational wave~(GW) signatures in two exotic scenarios (i) where a small black hole is emitted by a larger one ('black hole emission') and (ii) where a small black hole is emitted by a larger one and subsequently absorbed back ('black hole absorption'). While such scenarios are forbidden in general relativity!(GR), alternative theories (such as certain quantum gravity scenarios obeying the weak gravity conjecture, white holes, and Hawking radiation) may allow them. By leveraging the phenomenology of black hole emission and absorption signals, we introduce straightforward modifications to existing gravitational waveform models to mimic gravitational radiation associated with these exotic events. We anticipate that these (incomplete but) initial simulations, coupled with the adjusted waveform models, will aid in the development of null tests for GR using GWs., Comment: 12 pages, 7 figures

COVID 19 pandemic has disrupted the global market and workplace landscape. As a response, hybrid work situations have become popular in the software business sector. This way of working has an impact on software companies. This study investigates software companies responses to hybrid working. We conducted a large scale survey to achieve our objective. Our results are based on a qualitative analysis of 124 valid responses. The main result of our study is a taxonomy of software companies impacts on hybrid working at individual, team and organisation levels. We found higher positive responses at individual and organisational levels than negative responses. At the team level, both positive and negative impacts obtained a uniform number of responses. The results indicate that hybrid working became credible with the wave of COVID 19, with 83 positive responses outweighing the 41 negative responses. Software company respondents witnessed better work-life balance, productivity, and efficiency in hybrid working.

In order to develop robots that can effectively serve as versatile and capable home assistants, it is crucial for them to reliably perceive and interact with a wide variety of objects across diverse environments. To this end, we proposed Open Vocabulary Mobile Manipulation as a key benchmark task for robotics: finding any object in a novel environment and placing it on any receptacle surface within that environment. We organized a NeurIPS 2023 competition featuring both simulation and real-world components to evaluate solutions to this task. Our baselines on the most challenging version of this task, using real perception in simulation, achieved only an 0.8% success rate; by the end of the competition, the best participants achieved an 10.8\% success rate, a 13x improvement. We observed that the most successful teams employed a variety of methods, yet two common threads emerged among the best solutions: enhancing error detection and recovery, and improving the integration of perception with decision-making processes. In this paper, we detail the results and methodologies used, both in simulation and real-world settings. We discuss the lessons learned and their implications for future research. Additionally, we compare performance in real and simulated environments, emphasizing the necessity for robust generalization to novel settings.

Various generalizations of the scalar, axisymmetric Aretakis "horizon hair" for extremal black holes have recently appeared in the literature. In this paper, we present an expression for a non-axisymmetric Aretakis gravitational charge and its potentially observable imprint at a finite distance from the horizon (Ori-coefficient) in extremal Kerr black hole backgrounds. We offer strong and robust numerical evidence for the existence of this hair and its properties. Specifically, we consider the time evolution of horizon penetrating, quadrupolar and (subdominant) octupolar gravitational perturbations with compact support on extremal Kerr (EK) spacetime. We do this by numerically solving the Teukolsky equation and determining the Aretakis charge values on the horizon and at a finite distance from the black hole., Comment: 7 pages, 9 figures

We investigate the late-time tail behavior in gravitational waves from merging eccentric binary black holes (BBH) using black hole perturbation theory. For simplicity, we focus only on the dominant quadrupolar mode of the radiation. We demonstrate that such tails become more prominent as eccentricity increases. Exploring the phenomenology of the tails in both spinning and non-spinning eccentric binaries, with the spin magnitude varying from $\chi=-0.6$ to $\chi=+0.6$ and eccentricity as high as $e=0.98$, we find that these tails can be well approximated by a slowly decaying power law. We study the power law for varying systems and find that the power law exponent lies close to the theoretically expected value $-4$. Finally, using both plunge geodesic and radiation-reaction-driven orbits, we perform a series of numerical experiments to understand the origin of the tails in BBH simulations. Our results suggest that the late-time tails are strongly excited in eccentric BBH systems when the smaller black hole is in the neighborhood of the apocenter, as opposed to any structure in the strong field of the larger black hole. Our analysis framework is publicly available through the \texttt{gwtails} Python package., Comment: 16 pages, 18 figures

A $(1 \pm \epsilon)$-sparsifier of a hypergraph $G(V,E)$ is a (weighted) subgraph that preserves the value of every cut to within a $(1 \pm \epsilon)$-factor. It is known that every hypergraph with $n$ vertices admits a $(1 \pm \epsilon)$-sparsifier with $\tilde{O}(n/\epsilon^2)$ hyperedges. In this work, we explore the task of building such a sparsifier by using only linear measurements (a \emph{linear sketch}) over the hyperedges of $G$, and provide nearly-matching upper and lower bounds for this task. Specifically, we show that there is a randomized linear sketch of size $\widetilde{O}(n r \log(m) / \epsilon^2)$ bits which with high probability contains sufficient information to recover a $(1 \pm \epsilon)$ cut-sparsifier with $\tilde{O}(n/\epsilon^2)$ hyperedges for any hypergraph with at most $m$ edges each of which has arity bounded by $r$. This immediately gives a dynamic streaming algorithm for hypergraph cut sparsification with an identical space complexity, improving on the previous best known bound of $\widetilde{O}(n r^2 \log^4(m) / \epsilon^2)$ bits of space (Guha, McGregor, and Tench, PODS 2015). We complement our algorithmic result above with a nearly-matching lower bound. We show that for every $\epsilon \in (0,1)$, one needs $\Omega(nr \log(m/n) / \log(n))$ bits to construct a $(1 \pm \epsilon)$-sparsifier via linear sketching, thus showing that our linear sketch achieves an optimal dependence on both $r$ and $\log(m)$.

