Search

We present very-long-baseline interferometry (VLBI) observations of a continuum radio source potentially associated with the fast radio burst source FRB 20190520B. Using the European VLBI network (EVN), we find the source to be compact on VLBI scales with an angular size of $<2.3$ mas ($3\sigma$). This corresponds to a transverse physical size of $<9$ pc (at the $z=0.241$ redshift of the host galaxy), confirming it to be an FRB persistent radio source (PRS) like that associated with the first-known repeater FRB 20121102A. The PRS has a flux density of $201 \pm 34 \rm{\mu Jy}$ at 1.7 GHz and a spectral radio luminosity of $L_{1.7 \rm GHz} = (3.0 \pm 0.5) \times 10^{29}\,\mathrm{erg s^{-1} Hz^{-1}}$ (also similar to the FRB 20121102A PRS). Comparing to previous lower-resolution observations, we find that no flux is resolved out on milliarcsecond scales. We have refined the PRS position, improving its precision by an order of magnitude compared to previous results. We also report the detection of a FRB 20190520B burst at 1.4 GHz and find the burst position to be consistent with the PRS position, at $\lesssim20$ mas. This strongly supports their direct physical association and the hypothesis that a single central engine powers both the bursts and the PRS. We discuss the model of a magnetar in a wind nebula and present an allowed parameter space for its age and the radius of the putative nebula powering the observed PRS emission. Alternatively, we find that an accretion-powered 'hypernebula' model also fits our observational constraints., Comment: 13 pages, 6 figures, Submitted to ApJL

We show that in the $\mathbb{P}_{\max}$ extension of a certain Chang-type model of determinacy, if $\kappa\in\{\omega_1, \omega_2, \omega_3\}$, then the restriction of the club filter on $\kappa\cap\mathrm{Cof}(\omega)$ to HOD is an ultrafilter in HOD. This answers Question 4.11 of [BNH23] raised by Ben-Neria and Hayut., Comment: 22 pages. arXiv admin note: text overlap with arXiv:2307.08607

The InterPlanetary File System (IPFS) is currently the largest decentralized storage solution in operation, with thousands of active participants and millions of daily content transfers. IPFS is used as remote data storage for numerous blockchain-based smart contracts, Non-Fungible Tokens (NFT), and decentralized applications. We present a content censorship attack that can be executed with minimal effort and cost, and that prevents the retrieval of any chosen content in the IPFS network. The attack exploits a conceptual issue in a core component of IPFS, the Kademlia Distributed Hash Table (DHT), which is used to resolve content IDs to peer addresses. We provide efficient detection and mitigation mechanisms for this vulnerability. Our mechanisms achieve a 99.6\% detection rate and mitigate 100\% of the detected attacks with minimal signaling and computational overhead. We followed responsible disclosure procedures, and our countermeasures are scheduled for deployment in the future versions of IPFS., Comment: 17 pages (including references and appendices), 15 figures. Accepted to be published at the Network and Distributed System Security (NDSS) Symposium 2024

Recent developments in natural language processing have demonstrated the potential of large language models (LLMs) to improve a range of educational and learning outcomes. Of recent chatbots based on LLMs, ChatGPT and Bard have made it clear that artificial intelligence (AI) technology will have significant implications on the way we obtain and search for information. However, these tools sometimes produce text that is convincing, but often incorrect, known as hallucinations. As such, their use can distort scientific facts and spread misinformation. To counter polarizing responses on these tools, it is critical to provide an overview of such responses so stakeholders can determine which topics tend to produce more contentious responses -- key to developing targeted regulatory policy and interventions. In addition, there currently exists no annotated dataset of ChatGPT and Bard responses around possibly polarizing topics, central to the above aims. We address the indicated issues through the following contribution: Focusing on highly polarizing topics in the US, we created and described a dataset of ChatGPT and Bard responses. Broadly, our results indicated a left-leaning bias for both ChatGPT and Bard, with Bard more likely to provide responses around polarizing topics. Bard seemed to have fewer guardrails around controversial topics, and appeared more willing to provide comprehensive, and somewhat human-like responses. Bard may thus be more likely abused by malicious actors. Stakeholders may utilize our findings to mitigate misinformative and/or polarizing responses from LLMs

