Search

This paper presents preliminary work on a novel connection between certified robustness in machine learning and the modeling of 3D objects. We highlight an intriguing link between the Maximal Certified Radius (MCR) of a classifier representing a space's occupancy and the space's Signed Distance Function (SDF). Leveraging this relationship, we propose to use the certification method of randomized smoothing (RS) to compute SDFs. Since RS' high computational cost prevents its practical usage as a way to compute SDFs, we propose an algorithm to efficiently run RS in low-dimensional applications, such as 3D space, by expressing RS' fundamental operations as Gaussian smoothing on pre-computed voxel grids. Our approach offers an innovative and practical tool to compute SDFs, validated through proof-of-concept experiments in novel view synthesis. This paper bridges two previously disparate areas of machine learning, opening new avenues for further exploration and potential cross-domain advancements., Comment: This paper is an accepted extended abstract to the LatinX workshop at ICCV 2023. This was uploaded a year late

The goal of the PolarKID project is testing a new method for the measurement of polarized sources, in order to identify all the possible instrumental systematic effects that could impact the detection of CMB B-modes of polarization. It employs the KISS (KIDs Interferometer Spectrum Survey) instrument coupled to a sky simulator and to sources such as point-like black bodies (simulating planets), a dipole (extended source) and a polarizer. We use filled-arrays Lumped Element Kinetic Inductance Detectors (LEKIDs) since they have multiple advantages when observing both in a photometry and in a polarimetry configuration, Comment: 8 pages, 1 figure, Proceeding of the SPIE conference Millimeter, Submillimeter, and Far-Infrared Detectors and Instrumentation for Astronomy XII, SPIE Astronomical Telescopes + Instrumentation 2024

The study of Twitter as a means for analyzing social phenomena has gained interest in recent years due to the availability of large amounts of data in a relatively spontaneous environment. Within opinion-mining tasks, emotion detection is specially relevant, as it allows for the identification of people's subjective responses to different social events in a more granular way than traditional sentiment analysis based on polarity. In the particular case of political events, the analysis of emotions in social networks can provide valuable information on the perception of candidates, proposals, and other important aspects of the public debate. In spite of this importance, there are few studies on emotion detection in Spanish and, to the best of our knowledge, few resources are public for opinion mining in Colombian Spanish, highlighting the need for generating resources addressing the specific cultural characteristics of this variety. In this work, we present a small corpus of tweets in Spanish related to the 2022 Colombian presidential elections, manually labeled with emotions using a fine-grained taxonomy. We perform classification experiments using supervised state-of-the-art models (BERT models) and compare them with GPT-3.5 in few-shot learning settings. We make our dataset and code publicly available for research purposes.

Modern astrophysics relies on intricate instrument setups to meet the demands of sensitivity, sky coverage, and multi-channel observations. An example is the CONCERTO project, employing advanced technology like kinetic inductance detectors and a Martin-Puplett interferometer. This instrument, installed at the APEX telescope atop the Chajnantor plateau, began commissioning observations in April 2021. Following a successful commissioning phase that concluded in June 2021, CONCERTO was offered to the scientific community for observations, with a final observing run in December 2022. CONCERTO boasts an 18.5 arcmin field of view and a spectral resolution down to 1.45 GHz in the 130-310 GHz electromagnetic band. We developed a comprehensive instrument model of CONCERTO inspired by Fourier transform spectrometry principles to optimize performance and address systematic errors. This model integrates instrument noises, subsystem characteristics, and celestial signals, leveraging both physical data and simulations. Our methodology involves delineating simulation components, executing on-sky simulations, and comparing results with real observations. The resulting instrument model is pivotal, enabling a precise error correction and enhancing the reliability of astrophysical insights obtained from observational data. In this work, we focus on the description of three white-noise noise components included in the instrument model that characterize the white-noise level: the photon, the generation-recombination, and the amplifier noises., Comment: 8 pages, 1 figure, Proceeding of the SPIE conference Millimeter, Submillimeter, and Far-Infrared Detectors and Instrumentation for Astronomy XII, SPIE Astronomical Telescopes + Instrumentation 2024