Novel dataset creation for all multi-object tracking, crowd-counting, and industrial-based videos is arduous and time-consuming when faced with a unique class that densely populates a video sequence. We propose a time efficient method called POPCat that exploits the multi-target and temporal features of video data to produce a semi-supervised pipeline for segmentation or box-based video annotation. The method retains the accuracy level associated with human level annotation while generating a large volume of semi-supervised annotations for greater generalization. The method capitalizes on temporal features through the use of a particle tracker to expand the domain of human-provided target points. This is done through the use of a particle tracker to reassociate the initial points to a set of images that follow the labeled frame. A YOLO model is then trained with this generated data, and then rapidly infers on the target video. Evaluations are conducted on GMOT-40, AnimalTrack, and Visdrone-2019 benchmarks. These multi-target video tracking/detection sets contain multiple similar-looking targets, camera movements, and other features that would commonly be seen in "wild" situations. We specifically choose these difficult datasets to demonstrate the efficacy of the pipeline and for comparison purposes. The method applied on GMOT-40, AnimalTrack, and Visdrone shows a margin of improvement on recall/mAP50/mAP over the best results by a value of 24.5%/9.6%/4.8%, -/43.1%/27.8%, and 7.5%/9.4%/7.5% where metrics were collected., Comment: 10 pages, 5 figures, Accepted in "Conference on Robots and Vision 2024"

In a very recent breakthrough, Behnezhad and Ghafari [arXiv'24] developed a novel fully dynamic randomized algorithm for maintaining a $(1-\epsilon)$-approximation of maximum matching with amortized update time potentially much better than the trivial $O(n)$ update time. The runtime of the BG algorithm is parameterized via the following graph theoretical concept: * For any $n$, define $ORS(n)$ -- standing for Ordered RS Graph -- to be the largest number of edge-disjoint matchings $M_1,\ldots,M_t$ of size $\Theta(n)$ in an $n$-vertex graph such that for every $i \in [t]$, $M_i$ is an induced matching in the subgraph $M_{i} \cup M_{i+1} \cup \ldots \cup M_t$. Then, for any fixed $\epsilon > 0$, the BG algorithm runs in \[ O\left( \sqrt{n^{1+O(\epsilon)} \cdot ORS(n)} \right) \] amortized update time with high probability, even against an adaptive adversary. $ORS(n)$ is a close variant of a more well-known quantity regarding RS graphs (which require every matching to be induced regardless of the ordering). It is currently only known that $n^{o(1)} \leqslant ORS(n) \leqslant n^{1-o(1)}$, and closing this gap appears to be a notoriously challenging problem. In this work, we further strengthen the result of Behnezhad and Ghafari and push it to limit to obtain a randomized algorithm with amortized update time of \[ n^{o(1)} \cdot ORS(n) \] with high probability, even against an adaptive adversary. In the limit, i.e., if current lower bounds for $ORS(n) = n^{o(1)}$ are almost optimal, our algorithm achieves an $n^{o(1)}$ update time for $(1-\epsilon)$-approximation of maximum matching, almost fully resolving this fundamental question. In its current stage also, this fully reduces the algorithmic problem of designing dynamic matching algorithms to a purely combinatorial problem of upper bounding $ORS(n)$ with no algorithmic considerations., Comment: 24 pages, 2 figures

In this work, the Ginzburg-Landau theory is represented on a symplectic manifold with a phase space content. The order parameter is defined by a quasi-probability amplitude, which gives rise to a quasi-probability distribution function, i.e., a Wigner-type function. The starting point is the thermal group representation of Euclidean symmetries and gauge symmetry. Well-known basic results on the behavior of a superconductor are re-derived, providing the consistency of representation. The critical superconducting current density is determined and its usual behavior is inferred. The negativety factor associated with the quasi-distribution function is analyzed, providing information about the non-classicality nature of the superconductor state in the region closest to the edge of the superconducting material., Comment: 9 pages, 1 figures