We present high-resolution 1.5 $-$ 6 GHz Karl G. Jansky Very Large Array (VLA) and Hubble Space Telescope (HST) optical and infrared observations of the extremely active repeating fast radio burst (FRB) FRB 20201124A and its barred spiral host galaxy. We constrain the location and morphology of star formation in the host and search for a persistent radio source (PRS) coincident with FRB 20201124A. We resolve the morphology of the radio emission across all frequency bands and measure a star formation rate SFR $\approx 8.9\,M_{\odot}$ yr$^{-1}$, approximately $\approx 2.5-6$ times larger than optically-inferred SFRs, demonstrating dust-obscured star formation throughout the host. Compared to a sample of all known FRB hosts with radio emission, the host of FRB 20201124A has the most significantly obscured star formation. While HST observations show the FRB to be offset from the bar or spiral arms, the radio emission extends to the FRB location. We propose that the FRB progenitor could have formed in situ (e.g., a magnetar born from a massive star explosion). It is still plausible, although less likely, that the progenitor of FRB 20201124A migrated from the central bar of the host. We further place a limit on the luminosity of a putative PRS at the FRB position of $L_{\rm 6.0 \ GHz}$ $\lesssim$ 1.8 $\times 10^{27}$ erg s$^{-1}$ Hz$^{-1}$, among the deepest PRS luminosity limits to date. However, this limit is still broadly consistent with both magnetar nebulae and hypernebulae models assuming a constant energy injection rate of the magnetar and an age of $\gtrsim 10^{5}$ yr in each model, respectively., Comment: 21 pages, 6 figures, 3 tables, Accepted to ApJ; doi:10.3847/1538-4357/ad0cbd

We report significantly improved accuracy of grain boundary segmentation using Convolutional Neural Networks (CNN) trained on a combination of real and generated data. Manual segmentation is accurate but time-consuming, and existing computational methods are faster but often inaccurate. To combat this dilemma, machine learning models can be used to achieve the accuracy of manual segmentation and have the efficiency of a computational method. An extensive dataset of from 316L stainless steel samples is additively manufactured, prepared, polished, etched, and then microstructure grain images were systematically collected. Grain segmentation via existing computational methods and manual (by-hand) were conducted, to create "real" training data. A Voronoi tessellation pattern combined with random synthetic noise and simulated defects, is developed to create a novel artificial grain image fabrication method. This provided training data supplementation for data-intensive machine learning methods. The accuracy of the grain measurements from microstructure images segmented via computational methods and machine learning methods proposed in this work are calculated and compared to provide much benchmarks in grain segmentation. Over 400 images of the microstructure of stainless steel samples were manually segmented for machine learning training applications. This data and the artificial data is available on Kaggle.

The proliferation of discussion about fatherhood in Singapore attests to its significance, indicating the need for an exploration of how fatherhood is framed, aiding policy-making around fatherhood in Singapore. Sound and holistic policy around fatherhood in Singapore may reduce stigma and apprehension around being a parent, critical to improving the nations flagging birth rate. We analyzed 15,705 articles and 56,221 posts to study how fatherhood is framed in Singapore across a range of online platforms (news outlets, parenting forums, Twitter). We used NLP techniques to understand these differences. While fatherhood was framed in a range of ways on the Singaporean online environment, it did not seem that fathers were framed as central to the Singaporean family unit. A strength of our work is how the different techniques we have applied validate each other.

We infer the connection between the stellar mass of galaxies from the Subaru Hyper Suprime-Cam (HSC) survey, and their dark matter halo masses and its evolution in two bins of redshifts between $[0.3, 0.8]$. We use the measurements of the weak lensing signal of galaxies using background sources from the Year 1 shape catalog from the HSC survey. We bin galaxies in stellar mass with varying thresholds ranging from $8.6 \leq \log [ M_*/(h^{-2} {M_\odot})] \leq 11.2$ and use stringent cuts in the selection of source galaxies to measure the weak lensing signal. We model these measurements of the weak lensing signal together with the abundance of galaxies in the halo occupation distribution framework. We obtain constraints on the halo occupation parameters of central galaxies $M_{\rm min}$ and $\sigma_{\log M}$, which correspond to the halo mass at which central galaxies for each threshold sample reach half occupancy, and its scatter, respectively, along with parameters that describe the occupation of the satellite galaxies. The measurements of abundance and weak lensing individually constrain different degeneracy directions in the $M_{\rm min}$ and $\sigma_{\log M}$ plane, thus breaking the degeneracy in these parameters. We demonstrate that the weak lensing measurements are best able to constrain the average central halo masses, $\langle M_{\rm cen} \rangle$. We compare our measurements to those obtained using the abundance and clustering of these galaxies as well as the subhalo abundance matching measurements and demonstrate qualitative agreement. We find that the galaxy-dark matter connection does not vary significantly between redshift bins we explore in this study. Uncertainties in the photometric redshift of the lens galaxies imply that more efforts are required to understand the true underlying stellar mass-halo mass relation of galaxies and its evolution over cosmic epoch.

