Search

This paper presents a novel framework for full-waveform seismic source inversion using simulation-based inference (SBI). Traditional probabilistic approaches often rely on simplifying assumptions about data errors, which we show can lead to inaccurate uncertainty quantification. SBI addresses this limitation by building an empirical probabilistic model of the data errors using machine learning models, known as neural density estimators, which can then be integrated into the Bayesian inference framework. We apply the SBI framework to point-source moment tensor inversions as well as joint moment tensor and time-location inversions. We construct a range of synthetic examples to explore the quality of the SBI solutions, as well as to compare the SBI results with standard Gaussian likelihood-based Bayesian inversions. We then demonstrate that under real seismic noise, common Gaussian likelihood assumptions for treating full-waveform data yield overconfident posterior distributions that underestimate the moment tensor component uncertainties by up to a factor of 3. We contrast this with SBI, which produces well-calibrated posteriors that generally agree with the true seismic source parameters, and offers an order-of-magnitude reduction in the number of simulations required to perform inference compared to standard Monte Carlo techniques. Finally, we apply our methodology to a pair of moderate magnitude earthquakes in the North Atlantic. We utilise seismic waveforms recorded by the recent UPFLOW ocean bottom seismometer array as well as by regional land stations in the Azores, comparing full moment tensor and source-time location posteriors between SBI and a Gaussian likelihood approach. We find that our adaptation of SBI can be directly applied to real earthquake sources to efficiently produce high quality posterior distributions that significantly improve upon Gaussian likelihood approaches., Comment: 22 + 11 pages, 11 + 11 figures

In learning problems, the noise inherent to the task at hand hinders the possibility to infer without a certain degree of uncertainty. Quantifying this uncertainty, regardless of its wide use, assumes high relevance for security-sensitive applications. Within these scenarios, it becomes fundamental to guarantee good (i.e., trustworthy) uncertainty measures, which downstream modules can securely employ to drive the final decision-making process. However, an attacker may be interested in forcing the system to produce either (i) highly uncertain outputs jeopardizing the system's availability or (ii) low uncertainty estimates, making the system accept uncertain samples that would instead require a careful inspection (e.g., human intervention). Therefore, it becomes fundamental to understand how to obtain robust uncertainty estimates against these kinds of attacks. In this work, we reveal both empirically and theoretically that defending against adversarial examples, i.e., carefully perturbed samples that cause misclassification, additionally guarantees a more secure, trustworthy uncertainty estimate under common attack scenarios without the need for an ad-hoc defense strategy. To support our claims, we evaluate multiple adversarial-robust models from the publicly available benchmark RobustBench on the CIFAR-10 and ImageNet datasets.

We employ a novel framework for accelerated cosmological inference, based on neural emulators and gradient-based sampling methods, to forecast constraints on dark energy models from Stage IV cosmic shear surveys. We focus on dark scattering (DS), an interacting dark energy model with pure momentum exchange in the dark sector, and train COSMOPOWER emulators to accurately and efficiently model the DS non-linear matter power spectrum produced by the halo model reaction framework, including the effects of baryon feedback and massive neutrinos. We embed the emulators within a fully-differentiable pipeline for gradient-based cosmological inference for which the batch likelihood call is up to $O(10^5)$ times faster than with traditional approaches, producing parameter constraints from simulated Stage IV cosmic shear data running on a single graphics processing unit (GPU). We also perform model comparison on the output chains from the inference process, employing the learnt harmonic mean estimator implemented in the software HARMONIC. We investigate degeneracies between dark energy and systematics parameters and assess the impact of scale cuts on the final constraints. Assuming a DS model for the mock data vector, we find that a Stage IV survey cosmic shear analysis can constrain the DS amplitude parameter $A_{\mathrm{ds}}$ with an uncertainty roughly an order of magnitude smaller than current constraints from Stage III surveys, even after marginalising over baryonic feedback, intrinsic alignments and redshift distribution uncertainties. These results show great promise for constraining DS with Stage IV data; furthermore, our methodology can be straightforwardly extended to a wide range of dark energy and modified gravity models., Comment: 9 pages, 6 figures. COSMOPOWER available at https://github.com/alessiospuriomancini/cosmopower, COSMOPOWER-JAX available at https://github.com/dpiras/cosmopower-jax, emulators for Dark Scattering available at https://github.com/karimpsi22/DS-emulators