This study investigates whether Compressed-Language Models (CLMs), i.e. language models operating on raw byte streams from Compressed File Formats~(CFFs), can understand files compressed by CFFs. We focus on the JPEG format as a representative CFF, given its commonality and its representativeness of key concepts in compression, such as entropy coding and run-length encoding. We test if CLMs understand the JPEG format by probing their capabilities to perform along three axes: recognition of inherent file properties, handling of files with anomalies, and generation of new files. Our findings demonstrate that CLMs can effectively perform these tasks. These results suggest that CLMs can understand the semantics of compressed data when directly operating on the byte streams of files produced by CFFs. The possibility to directly operate on raw compressed files offers the promise to leverage some of their remarkable characteristics, such as their ubiquity, compactness, multi-modality and segment-nature.

Climate risk assessment is becoming increasingly important. For organisations, identifying and assessing climate-related risks is challenging, as they can come from multiple sources. This study identifies and assesses the main climate transition risks in the colombian processed food sector. As transition risks are vague, our approach uses Fuzzy Logic and compares it to various multi-criteria decision-making methods to classify the different climate transition risks an organisation may be exposed to. This approach allows us to use linguistic expressions for risk analysis and to better describe risks and their consequences. The results show that the risks ranked as the most critical for this organisation in their order were price volatility and raw materials availability, the change to less carbon-intensive production or consumption patterns, the increase in carbon taxes and technological change, and the associated development or implementation costs. These risks show a critical risk level, which implies that they are the most significant risks for the organisation in the case study. These results highlight the importance of investments needed to meet regulatory requirements, which are the main drivers for organisations at the financial level.

Synthetic data from generative models emerges as the privacy-preserving data-sharing solution. Such a synthetic data set shall resemble the original data without revealing identifiable private information. The backbone technology of tabular synthesizers is rooted in image generative models, ranging from Generative Adversarial Networks (GANs) to recent diffusion models. Recent prior work sheds light on the utility-privacy tradeoff on tabular data, revealing and quantifying privacy risks on synthetic data. We first conduct an exhaustive empirical analysis, highlighting the utility-privacy tradeoff of five state-of-the-art tabular synthesizers, against eight privacy attacks, with a special focus on membership inference attacks. Motivated by the observation of high data quality but also high privacy risk in tabular diffusion, we propose DP-TLDM, Differentially Private Tabular Latent Diffusion Model, which is composed of an autoencoder network to encode the tabular data and a latent diffusion model to synthesize the latent tables. Following the emerging f-DP framework, we apply DP-SGD to train the auto-encoder in combination with batch clipping and use the separation value as the privacy metric to better capture the privacy gain from DP algorithms. Our empirical evaluation demonstrates that DP-TLDM is capable of achieving a meaningful theoretical privacy guarantee while also significantly enhancing the utility of synthetic data. Specifically, compared to other DP-protected tabular generative models, DP-TLDM improves the synthetic quality by an average of 35% in data resemblance, 15% in the utility for downstream tasks, and 50% in data discriminability, all while preserving a comparable level of privacy risk.

The conservation and authentication of pictorial artworks is considered an important part of the preservation of the cultural heritage. The use of non-destructive testing allows the obtaining of accurate information about the state of pictorial artworks, without direct contact between the equipment used and the sample. In particular, the use of this kind of technology is recommended in obtaining three-dimensional surface digital models, as it provides high-resolution information that constitutes a kind of fingerprint of the samples. In the case of pictorial artworks with some kind of surface relief, one of the most useful technologies is structured light (SL). In this paper the minimum difference in height that can be distinguished with this technology is estimated, establishing experimentally both the error committed in the measurement process and the precision in the use of this technology. The study, focused on the case of oil paintings on canvas, has been developed using a low-cost system to ensure its wide use.