Machine learning algorithms can sometimes exacerbate health disparities based on ethnicity, gender, and other factors. There has been limited work at exploring potential biases within algorithms deployed on a small scale, and/or within minoritized communities. Understanding the nature of potential biases may improve the prediction of various health outcomes. As a case study, we used data from a sample of 539 young adults from minoritized communities who engaged in nonmedical use of prescription opioids and/or heroin. We addressed the indicated issues through the following contributions: 1) Using machine learning techniques, we predicted a range of opioid use outcomes for participants in our dataset; 2) We assessed if algorithms trained only on a majority sub-sample (e.g., Non-Hispanic/Latino, male), could accurately predict opioid use outcomes for a minoritized sub-sample (e.g., Latino, female). Results indicated that models trained on a random sample of our data could predict a range of opioid use outcomes with high precision. However, we noted a decrease in precision when we trained our models on data from a majority sub-sample, and tested these models on a minoritized sub-sample. We posit that a range of cultural factors and systemic forms of discrimination are not captured by data from majority sub-samples. Broadly, for predictions to be valid, models should be trained on data that includes adequate representation of the groups of people about whom predictions will be made. Stakeholders may utilize our findings to mitigate biases in models for predicting opioid use outcomes within minoritized communities.

Cell line authentication plays a crucial role in the biomedical field, ensuring researchers work with accurately identified cells. Supervised deep learning has made remarkable strides in cell line identification by studying cell morphological features through cell imaging. However, batch effects, a significant issue stemming from the different times at which data is generated, lead to substantial shifts in the underlying data distribution, thus complicating reliable differentiation between cell lines from distinct batch cultures. To address this challenge, we introduce CLANet, a pioneering framework for cross-batch cell line identification using brightfield images, specifically designed to tackle three distinct batch effects. We propose a cell cluster-level selection method to efficiently capture cell density variations, and a self-supervised learning strategy to manage image quality variations, thus producing reliable patch representations. Additionally, we adopt multiple instance learning(MIL) for effective aggregation of instance-level features for cell line identification. Our innovative time-series segment sampling module further enhances MIL's feature-learning capabilities, mitigating biases from varying incubation times across batches. We validate CLANet using data from 32 cell lines across 93 experimental batches from the AstraZeneca Global Cell Bank. Our results show that CLANet outperforms related approaches (e.g. domain adaptation, MIL), demonstrating its effectiveness in addressing batch effects in cell line identification., Comment: 15 pages, 10 figures

We calculate chirally enhanced corrections to the muon's electric and magnetic dipole moments in two-Higgs-doublet models extended by vector-like leptons, and we explore a sharp correlation between $h \rightarrow \mu^+ \mu^-$ and the muon's dipole moments in these models. Among many detailed predictions, for a model with new leptons with the same quantum numbers as standard model leptons, we find that $0.38 \lesssim \tan \beta \lesssim 21$ necessarily requires a muon electric dipole moment to be observed at near-future experiments, assuming $h \rightarrow \mu^+ \mu^-$ is measured within $1\%$ of the standard model prediction for the current central value of the measured muon magnetic moment. In all studied models, the predicted values of the electric dipole moment can reach up to current experimental limits. Moreover, we show that in some models there can be two sources of chiral enhancement, parametrizing the correlation between $h \rightarrow \mu^+ \mu^-$ and the dipole moments by a complex number. This leads to sign-preferred predictions for the electric dipole moment., Comment: 26 pages + 25 pages of appendices, 9 figures. v2: Matches published version

Language models of code (LMs) work well when the surrounding code provides sufficient context. This is not true when it becomes necessary to use types, functionality or APIs defined elsewhere in the repository or a linked library, especially those not seen during training. LMs suffer from limited awareness of such global context and end up hallucinating. Integrated development environments (IDEs) assist developers in understanding repository context using static analysis. We extend this assistance, enjoyed by developers, to LMs. We propose monitor-guided decoding (MGD) where a monitor uses static analysis to guide the decoding. We construct a repository-level dataset PragmaticCode for method-completion in Java and evaluate MGD on it. On models of varying parameter scale, by monitoring for type-consistent object dereferences, MGD consistently improves compilation rates and agreement with ground truth. Further, LMs with fewer parameters, when augmented with MGD, can outperform larger LMs. With MGD, SantaCoder-1.1B achieves better compilation rate and next-identifier match than the much larger text-davinci-003 model. We also conduct a generalizability study to evaluate the ability of MGD to generalize to multiple programming languages (Java, C# and Rust), coding scenarios (e.g., correct number of arguments to method calls), and to enforce richer semantic constraints (e.g., stateful API protocols). Our data and implementation are available at https://github.com/microsoft/monitors4codegen ., Comment: Accepted to NeurIPS 2023 and to appear as "Monitor-Guided Decoding of Code LMs with Static Analysis of Repository Context" at https://neurips.cc/virtual/2023/poster/70362 . Contents: 11 pages, 15 additional pages of appendix, 13 figures, 3 tables