Simulations of the dark matter distribution throughout the Universe are essential in order to analyse data from cosmological surveys. $N$-body simulations are computationally expensive, and many cheaper alternatives (such as lognormal random fields) fail to reproduce accurate statistics of the smaller, non-linear scales. In this work, we present Psi-GAN (Power-spectrum-informed generative adversarial network), a machine learning model which takes a two-dimensional lognormal dark matter density field and transforms it into a more realistic field. We construct Psi-GAN so that it is continuously conditional, and can therefore generate realistic realisations of the dark matter density field across a range of cosmologies and redshifts in $z \in [0, 3]$. We train Psi-GAN as a generative adversarial network on $2\,000$ simulation boxes from the Quijote simulation suite. We use a novel critic architecture that utilises the power spectrum as the basis for discrimination between real and generated samples. Psi-GAN shows agreement with $N$-body simulations over a range of redshifts and cosmologies, consistently outperforming the lognormal approximation on all tests of non-linear structure, such as being able to reproduce both the power spectrum up to wavenumbers of $1~h~\mathrm{Mpc}^{-1}$, and the bispectra of target $N$-body simulations to within ${\sim}5$ per cent. Our improved ability to model non-linear structure should allow more robust constraints on cosmological parameters when used in techniques such as simulation-based inference., Comment: 20 pages, 11 figures, 3 tables, 1 appendix. Submitted for publication by Monthly Notices of the Royal Astronomical Society

Vision foundation models, which have demonstrated significant potential in multimedia applications, are often underutilized in the natural sciences. This is primarily due to mismatches between the nature of domain-specific scientific data and the typical training data used for foundation models, leading to distribution shift. Scientific images can have unique structure and characteristics, and researchers often face the challenge of optimizing model performance with only a few hundred or thousand labeled examples. Furthermore, the choice of vision foundation model can be non-trivial, as each model exhibits unique strengths and limitations, influenced by differences in architecture, training procedures, and the datasets used for training. In this work, we evaluate the application of various vision foundation models to images from optical and radio astronomy. Our results show that use of certain foundation models improves the classification accuracy of optical galaxy images compared to conventional supervised training. Similarly, we achieve equivalent or better performance in object detection tasks with radio images. However, their performance in classifying radio galaxy images is generally poor and often inferior to traditional supervised training results. These findings suggest that selecting suitable vision foundation models for astrophysics applications requires careful consideration of the model characteristics and alignment with the specific requirements of the downstream tasks., Comment: 10 pages, 4 figures, submitted to Foundation Models for Science Workshop at NeurIPS 2024

Recent work has proposed neural network pruning techniques to reduce the size of a network while preserving robustness against adversarial examples, i.e., well-crafted inputs inducing a misclassification. These methods, which we refer to as adversarial pruning methods, involve complex and articulated designs, making it difficult to analyze the differences and establish a fair and accurate comparison. In this work, we overcome these issues by surveying current adversarial pruning methods and proposing a novel taxonomy to categorize them based on two main dimensions: the pipeline, defining when to prune; and the specifics, defining how to prune. We then highlight the limitations of current empirical analyses and propose a novel, fair evaluation benchmark to address them. We finally conduct an empirical re-evaluation of current adversarial pruning methods and discuss the results, highlighting the shared traits of top-performing adversarial pruning methods, as well as common issues. We welcome contributions in our publicly-available benchmark at https://github.com/pralab/AdversarialPruningBenchmark

Self-supervised learning (SSL) applied to natural images has demonstrated a remarkable ability to learn meaningful, low-dimension representations without labels, resulting in models that are adaptable to many different tasks. Until now, applications of SSL to astronomical images have been limited to Galaxy Zoo datasets, which require a significant amount of pre-processing to prepare sparse images centered on a single galaxy. With wide-field survey instruments at the forefront of the Square Kilometer Array (SKA) era, this approach to gathering training data is impractical. We demonstrate that continuum images from surveys like the MeerKAT Galactic Cluster Legacy Survey (MGCLS) can be successfully used with SSL, without extracting single-galaxy cutouts. Using the SSL framework DINO, we experiment with various preprocessing steps, augmentations, and architectures to determine the optimal approach for this data. We train both ResNet50 and Vision Transformer (ViT) backbones. Our models match state-of-the-art results (trained on Radio Galaxy Zoo) for FRI/FRII morphology classification. Furthermore, they predict the number of compact sources via linear regression with much higher accuracy. However, fine-tuning results in similar performance between our models, the state-of-the-art, and open-source models on multi-class morphology classification. Using source-rich crops from wide-field images to train multi-purpose models is an easily scalable approach that significantly reduces data preparation time. For the tasks evaluated in this work, twenty thousand crops is sufficient training data for models that produce results similar to state-of-the-art. In the future, complex tasks like source detection and characterization, together with domain-specific tasks, ought to demonstrate the true advantages of training models with radio astronomy data over natural-image foundation models., Comment: 16 pages, 11 figures. A version of this paper has been accepted for publication in Astronomy and Astrophysics. All datasets used in this work are public, and the code and pre-trained checkpoints are available at https://github.com/elastufka/mgcls_dino