Parameter-efficient fine-tuning (PEFT) techniques such as low-rank adaptation (LoRA) can effectively adapt large pre-trained foundation models to downstream tasks using only a small fraction (0.1%-10%) of the original trainable weights. An under-explored question of PEFT is in extending the pre-training phase without supervised labels; that is, can we adapt a pre-trained foundation model to a new domain via efficient self-supervised pre-training on this new domain? In this work, we introduce ExPLoRA, a highly effective technique to improve transfer learning of pre-trained vision transformers (ViTs) under domain shifts. Initializing a ViT with pre-trained weights on large, natural-image datasets such as from DinoV2 or MAE, ExPLoRA continues the unsupervised pre-training objective on a new domain. In this extended pre-training phase, ExPLoRA only unfreezes 1-2 pre-trained ViT blocks and all normalization layers, and then tunes all other layers with LoRA. Finally, we fine-tune the resulting model only with LoRA on this new domain for supervised learning. Our experiments demonstrate state-of-the-art results on satellite imagery, even outperforming fully pre-training and fine-tuning ViTs. Using the DinoV2 training objective, we demonstrate up to 7% improvement in linear probing top-1 accuracy on downstream tasks while using <10% of the number of parameters that are used in prior fully-tuned state-of-the art approaches. Our ablation studies confirm the efficacy of our approach over other baselines, including PEFT and simply unfreezing more transformer blocks.

As a relatively young and bright population, and the archetype of standard candles, classical Cepheids make an ideal population to trace non-axisymmetric structure in the young stellar disk to large distances. We use the new distances derived in Paper I based on mid-IR WISE photometry for a selected sample of 2857 dynamically young Cepheids to trace the spiral arms of the Milky Way. The Perseus and Sagittarius-Carina arms are clearly evident in the third and fourth Galactic quadrants, while the Local and Scutum arms are much weaker, with extinction severely limiting our view of the latter, inner-most spiral arm. Pitch angles are derived for each arm over various ranges of galactic azimuth, each covering at least 90{\deg} in azimuth. Our method of detecting spiral arms and deriving pitch angles does not rely on pre-assigning sources to specific arms. While spiral structure in the first and second quadrant is not obvious, in part due to extinction effects, it is not inconsistent with that seen in the third and fourth quadrants. In summary, the Cepheids allow us to map spiral structure in the third and fourth Galactic quadrants where there are currently few masers with astrometric parallaxes, thus significantly extending our picture of the Milky Way on large-scales., Comment: (10 pages, 9 figures, submitted)

Classical Cepheids are the archetype of the standard candle, thanks to the period-luminosity relation which allows to measure their intrinsic brightness. They are also relatively young and bright, potentially making them excellent tracers of the young stellar population that is responsible for shaping the visible aspect of our Galaxy, the Milky Way. However, being observers embedded in the dusty interstellar medium of the Galaxy, deriving reliable photometric distances to classical Cepheids of the Milky Way is a challenge. The typical approach is to use reddening-free indices, such as Wesenheit magnitudes, to obviate the need for an extinction correction. However, this approach is not reliable - especially toward the inner Galaxy - as its assumption of a universal total-to-selective extinction ratio is not satisfied, particularly in lines-of-sight where the extinction is high and crosses spiral arms. We instead estimate new distances for 3425 Cepheids based on mid-IR photometry from WISE, which suffers minimally from extinction, and by adopting a 3D extinction map to calculate the necessary (albeit small) extinction corrections. We show that our distances are consistent with Gaia's parallaxes for the subset with relative parallax errors smaller than 10%, verifying that our mean distance errors are of the order of 13% and that the mean parallax zero point for this subsample is 7 $\mu$as. The catalog of Cepheid distances is made available online., Comment: 21 pages, 21 figures, 3 tables; submitted to ApJS

In the era of Large Language Models (LLMs), predicting judicial outcomes poses significant challenges due to the complexity of legal proceedings and the scarcity of expert-annotated datasets. Addressing this, we introduce \textbf{Pred}iction with \textbf{Ex}planation (\texttt{PredEx}), the largest expert-annotated dataset for legal judgment prediction and explanation in the Indian context, featuring over 15,000 annotations. This groundbreaking corpus significantly enhances the training and evaluation of AI models in legal analysis, with innovations including the application of instruction tuning to LLMs. This method has markedly improved the predictive accuracy and explanatory depth of these models for legal judgments. We employed various transformer-based models, tailored for both general and Indian legal contexts. Through rigorous lexical, semantic, and expert assessments, our models effectively leverage \texttt{PredEx} to provide precise predictions and meaningful explanations, establishing it as a valuable benchmark for both the legal profession and the NLP community.

Decoherence is an undesirable, but ubiquitous phenomenon in quantum systems. Here, we study the effect of partial decoherence, induced via a B\"uttiker probe, on two-terminal electronic transport across one-dimensional quantum wires and rings, in both the linear and non-linear regimes. We find that dephasing causes backscattering when introduced locally in a ballistic channel. Further, we find that decoherence results in rectification when inversion is broken in the two-terminal transport set-up by a combination of a local dephasing centre and a static impurity. Interestingly, the rectification strength and even its direction varies strongly with the relative distance between the probe and the scatterer. We further analyze how decoherence affects characteristic quantum effects in electronic transport, such as, Fabry-P\'erot oscillations in double-barrier setups, and Aharonov-Bohm interference in one-dimensional rings, and find that the amplitude of oscillations in conductance is reduced by decoherence., Comment: 8 pages, 8 captioned figures