In this work we present an integrated photogrammetric-acoustic technique that, together with the construction of a scaled wind tunnel, allows us to experimentally analyze the permeability behavior of a new type of acoustic screens based on a material called sonic crystal. Acoustic screens are devices used to reduce the noise, mostly due to the communication infrastructures, in its transmission phase from the source to the receiver. The main constructive difference between these new screens and the classic ones is that the first ones are formed by arrays of acoustic scatterers while the second ones are formed by continuous walls. This implies that, due to their geometry, screens based on sonic crystals are permeable to wind and water, unlike the classic ones. This fact may allow the use of these new screens in sandy soils, where sand would pass through the screen avoiding the formation of sand dunes that are formed in classic screens and drastically reduce their acoustic performance. In this work, the movement of the sand and the resulting acoustic attenuation in these new screens are analyzed qualitatively comparing the results with those obtained with the classic ones, obtaining interesting results under the acoustic point of view, Comment: 17 pages, 8 figures

There are several new projects to survey the sky with millimetre eyes, the biggest being Simons Observatory and CMB-S4, in the Southern Hemisphere. The NIKA2 collaboration has acquired sufficient knowledge to build a large focal plane KID camera for a 15~m antenna. This would allow covering the whole Northern Hemisphere in five years at subarcminute resolution and with milliJansky point-source sensitivity. We describe the main scientific drivers for such a project: the SZ sky, the high-redshift millimetre Universe and the interstellar medium in our Galaxy and the nearby galaxies. We also show briefly the main difficulties (scientific, organisational, technical and financial)., Comment: 6 pages, 4 figures, to appear in Proc. of the mm Universe 2023 conference, Grenoble (France), June 2023, published by F. Mayet et al. (Eds), EPJ Web of conferences, EDP Sciences

The correction of transcranial focused ultrasound aberrations is a relevant topic for enhancing various non-invasive medical treatments. Nowadays, the most widely accepted method to improve focusing is the emission through multi-element phased arrays; however, a new disruptive technology, based on 3D printed holographic acoustic lenses, has recently been proposed overcoming the spatial limitations of phased arrays due to the submillimetric precision of the latest generation of 3D printers. This works aims to optimize this recent solution; particularly, the preferred acoustic properties of the polymers used for printing the lens are systematically analyzed, paying special attention to the effect of p-wave speed and its relationship to the achievable voxel size of 3D printers. Results from simulations and experiments clearly show that there are optimal ranges for lens thickness and p-wave speed, fairly independent of the emitted frequency, the transducer aperture, or the transducer-target distance, given a particular voxel size, Comment: 24 pages, 9 figures

The behaviour of many dynamic real phenomena shows different phases, with each one following a sigmoidal type pattern. This requires studying sigmoidal curves with more than one inflection point. In this work, a diffusion process is introduced whose mean function is a curve of this type, concretely a transformation of the well-known Gompertz model after introducing in its expression a polynomial term. The maximum likelihood estimation of the parameters of the model is studied, and various criteria are provided for the selection of the degree of the polynomial when real situations are addressed. Finally, some simulated examples are presented., Comment: 17 pages, 28 figures

We consider a non homogeneous Gompertz diffusion process whose parameters are modified by generally time-dependent exogenous factors included in the infinitesimal moments. The proposed model is able to describe tumor dynamics under the effect of anti-proliferative and/or cell death-induced therapies. We assume that such therapies can modify also the infinitesimal variance of the diffusion process. An estimation procedure, based on a control group and two treated groups, is proposed to infer the model by estimating the constant parameters and the time-dependent terms. Moreover, several concatenated hypothesis tests are considered in order to confirm or reject the need to include time-dependent functions in the infinitesimal moments. Simulations are provided to evaluate the efficiency of the suggested procedures and to validate the testing hypothesis. Finally, an application to real data is considered., Comment: 44 pages, 63 figures. Includes Material Supplementary