Gradient-based attacks are a primary tool to evaluate robustness of machine-learning models. However, many attacks tend to provide overly-optimistic evaluations as they use fixed loss functions, optimizers, step-size schedulers, and default hyperparameters. In this work, we tackle these limitations by proposing a parametric variation of the well-known fast minimum-norm attack algorithm, whose loss, optimizer, step-size scheduler, and hyperparameters can be dynamically adjusted. We re-evaluate 12 robust models, showing that our attack finds smaller adversarial perturbations without requiring any additional tuning. This also enables reporting adversarial robustness as a function of the perturbation budget, providing a more complete evaluation than that offered by fixed-budget attacks, while remaining efficient. We release our open-source code at https://github.com/pralab/HO-FMN.

ModSecurity is widely recognized as the standard open-source Web Application Firewall (WAF), maintained by the OWASP Foundation. It detects malicious requests by matching them against the Core Rule Set (CRS), identifying well-known attack patterns. Each rule is manually assigned a weight based on the severity of the corresponding attack, and a request is blocked if the sum of the weights of matched rules exceeds a given threshold. However, we argue that this strategy is largely ineffective against web attacks, as detection is only based on heuristics and not customized on the application to protect. In this work, we overcome this issue by proposing a machine-learning model that uses the CRS rules as input features. Through training, ModSec-Learn is able to tune the contribution of each CRS rule to predictions, thus adapting the severity level to the web applications to protect. Our experiments show that ModSec-Learn achieves a significantly better trade-off between detection and false positive rates. Finally, we analyze how sparse regularization can reduce the number of rules that are relevant at inference time, by discarding more than 30% of the CRS rules. We release our open-source code and the dataset at https://github.com/pralab/modsec-learn and https://github.com/pralab/http-traffic-dataset, respectively., Comment: arXiv admin note: text overlap with arXiv:2308.04964

The aim of this study is to probe the sub-mJy polarized source population with LOFAR. We present the method used to stack LOFAR polarization datasets, the resulting catalog of polarized sources, and the derived polarized source counts. The ELAIS-N1 field was selected for a polarimetric study at 114.9-177.4 MHz. A total area of 25 deg2 was imaged at 6"- resolution in the Stokes Q and U parameters. Alignment of polarization angles was done both in frequency and in Faraday space before stacking datasets from 19 eight-hour-long epochs. A search for polarized sources was carried out in the final, stacked dataset, and the properties of the detected sources were examined. The depolarization level of sources known to be polarized at 1.4 GHz was quantified. A one-sigma noise level of 19 {\mu}Jy/beam was reached in the central part of the field after stacking. Twenty-five polarized sources were detected above 8\sigma, five of which had not been detected in polarization at any other radio frequencies before. Seven additional polarized components were found by lowering the threshold to 6\sigma at positions corresponding to sources known to be polarized at 1.4 GHz. In two radio galaxies, polarization was detected from both radio lobes, so the final number of associated radio continuum sources is 31. The detected sources are weakly polarized, with a median degree of polarization of 1.75% for the sample of sources detected in polarized emission. The sources previously detected in polarization at 1.4 GHz are significantly depolarized at 150 MHz. The catalog is used to derive the polarized source counts at 150 MHz. This is the deepest and highest-resolution polarization study at 150 MHz to date., Comment: 18 pages, 10 figures, accepted for publication in A&A

The abundance of dark matter haloes is a key cosmological probe in forthcoming galaxy surveys. The theoretical understanding of the halo mass function (HMF) is limited by our incomplete knowledge of the origin of non-universality and its cosmological parameter dependence. We present a deep learning model which compresses the linear matter power spectrum into three independent factors which are necessary and sufficient to describe the $z=0$ HMF from the state-of-the-art AEMULUS emulator to sub-per cent accuracy in a $w$CDM$+N_\mathrm{eff}$ parameter space. Additional information about growth history does not improve the accuracy of HMF predictions if the matter power spectrum is already provided as input, because required aspects of the former can be inferred from the latter. The three factors carry information about the universal and non-universal aspects of the HMF, which we interrogate via the information-theoretic measure of mutual information. We find that non-universality is captured by recent growth history after matter-dark-energy equality and $N_\mathrm{eff}$ for $M\sim 10^{13} \, \mathrm{M_\odot}\, h^{-1}$ haloes, and by $\Omega_{\rm m}$ for $M\sim 10^{15} \, \mathrm{M_\odot}\, h^{-1}$. The compact representation learnt by our model can inform the design of emulator training sets to achieve high emulator accuracy with fewer simulations., Comment: 17 pages, 13 figures. Minor changes to match version accepted for publication in MNRAS