Maximum bipartite matching (MBM) is a fundamental problem in combinatorial optimization with a long and rich history. A classic result of Hopcroft and Karp (1973) provides an $O(m \sqrt{n})$-time algorithm for the problem, where $n$ and $m$ are the number of vertices and edges in the input graph, respectively. For dense graphs, an approach based on fast matrix multiplication achieves a running time of $O(n^{2.371})$. For several decades, these results represented state-of-the-art algorithms, until, in 2013, Madry introduced a powerful new approach for solving MBM using continuous optimization techniques. This line of research led to several spectacular results, culminating in a breakthrough $m^{1+o(1)}$-time algorithm for min-cost flow, that implies an $m^{1+o(1)}$-time algorithm for MBM as well. These striking advances naturally raise the question of whether combinatorial algorithms can match the performance of the algorithms that are based on continuous techniques for MBM. A recent work of the authors (2024) made progress on this question by giving a combinatorial $\tilde{O}(m^{1/3}n^{5/3})$-time algorithm for MBM, thus outperforming both the Hopcroft-Karp algorithm and matrix multiplication based approaches, on sufficiently dense graphs. Still, a large gap remains between the running time of their algorithm and the almost linear-time achievable by algorithms based on continuous techniques. In this work, we take another step towards narrowing this gap, and present a randomized $n^{2+o(1)}$-time combinatorial algorithm for MBM. Thus in dense graphs, our algorithm essentially matches the performance of algorithms that are based on continuous methods. We also obtain a randomized $n^{2+o(1)}$-time combinatorial algorithm for maximum vertex-capacitated $s$-$t$ flow in directed graphs when all vertex capacities are identical, using a standard reduction from this problem to MBM.

In this work, we test an effective-one-body radiation-reaction force for eccentric planar orbits of a test mass in a Kerr background, which contains third-order post-Newtonian (PN) non-spinning and second-order PN spin contributions. We compare the analytical fluxes connected to two different resummations of this force, truncated at different PN orders in the eccentric sector, with the numerical fluxes computed through the use of frequency- and time-domain Teukolsky-equation codes. We find that the different PN truncations of the radiation-reaction force show the expected scaling in the weak gravitational-field regime, and we observe a fractional difference with the numerical fluxes that is $<5 \%$, for orbits characterized by eccentricity $0 \le e \le 0.7$, central black-hole spin $-0.99 M \le a \le 0.99 M$ and fixed orbital-averaged quantity $x=\langle M\Omega \rangle^{2/3} = 0.06$, corresponding to the mildly strong-field regime with semilatera recta $9 M

In this paper, we propose SpotNet: a fast, single stage, image-centric but LiDAR anchored approach for long range 3D object detection. We demonstrate that our approach to LiDAR/image sensor fusion, combined with the joint learning of 2D and 3D detection tasks, can lead to accurate 3D object detection with very sparse LiDAR support. Unlike more recent bird's-eye-view (BEV) sensor-fusion methods which scale with range $r$ as $O(r^2)$, SpotNet scales as $O(1)$ with range. We argue that such an architecture is ideally suited to leverage each sensor's strength, i.e. semantic understanding from images and accurate range finding from LiDAR data. Finally we show that anchoring detections on LiDAR points removes the need to regress distances, and so the architecture is able to transfer from 2MP to 8MP resolution images without re-training.

We describe the design and results from the BraTS 2023 Intracranial Meningioma Segmentation Challenge. The BraTS Meningioma Challenge differed from prior BraTS Glioma challenges in that it focused on meningiomas, which are typically benign extra-axial tumors with diverse radiologic and anatomical presentation and a propensity for multiplicity. Nine participating teams each developed deep-learning automated segmentation models using image data from the largest multi-institutional systematically expert annotated multilabel multi-sequence meningioma MRI dataset to date, which included 1000 training set cases, 141 validation set cases, and 283 hidden test set cases. Each case included T2, T2/FLAIR, T1, and T1Gd brain MRI sequences with associated tumor compartment labels delineating enhancing tumor, non-enhancing tumor, and surrounding non-enhancing T2/FLAIR hyperintensity. Participant automated segmentation models were evaluated and ranked based on a scoring system evaluating lesion-wise metrics including dice similarity coefficient (DSC) and 95% Hausdorff Distance. The top ranked team had a lesion-wise median dice similarity coefficient (DSC) of 0.976, 0.976, and 0.964 for enhancing tumor, tumor core, and whole tumor, respectively and a corresponding average DSC of 0.899, 0.904, and 0.871, respectively. These results serve as state-of-the-art benchmarks for future pre-operative meningioma automated segmentation algorithms. Additionally, we found that 1286 of 1424 cases (90.3%) had at least 1 compartment voxel abutting the edge of the skull-stripped image edge, which requires further investigation into optimal pre-processing face anonymization steps., Comment: 16 pages, 11 tables, 10 figures, MICCAI