In-context learning (ICL) is one of the surprising and useful features of large language models and subject of intense research. Recently, stylized meta-learning-like ICL setups have been devised that train transformers on sequences of input-output pairs $(x, f(x))$. The function $f$ comes from a function class and generalization is checked by evaluating on sequences generated from unseen functions from the same class. One of the main discoveries in this line of research has been that for several function classes, such as linear regression, transformers successfully generalize to new functions in the class. However, the inductive biases of these models resulting in this behavior are not clearly understood. A model with unlimited training data and compute is a Bayesian predictor: it learns the pretraining distribution. In this paper we empirically examine how far this Bayesian perspective can help us understand ICL. To this end, we generalize the previous meta-ICL setup to hierarchical meta-ICL setup which involve unions of multiple task families. We instantiate this setup on a diverse range of linear and nonlinear function families and find that transformers can do ICL in this setting as well. Where Bayesian inference is tractable, we find evidence that high-capacity transformers mimic the Bayesian predictor. The Bayesian perspective provides insights into the inductive bias of ICL and how transformers perform a particular task when they are trained on multiple tasks. We also find that transformers can learn to generalize to new function classes that were not seen during pretraining. This involves deviation from the Bayesian predictor. We examine these deviations in more depth offering new insights and hypotheses., Comment: ICLR 2024

The permanent of a non-negative square matrix can be well approximated by finding the minimum of the Bethe free energy functions associated with some suitably defined factor graph; the resulting approximation to the permanent is called the Bethe permanent. Vontobel gave a combinatorial characterization of the Bethe permanent via degree-$M$ Bethe permanents, which are based on degree-$M$ covers of the underlying factor graph. In this paper, we prove a degree-$M$-Bethe-permanent-based lower bound on the permanent of a non-negative matrix, which solves a conjecture proposed by Vontobel in [IEEE Trans. Inf. Theory, Mar. 2013]. We also prove a degree-$M$-Bethe-permanent-based upper bound on the permanent of a non-negative matrix. In the limit $M \to \infty$, these lower and upper bounds yield known Bethe-permanent-based lower and upper bounds on the permanent of a non-negative matrix. Moreover, we prove similar results for an approximation to the permanent known as the (scaled) Sinkhorn permanent., Comment: to appear in the IEEE Transactions on Information Theory

Speech emotion conversion aims to convert the expressed emotion of a spoken utterance to a target emotion while preserving the lexical information and the speaker's identity. In this work, we specifically focus on in-the-wild emotion conversion where parallel data does not exist, and the problem of disentangling lexical, speaker, and emotion information arises. In this paper, we introduce a methodology that uses self-supervised networks to disentangle the lexical, speaker, and emotional content of the utterance, and subsequently uses a HiFiGAN vocoder to resynthesise the disentangled representations to a speech signal of the targeted emotion. For better representation and to achieve emotion intensity control, we specifically focus on the aro\-usal dimension of continuous representations, as opposed to performing emotion conversion on categorical representations. We test our methodology on the large in-the-wild MSP-Podcast dataset. Results reveal that the proposed approach is aptly conditioned on the emotional content of input speech and is capable of synthesising natural-sounding speech for a target emotion. Results further reveal that the methodology better synthesises speech for mid-scale arousal (2 to 6) than for extreme arousal (1 and 7)., Comment: Submitted to 15th ITG Conference on Speech Communication

We propose a novel algorithm called Unique Brain Network Identification Number, UBNIN for encoding the brain networks of individual subjects. To realize this objective, we employed structural MRI on 180 Parkinsons disease PD patients and 70 healthy controls HC from the National Institute of Mental Health and Neurosciences, India. We parcellated each subjects brain volume and constructed an individual adjacency matrix using the correlation between the gray matter volumes of every pair of regions. The unique code is derived from values representing connections for every node i, weighted by a factor of 2^1-i. The numerical representation UBNIN was observed to be distinct for each individual brain network, which may also be applied to other neuroimaging modalities. This model may be implemented as a neural signature of a persons unique brain connectivity, thereby making it useful for brainprinting applications. Additionally, we segregated the above datasets into five age cohorts to study the variation in network topology over age. Sparsity was adopted as the threshold estimate to binarize each age-based correlation matrix. For each age cohort, a decreasing trend was observed in the mean clustering coefficient with increasing sparsity. Significantly different clustering coefficients were noted in PD between age cohort B and C, C and E, and in HC between E and B, E and C, E and D, and C and D. Our findings suggest network connectivity patterns change with age, indicating network disruption may be due to the underlying neuropathology. Varying clustering coefficients for different cohorts indicate that information transfer between neighboring nodes changes with age. This provides evidence of age related brain shrinkage and network degeneration. We also discuss limitations and provide an open-access link to software codes and a help file for the entire study., Comment: 15 pages, 5 figures,1 algorithm, 1 main table, 1 appendix table