We advocate for a new paradigm of cosmological likelihood-based inference, leveraging recent developments in machine learning and its underlying technology, to accelerate Bayesian inference in high-dimensional settings. Specifically, we combine (i) emulation, where a machine learning model is trained to mimic cosmological observables, e.g. CosmoPower-JAX; (ii) differentiable and probabilistic programming, e.g. JAX and NumPyro, respectively; (iii) scalable Markov chain Monte Carlo (MCMC) sampling techniques that exploit gradients, e.g. Hamiltonian Monte Carlo; and (iv) decoupled and scalable Bayesian model selection techniques that compute the Bayesian evidence purely from posterior samples, e.g. the learned harmonic mean implemented in harmonic. This paradigm allows us to carry out a complete Bayesian analysis, including both parameter estimation and model selection, in a fraction of the time of traditional approaches. First, we demonstrate the application of this paradigm on a simulated cosmic shear analysis for a Stage IV survey in 37- and 39-dimensional parameter spaces, comparing $\Lambda$CDM and a dynamical dark energy model ($w_0w_a$CDM). We recover posterior contours and evidence estimates that are in excellent agreement with those computed by the traditional nested sampling approach while reducing the computational cost from 8 months on 48 CPU cores to 2 days on 12 GPUs. Second, we consider a joint analysis between three simulated next-generation surveys, each performing a 3x2pt analysis, resulting in 157- and 159-dimensional parameter spaces. Standard nested sampling techniques are simply unlikely to be feasible in this high-dimensional setting, requiring a projected 12 years of compute time on 48 CPU cores; on the other hand, the proposed approach only requires 8 days of compute time on 24 GPUs. All packages used in our analyses are publicly available., Comment: 14 pages, 6 figures. Accepted for publication in the Open Journal of Astrophysics. Codes available at https://github.com/alessiospuriomancini/cosmopower, https://github.com/dpiras/cosmopower-jax, https://github.com/astro-informatics/harmonic/

We present the learned harmonic mean estimator with normalizing flows - a robust, scalable and flexible estimator of the Bayesian evidence for model comparison. Since the estimator is agnostic to sampling strategy and simply requires posterior samples, it can be applied to compute the evidence using any Markov chain Monte Carlo (MCMC) sampling technique, including saved down MCMC chains, or any variational inference approach. The learned harmonic mean estimator was recently introduced, where machine learning techniques were developed to learn a suitable internal importance sampling target distribution to solve the issue of exploding variance of the original harmonic mean estimator. In this article we present the use of normalizing flows as the internal machine learning technique within the learned harmonic mean estimator. Normalizing flows can be elegantly coupled with the learned harmonic mean to provide an approach that is more robust, flexible and scalable than the machine learning models considered previously. We perform a series of numerical experiments, applying our method to benchmark problems and to a cosmological example in up to 21 dimensions. We find the learned harmonic mean estimator is in agreement with ground truth values and nested sampling estimates. The open-source harmonic Python package implementing the learned harmonic mean, now with normalizing flows included, is publicly available., Comment: 14 pages, 8 figures, harmonic code available at https://github.com/astro-informatics/harmonic

Smart contracts have played a pivotal role in the evolution of blockchains and Decentralized Applications (DApps). As DApps continue to gain widespread adoption, multiple smart contract languages have been and are being made available to developers, each with its distinctive features, strengths, and weaknesses. In this paper, we examine the smart contract languages used in major blockchain platforms, with the goal of providing a comprehensive assessment of their main properties. Our analysis targets the programming languages rather than the underlying architecture: as a result, while we do consider the interplay between language design and blockchain model, our main focus remains on language-specific features such as usability, programming style, safety and security. To conduct our assessment, we propose an original benchmark which encompasses a wide, yet manageable, spectrum of key use cases that cut across all the smart contract languages under examination.