Medical image interpretation using deep learning has shown promise but often requires extensive expert-annotated datasets. To reduce this annotation burden, we develop an Image-Graph Contrastive Learning framework that pairs chest X-rays with structured report knowledge graphs automatically extracted from radiology notes. Our approach uniquely encodes the disconnected graph components via a relational graph convolution network and transformer attention. In experiments on the CheXpert dataset, this novel graph encoding strategy enabled the framework to outperform existing methods that use image-text contrastive learning in 1% linear evaluation and few-shot settings, while achieving comparable performance to radiologists. By exploiting unlabeled paired images and text, our framework demonstrates the potential of structured clinical insights to enhance contrastive learning for medical images. This work points toward reducing demands on medical experts for annotations, improving diagnostic precision, and advancing patient care through robust medical image understanding., Comment: Accepted into Machine Learning for Health (ML4H) 2023

In recent times Large Language Models have exhibited tremendous capabilities, especially in the areas of mathematics, code generation and general-purpose reasoning. However for specialized domains especially in applications that require parsing and analyzing large chunks of numeric or tabular data even state-of-the-art (SOTA) models struggle. In this paper, we introduce a new approach to solving domain-specific tabular data analysis tasks by presenting a unique RAG workflow that mitigates the scalability issues of existing tabular LLM solutions. Specifically, we present Tabular Embedding Model (TEM), a novel approach to fine-tune embedding models for tabular Retrieval-Augmentation Generation (RAG) applications. Embedding models form a crucial component in the RAG workflow and even current SOTA embedding models struggle as they are predominantly trained on textual datasets and thus underperform in scenarios involving complex tabular data. The evaluation results showcase that our approach not only outperforms current SOTA embedding models in this domain but also does so with a notably smaller and more efficient model structure., Comment: 11 pages, 5 figures

We show that the magnetic properties of high-entropy alloys (HEAs) can be captured by \textit{ab initio} calculations within the coherent potential approximation, where the atomic details of the high-entropy mixing are considered as an effective medium that possesses the translational symmetry of the lattice. This is demonstrated using the face-centered cubic (FCC) phase of $\textrm{FeCoNiMnCu}$ and the $L1_0$ phase of $\textrm{(FeCoNiMnCu)Pt}$ by comparing the density functional theory (DFT) results with the experimental values. Working within the first Brillouin zone and the primitive unit cell, we show that DFT can capture the smooth profile of magnetic properties such as the saturation magnetization, the Curie temperature and the magnetic anisotropy, using only a sparse set of sampling points in the vast compositional space. The smooth profiles given by DFT indeed follow the experimental trend, demonstrating the promising potential of using machine learning to explore the magnetic properties of HEAs, by establishing reasonably large datasets with high-throughput calculations using density-functional theory.

Gaia can not individually resolve very close binary systems, however, the collected data can still be used to identify them. A powerful indicator of stellar multiplicity is the sources reported Renormalized Unit Weight Error (ruwe), which effectively captures the astrometric deviations from single-source solutions. We aim to characterise the imprints left on ruwe caused by binarity. By flagging potential binary systems based on ruwe, we aim to characterise which of their properties will contribute the most to their detectability. We develop a model to estimate ruwe values for observations of Gaia sources, based on the biases to the single-source astrometric track arising from the presence of an unseen companion. Then, using the recipes from previous GaiaUnlimited selection functions, we estimate the selection probability of sources with high ruwe, and discuss what binary properties contribute to increasing the sources ruwe. We compute the maximum ruwe value which is compatible with single-source solutions as a function of their location on-sky. We see that binary systems selected as sources with a ruwe higher than this sky-varying threshold have a strong detectability window in their orbital period distribution, which peaks at periods equal to the Gaia observation time baseline. We demonstrate how our sky-varying ruwe threshold provides a more complete sample of binary systems when compared to single sky-averaged values by studying the unresolved binary population in the Gaia Catalogue of Nearby Stars. We provide the code and tools used in this study, as well as the sky-varying ruwe threshold through the GaiaUnlimited Python package, Comment: 11 pages, 10 figures. Accepted for publication in A&A

Gravitational waves from black-hole merging events have revealed a population of extra-galactic BHs residing in short-period binaries with masses that are higher than expected based on most stellar evolution models - and also higher than known stellar-origin black holes in our Galaxy. It has been proposed that those high-mass BHs are the remnants of massive metal-poor stars. Gaia astrometry is expected to uncover many Galactic wide-binary systems containing dormant BHs, which may not have been detected before. The study of this population will provide new information on the BH-mass distribution in binaries and shed light on their formation mechanisms and progenitors. As part of the validation efforts in preparation for the fourth Gaia data release (DR4), we analysed the preliminary astrometric binary solutions, obtained by the Gaia Non-Single Star pipeline, to verify their significance and to minimise false-detection rates in high-mass-function orbital solutions. The astrometric binary solution of one source, Gaia BH3, implies the presence of a 32.70 \pm 0.82 M\odot BH in a binary system with a period of 11.6 yr. Gaia radial velocities independently validate the astrometric orbit. Broad-band photometric and spectroscopic data show that the visible component is an old, very metal-poor giant of the Galactic halo, at a distance of 590 pc. The BH in the Gaia BH3 system is more massive than any other Galactic stellar-origin BH known thus far. The low metallicity of the star companion supports the scenario that metal-poor massive stars are progenitors of the high-mass BHs detected by gravitational-wave telescopes. The Galactic orbit of the system and its metallicity indicate that it might belong to the Sequoia halo substructure. Alternatively, and more plausibly, it could belong to the ED-2 stream, which likely originated from a globular cluster that had been disrupted by the Milky Way., Comment: 23 pages, accepted fro publication in A&A Letters. New version with small fixes