This paper presents a comprehensive study on the role of Classifier-Free Guidance (CFG) in text-conditioned diffusion models from the perspective of inference efficiency. In particular, we relax the default choice of applying CFG in all diffusion steps and instead search for efficient guidance policies. We formulate the discovery of such policies in the differentiable Neural Architecture Search framework. Our findings suggest that the denoising steps proposed by CFG become increasingly aligned with simple conditional steps, which renders the extra neural network evaluation of CFG redundant, especially in the second half of the denoising process. Building upon this insight, we propose "Adaptive Guidance" (AG), an efficient variant of CFG, that adaptively omits network evaluations when the denoising process displays convergence. Our experiments demonstrate that AG preserves CFG's image quality while reducing computation by 25%. Thus, AG constitutes a plug-and-play alternative to Guidance Distillation, achieving 50% of the speed-ups of the latter while being training-free and retaining the capacity to handle negative prompts. Finally, we uncover further redundancies of CFG in the first half of the diffusion process, showing that entire neural function evaluations can be replaced by simple affine transformations of past score estimates. This method, termed LinearAG, offers even cheaper inference at the cost of deviating from the baseline model. Our findings provide insights into the efficiency of the conditional denoising process that contribute to more practical and swift deployment of text-conditioned diffusion models.

We provide an account of the construction of the moduli stack of elliptic curves as an analytic orbifold. While intimately linked to Thurston's point of view on the subject (discrete groups acting properly and effectively on differentiable manifolds), the construction of the modular orbi-curve and its universal family of elliptic curves ends up requiring a bit more technology, in order to allow for non-effective actions. The paper is entirely expository and makes no claims to originality: its main goal is to be self-contained enough in order to be useful to young researchers who are entering the field and are interested in the interactions between differential and algebraic geometry., Comment: To appear in 'In the tradition of Thurston III. Geometry and dynamics'

CONCERTO (CarbON CII line in post-rEionization and ReionizaTiOn) is a low-resolution Fourier transform spectrometer dedicated to the study of star-forming galaxies and clusters of galaxies in the transparent millimeter windows from the ground. It is characterized by a wide instantaneous 18.6 arcmin field of view, operates at 130-310 GHz, and was installed on the 12-meter Atacama Pathfinder Experiment (APEX) telescope at 5100 m above sea level. CONCERTO's double focal planes host two arrays of 2152 kinetic inductance detectors and represent a pioneering instrument to meet a state-of-the-art scientific challenge. This paper introduces the CONCERTO instrument and explains its status, shows the first CONCERTO spectral maps of Orion, and describes the perspectives of the project., Comment: 6 pages, 2 figures, to appear in Proc. of the mm Universe 2023 conference, Grenoble (France), June 2023, published by F. Mayet et al. (Eds), EPJ Web of conferences, EDP Sciences

Diffusion or flow-based models are powerful generative paradigms that are notoriously hard to sample as samples are defined as solutions to high-dimensional Ordinary or Stochastic Differential Equations (ODEs/SDEs) which require a large Number of Function Evaluations (NFE) to approximate well. Existing methods to alleviate the costly sampling process include model distillation and designing dedicated ODE solvers. However, distillation is costly to train and sometimes can deteriorate quality, while dedicated solvers still require relatively large NFE to produce high quality samples. In this paper we introduce "Bespoke solvers", a novel framework for constructing custom ODE solvers tailored to the ODE of a given pre-trained flow model. Our approach optimizes an order consistent and parameter-efficient solver (e.g., with 80 learnable parameters), is trained for roughly 1% of the GPU time required for training the pre-trained model, and significantly improves approximation and generation quality compared to dedicated solvers. For example, a Bespoke solver for a CIFAR10 model produces samples with Fr\'echet Inception Distance (FID) of 2.73 with 10 NFE, and gets to 1% of the Ground Truth (GT) FID (2.59) for this model with only 20 NFE. On the more challenging ImageNet-64$\times$64, Bespoke samples at 2.2 FID with 10 NFE, and gets within 2% of GT FID (1.71) with 20 NFE.