We present DE-VAE, a variational autoencoder (VAE) architecture to search for a compressed representation of dynamical dark energy (DE) models in observational studies of the cosmic large-scale structure. DE-VAE is trained on matter power spectra boosts generated at wavenumbers $k\in(0.01-2.5) \ h/\rm{Mpc}$ and at four redshift values $z\in(0.1,0.48,0.78,1.5)$ for the most typical dynamical DE parametrization with two extra parameters describing an evolving DE equation of state. The boosts are compressed to a lower-dimensional representation, which is concatenated with standard cold dark matter (CDM) parameters and then mapped back to reconstructed boosts; both the compression and the reconstruction components are parametrized as neural networks. Remarkably, we find that a single latent parameter is sufficient to predict 95% (99%) of DE power spectra generated over a broad range of cosmological parameters within $1\sigma$ ($2\sigma$) of a Gaussian error which includes cosmic variance, shot noise and systematic effects for a Stage IV-like survey. This single parameter shows a high mutual information with the two DE parameters, and these three variables can be linked together with an explicit equation through symbolic regression. Considering a model with two latent variables only marginally improves the accuracy of the predictions, and adding a third latent variable has no significant impact on the model's performance. We discuss how the DE-VAE architecture can be extended from a proof of concept to a general framework to be employed in the search for a common lower-dimensional parametrization of a wide range of beyond-$\Lambda$CDM models and for different cosmological datasets. Such a framework could then both inform the development of cosmological surveys by targeting optimal probes, and provide theoretical insight into the common phenomenological aspects of beyond-$\Lambda$CDM models., Comment: 11 pages, 5 figures. Minor changes to match version published in PRD

Evaluating the adversarial robustness of machine learning models using gradient-based attacks is challenging. In this work, we show that hyperparameter optimization can improve fast minimum-norm attacks by automating the selection of the loss function, the optimizer and the step-size scheduler, along with the corresponding hyperparameters. Our extensive evaluation involving several robust models demonstrates the improved efficacy of fast minimum-norm attacks when hyper-up with hyperparameter optimization. We release our open-source code at https://github.com/pralab/HO-FMN., Comment: Accepted at ESANN23

Neural network pruning has shown to be an effective technique for reducing the network size, trading desirable properties like generalization and robustness to adversarial attacks for higher sparsity. Recent work has claimed that adversarial pruning methods can produce sparse networks while also preserving robustness to adversarial examples. In this work, we first re-evaluate three state-of-the-art adversarial pruning methods, showing that their robustness was indeed overestimated. We then compare pruned and dense versions of the same models, discovering that samples on thin ice, i.e., closer to the unpruned model's decision boundary, are typically misclassified after pruning. We conclude by discussing how this intuition may lead to designing more effective adversarial pruning methods in future work.

Machine-learning models can be fooled by adversarial examples, i.e., carefully-crafted input perturbations that force models to output wrong predictions. While uncertainty quantification has been recently proposed to detect adversarial inputs, under the assumption that such attacks exhibit a higher prediction uncertainty than pristine data, it has been shown that adaptive attacks specifically aimed at reducing also the uncertainty estimate can easily bypass this defense mechanism. In this work, we focus on a different adversarial scenario in which the attacker is still interested in manipulating the uncertainty estimate, but regardless of the correctness of the prediction; in particular, the goal is to undermine the use of machine-learning models when their outputs are consumed by a downstream module or by a human operator. Following such direction, we: \textit{(i)} design a threat model for attacks targeting uncertainty quantification; \textit{(ii)} devise different attack strategies on conceptually different UQ techniques spanning for both classification and semantic segmentation problems; \textit{(iii)} conduct a first complete and extensive analysis to compare the differences between some of the most employed UQ approaches under attack. Our extensive experimental analysis shows that our attacks are more effective in manipulating uncertainty quantification measures than attacks aimed to also induce misclassifications.

ModSecurity is widely recognized as the standard open-source Web Application Firewall (WAF), maintained by the OWASP Foundation. It detects malicious requests by matching them against the Core Rule Set, identifying well-known attack patterns. Each rule in the CRS is manually assigned a weight, based on the severity of the corresponding attack, and a request is detected as malicious if the sum of the weights of the firing rules exceeds a given threshold. In this work, we show that this simple strategy is largely ineffective for detecting SQL injection (SQLi) attacks, as it tends to block many legitimate requests, while also being vulnerable to adversarial SQLi attacks, i.e., attacks intentionally manipulated to evade detection. To overcome these issues, we design a robust machine learning model, named AdvModSec, which uses the CRS rules as input features, and it is trained to detect adversarial SQLi attacks. Our experiments show that AdvModSec, being trained on the traffic directed towards the protected web services, achieves a better trade-off between detection and false positive rates, improving the detection rate of the vanilla version of ModSecurity with CRS by 21%. Moreover, our approach is able to improve its adversarial robustness against adversarial SQLi attacks by 42%, thereby taking a step forward towards building more robust and trustworthy WAFs.