Hybrid work, a fusion of different work environments that allow employees to work in and outside their offices, represents a new frontier for agile researchers to explore. However, due to the nascent nature of the research phenomena, we are yet to achieve a good understanding of the research terrain formulated when hybrid work meets agile software development. This systematic mapping study, we aimed to provide a good understanding of this emerging research area. The systematic process we followed led to a collection of 12 primary studies, which is less than what we expected. All the papers are empirical studies, with most of them employing case studies as the research methodology. The people-centric nature of agile methods is yet to be adequately reflected in the studies in this area. Similarly, there is a lack of a richer understanding of hybrid work in terms of flexible work arrangements. Our mapping study identified various research opportunities that can be explored in future research., Comment: 11 pages

The Embodied AI community has made significant strides in visual navigation tasks, exploring targets from 3D coordinates, objects, language descriptions, and images. However, these navigation models often handle only a single input modality as the target. With the progress achieved so far, it is time to move towards universal navigation models capable of handling various goal types, enabling more effective user interaction with robots. To facilitate this goal, we propose GOAT-Bench, a benchmark for the universal navigation task referred to as GO to AnyThing (GOAT). In this task, the agent is directed to navigate to a sequence of targets specified by the category name, language description, or image in an open-vocabulary fashion. We benchmark monolithic RL and modular methods on the GOAT task, analyzing their performance across modalities, the role of explicit and implicit scene memories, their robustness to noise in goal specifications, and the impact of memory in lifelong scenarios.

CSP sparsification, introduced by Kogan and Krauthgamer (ITCS 2015), considers the following question: when can an instance of a constraint satisfaction problem be sparsified (by retaining a weighted subset of the constraints) while still roughly capturing the weight of constraints satisfied by {\em every} assignment. CSP sparsification generalizes and abstracts other commonly studied problems including graph cut-sparsification, hypergraph cut-sparsification and hypergraph XOR-sparsification. A central question here is to understand what properties of a constraint predicate $P:\Sigma^r \to \{0,1\}$ (where variables are assigned values in $\Sigma$) allow for nearly linear-size sparsifiers (in the number of variables). In this work (1) we significantly extend the class of CSPs for which nearly linear-size, and other non-trivial, sparsifications exist and give classifications in some broad settings and (2) give a polynomial-time algorithm to extract this sparsification. Our results captured in item (1) completely classify all symmetric Boolean predicates $P$ (i.e., on the Boolean domain $\Sigma = \{0,1\}$) that allow nearly-linear-size sparsifications. Symmetric Boolean CSPs already capture all the special classes of sparisifcation listed above including hypergraph cut-sparsification and variants. Our study of symmetric CSPs reveals an inherent, previously undetected, number-theoretic phenomenon that determines near-linear size sparsifiability. We also completely classify the set of Boolean predicates $P$ that allow non-trivial ($o(n^r)$-size) sparsifications, thus answering an open question from the work of Kogan and Krauthgamer.

Linear models are ubiquitous in data science, but are particularly prone to overfitting and data memorization in high dimensions. To guarantee the privacy of training data, differential privacy can be used. Many papers have proposed optimization techniques for high-dimensional differentially private linear models, but a systematic comparison between these methods does not exist. We close this gap by providing a comprehensive review of optimization methods for private high-dimensional linear models. Empirical tests on all methods demonstrate robust and coordinate-optimized algorithms perform best, which can inform future research. Code for implementing all methods is released online., Comment: 21 pages, 7 figures. To be published at the 2nd IEEE Conference on Secure and Trustworthy Machine Learning (SaTML)

This work introduces Extent of Violation Indices (EVIs), a novel metric for quantifying how well exchange-correlation functionals adhere to local conditions. Applying EVIs to a diverse set of molecules for GGA functionals reveals widespread violations, particularly for semi-empirical functionals. We leverage EVIs to explore potential connections between these violations and errors in chemical properties. While no correlation is observed for atomization energies, a link emerges between EVIs and total energies. Similarly, the analysis of reaction energies suggests weak positive correlations for specific conditions, but definitive conclusions about error cancellation require advancements in both functional accuracy and our understanding of cancellation mechanisms. Overall, this study highlights EVIs as a powerful tool for analyzing functional behavior and adherence to local conditions, paving the way for future research to fully elucidate the impact of violations on energy errors.

The exponential growth in scale and relevance of social networks enable them to provide expansive insights. Predicting missing links in social networks efficiently can help in various modern-day business applications ranging from generating recommendations to influence analysis. Several categories of solutions exist for the same. Here, we explore various feature extraction techniques to generate representations of nodes and edges in a social network that allow us to predict missing links. We compare the results of using ten feature extraction techniques categorized across Structural embeddings, Neighborhood-based embeddings, Graph Neural Networks, and Graph Heuristics, followed by modeling with ensemble classifiers and custom Neural Networks. Further, we propose combining heuristic-based features and learned representations that demonstrate improved performance for the link prediction task on social network datasets. Using this method to generate accurate recommendations for many applications is a matter of further study that appears very promising. The code for all the experiments has been made public., Comment: Accepted to the MAISoN Workshop at IJCAI 2023

We present a parallel algorithm for the $(1-\epsilon)$-approximate maximum flow problem in capacitated, undirected graphs with $n$ vertices and $m$ edges, achieving $O(\epsilon^{-3}\text{polylog} n)$ depth and $O(m \epsilon^{-3} \text{polylog} n)$ work in the PRAM model. Although near-linear time sequential algorithms for this problem have been known for almost a decade, no parallel algorithms that simultaneously achieved polylogarithmic depth and near-linear work were known. At the heart of our result is a polylogarithmic depth, near-linear work recursive algorithm for computing congestion approximators. Our algorithm involves a recursive step to obtain a low-quality congestion approximator followed by a "boosting" step to improve its quality which prevents a multiplicative blow-up in error. Similar to Peng [SODA'16], our boosting step builds upon the hierarchical decomposition scheme of R\"acke, Shah, and T\"aubig [SODA'14]. A direct implementation of this approach, however, leads only to an algorithm with $n^{o(1)}$ depth and $m^{1+o(1)}$ work. To get around this, we introduce a new hierarchical decomposition scheme, in which we only need to solve maximum flows on subgraphs obtained by contracting vertices, as opposed to vertex-induced subgraphs used in R\"acke, Shah, and T\"aubig [SODA'14]. In particular, we are able to directly extract congestion approximators for the subgraphs from a congestion approximator for the entire graph, thereby avoiding additional recursion on those subgraphs. Along the way, we also develop a parallel flow-decomposition algorithm that is crucial to achieving polylogarithmic depth and may be of independent interest.

Recently, a number of variants of the notion of cut-preserving hypergraph sparsification have been studied in the literature. These variants include directed hypergraph sparsification, submodular hypergraph sparsification, general notions of approximation including spectral approximations, and more general notions like sketching that can answer cut queries using more general data structures than just sparsifiers. In this work, we provide reductions between these different variants of hypergraph sparsification and establish new upper and lower bounds on the space complexity of preserving their cuts. At a high level, our results use the same general principle, namely, by showing that cuts in one class of hypergraphs can be simulated by cuts in a simpler class of hypergraphs, we can leverage sparsification results for the simpler class of hypergraphs.

In a recent paper [Carcassi, Oldofredi and Aidala, Found Phys 54, 14 (2024)] it is claimed that the whole Harrigan--Spekkens framework of ontological models is inconsistent with quantum theory. They show this by showing that all pure quantum states in $\psi$-ontic models must be orthogonal. In this note, we identify some crucial mistakes in their argument to the extent that the main claim is incorrect., Comment: 5 pages including references All comments welcome!

Diabetes, a chronic condition that impairs how the body turns food into energy, i.e. blood glucose, affects 38 million people in the US alone. The standard treatment is to supplement carbohydrate intake with an artificial pancreas, i.e. a continuous insulin pump (basal shots), as well as occasional insulin injections (bolus shots). The goal of the treatment is to keep blood glucose at the center of an acceptable range, as measured through a continuous glucose meter. A secondary goal is to minimize injections, which are unpleasant and difficult for some patients to implement. In this study, neuroevolution was used to discover an optimal strategy for the treatment. Based on a dataset of 30 days of treatment and measurements of a single patient, a random forest was first trained to predict future glucose levels. A neural network was then evolved to prescribe carbohydrates, basal pumping levels, and bolus injections. Evolution discovered a Pareto front that reduced deviation from the target and number of injections compared to the original data, thus improving patients' quality of life. To make the system easier to adopt, a language interface was developed with a large language model. Thus, these technologies not only improve patient care but also adoption in a broader population.

Craters are one of the most studied planetary features used for different scientific analyses, such as estimation of surface age and surface processes. Satellite images utilized for crater detection often have low resolution (LR) due to hardware constraints and transmission time. Super-resolution (SR) is a practical and cost-effective solution; however, most SR approaches work on fixed integer scale factors, i.e., a single model can generate images of a specific resolution. In practical applications, SR on multiple scales provides various levels of detail, but training for each scale is resource-intensive. Therefore, this paper proposes a system for crater detection assisted with an arbitrary scale super-resolution approach (i.e., a single model can be used for multiple scale factors) for the lunar surface. Our work is composed of two subsystems. The first sub-system employs an arbitrary scale SR approach to generate super-resolved images of multiple resolutions. Subsequently, the second sub-system passes super-resolved images of multiple resolutions to a deep learning-based crater detection framework for identifying craters on the lunar surface. Employed arbitrary scale SR approach is based on a combination of convolution and transformer modules. For the crater detection sub-system, we utilize the Mask-RCNN framework. Using SR images of multiple resolutions, the proposed system detects 13.47% more craters from the ground truth than the craters detected using only the LR images. Further, in complex crater settings, specifically in overlapping and degraded craters, 11.84% and 15.01% more craters are detected as compared to the crater detection networks using only the LR images. The proposed system also leads to better localization performance, 3.19% IoU increment compared to the LR images, Comment: 15 pages, 8 figures, 8 Tables

Large Language Models (LLMs) inherently carry the biases contained in their training corpora, which can lead to the perpetuation of societal harm. As the impact of these foundation models grows, understanding and evaluating their biases becomes crucial to achieving fairness and accuracy. We propose to study what LLMs know about the world we live in through the lens of geography. This approach is particularly powerful as there is ground truth for the numerous aspects of human life that are meaningfully projected onto geographic space such as culture, race, language, politics, and religion. We show various problematic geographic biases, which we define as systemic errors in geospatial predictions. Initially, we demonstrate that LLMs are capable of making accurate zero-shot geospatial predictions in the form of ratings that show strong monotonic correlation with ground truth (Spearman's $\rho$ of up to 0.89). We then show that LLMs exhibit common biases across a range of objective and subjective topics. In particular, LLMs are clearly biased against locations with lower socioeconomic conditions (e.g. most of Africa) on a variety of sensitive subjective topics such as attractiveness, morality, and intelligence (Spearman's $\rho$ of up to 0.70). Finally, we introduce a bias score to quantify this and find that there is significant variation in the magnitude of bias across existing LLMs.

Retrieval-augmented language models can better adapt to changes in world state and incorporate long-tail knowledge. However, most existing methods retrieve only short contiguous chunks from a retrieval corpus, limiting holistic understanding of the overall document context. We introduce the novel approach of recursively embedding, clustering, and summarizing chunks of text, constructing a tree with differing levels of summarization from the bottom up. At inference time, our RAPTOR model retrieves from this tree, integrating information across lengthy documents at different levels of abstraction. Controlled experiments show that retrieval with recursive summaries offers significant improvements over traditional retrieval-augmented LMs on several tasks. On question-answering tasks that involve complex, multi-step reasoning, we show state-of-the-art results; for example, by coupling RAPTOR retrieval with the use of GPT-4, we can improve the best performance on the QuALITY benchmark by 20% in absolute accuracy.

Stochastic Gradient Descent (SGD) stands as a cornerstone optimization algorithm with proven real-world empirical successes but relatively limited theoretical understanding. Recent research has illuminated a key factor contributing to its practical efficacy: the implicit regularization it instigates. Several studies have investigated the linear stability property of SGD in the vicinity of a stationary point as a predictive proxy for sharpness and generalization error in overparameterized neural networks (Wu et al., 2022; Jastrzebski et al., 2019; Cohen et al., 2021). In this paper, we delve deeper into the relationship between linear stability and sharpness. More specifically, we meticulously delineate the necessary and sufficient conditions for linear stability, contingent on hyperparameters of SGD and the sharpness at the optimum. Towards this end, we introduce a novel coherence measure of the loss Hessian that encapsulates pertinent geometric properties of the loss function that are relevant to the linear stability of SGD. It enables us to provide a simplified sufficient condition for identifying linear instability at an optimum. Notably, compared to previous works, our analysis relies on significantly milder assumptions and is applicable for a broader class of loss functions than known before, encompassing not only mean-squared error but also cross-entropy loss., Comment: To appear at ICLR 2024

The spatial distribution of Galactic stars with different chemical abundances encodes information on the processes that drove the formation and evolution of the Milky Way. Survey selection functions are indispensable for analysing astronomical catalogues produced by large-scale surveys. The use of these selection functions in data modelling is more complex when data from different surveys are to be modelled simultaneously. We introduce a procedure for constructing the selection function of a sample of red clump stars that have parallaxes and elemental abundances from the Gaia mission. We separately constructed the selection function of the APOGEE DR17 red clump stars, which depends on very different observables and has a very different spatial coverage. We combined the two surveys and accounted for their joint selection function to provide strong constraints on the radial and vertical density distribution of mono-abundance populations, with Gaia offering a dense coverage of the solar neighbourhood, while APOGEE reaches larger distances near the Galactic plane. We confirm that the radial density profile steepens with increasing metallicity. The combined sample also indicates a metallicity-dependent flaring of the alpha-poor disc. We provide the code for constructing the Gaia selection function we used in this study through the GaiaUnlimited Python package., Comment: 21 pages, 18 figures, accepted for publication in Astronomy & Astrophysics

The major source of ground-based gravitational wave detectors, the inspiral and merger of comparable mass binary black holes (BBH), consists of a slow quasicircular inspiral, a merger to form a single remnant hole, and the quasinormal ringing of that remnant. The first and last of these epochs are amenable to well developed and familiar approximations: Newtonian or post-Newtonian for the first, and BH perturbation methods for the last. Ironically, the middle epoch, the merger, generates most of the GW emission, yet has been accessible so far only to numerical relativity. Here we add the close-limit approximation for the phenomenology of the merger. With the completed set of methods (Newtonian/post-Newtonian; close limit; BH perturbation theory) we show that not only can we understand the results of BBH coalescence, but - with reasonable accuracy - we can {\it predict} the resulting radiation and remnant., Comment: 10 pages, 12 figures, 2 tables