Quantum computing is a rapidly emerging and promising field that has the potential to revolutionize numerous research domains, including drug design, network technologies and sustainable energy. Due to the inherent complexity and divergence from classical computing, several major quantum computing libraries have been developed to implement quantum algorithms, namely IBM Qiskit, Amazon Braket, Cirq, PyQuil, and PennyLane. These libraries allow for quantum simulations on classical computers and facilitate program execution on corresponding quantum hardware, e.g., Qiskit programs on IBM quantum computers. While all platforms have some differences, the main concepts are the same. QInterpreter is a tool embedded in the Quantum Science Gateway QubitHub using Jupyter Notebooks that translates seamlessly programs from one library to the other and visualizes the results. It combines the five well-known quantum libraries: into a unified framework. Designed as an educational tool for beginners, Qinterpreter enables the development and execution of quantum circuits across various platforms in a straightforward way. The work highlights the versatility and accessibility of Qinterpreter in quantum programming and underscores our ultimate goal of pervading Quantum Computing through younger, less specialized, and diverse cultural and national communities.

The negative impact of label noise is well studied in classical supervised learning yet remains an open research question in meta-learning. Meta-learners aim to adapt to unseen learning tasks by learning a good initial model in meta-training and consecutively fine-tuning it according to new tasks during meta-testing. In this paper, we present the first extensive analysis of the impact of varying levels of label noise on the performance of state-of-the-art meta-learners, specifically gradient-based $N$-way $K$-shot learners. We show that the accuracy of Reptile, iMAML, and foMAML drops by up to 42% on the Omniglot and CifarFS datasets when meta-training is affected by label noise. To strengthen the resilience against label noise, we propose two sampling techniques, namely manifold (Man) and batch manifold (BatMan), which transform the noisy supervised learners into semi-supervised ones to increase the utility of noisy labels. We first construct manifold samples of $N$-way $2$-contrastive-shot tasks through augmentation, learning the embedding via a contrastive loss in meta-training, and then perform classification through zeroing on the embedding in meta-testing. We show that our approach can effectively mitigate the impact of meta-training label noise. Even with 60% wrong labels \batman and \man can limit the meta-testing accuracy drop to ${2.5}$, ${9.4}$, ${1.1}$ percent points, respectively, with existing meta-learners across the Omniglot, CifarFS, and MiniImagenet datasets., Comment: 10 pages,3 figures

Faithfully reconstructing 3D geometry and generating novel views of scenes are critical tasks in 3D computer vision. Despite the widespread use of image augmentations across computer vision applications, their potential remains underexplored when learning neural rendering methods (NRMs) for 3D scenes. This paper presents a comprehensive analysis of the use of image augmentations in NRMs, where we explore different augmentation strategies. We found that introducing image augmentations during training presents challenges such as geometric and photometric inconsistencies for learning NRMs from images. Specifically, geometric inconsistencies arise from alterations in shapes, positions, and orientations from the augmentations, disrupting spatial cues necessary for accurate 3D reconstruction. On the other hand, photometric inconsistencies arise from changes in pixel intensities introduced by the augmentations, affecting the ability to capture the underlying 3D structures of the scene. We alleviate these issues by focusing on color manipulations and introducing learnable appearance embeddings that allow NRMs to explain away photometric variations. Our experiments demonstrate the benefits of incorporating augmentations when learning NRMs, including improved photometric quality and surface reconstruction, as well as enhanced robustness against data quality issues, such as reduced training data and image degradations.

This paper proposes a novel online evaluation protocol for Test Time Adaptation (TTA) methods, which penalizes slower methods by providing them with fewer samples for adaptation. TTA methods leverage unlabeled data at test time to adapt to distribution shifts. Although many effective methods have been proposed, their impressive performance usually comes at the cost of significantly increased computation budgets. Current evaluation protocols overlook the effect of this extra computation cost, affecting their real-world applicability. To address this issue, we propose a more realistic evaluation protocol for TTA methods, where data is received in an online fashion from a constant-speed data stream, thereby accounting for the method's adaptation speed. We apply our proposed protocol to benchmark several TTA methods on multiple datasets and scenarios. Extensive experiments show that, when accounting for inference speed, simple and fast approaches can outperform more sophisticated but slower methods. For example, SHOT from 2020, outperforms the state-of-the-art method SAR from 2023 in this setting. Our results reveal the importance of developing practical TTA methods that are both accurate and efficient., Comment: Accepted to ICML 2024

This paper proposes a novel approach for rendering a pre-trained Neural Radiance Field (NeRF) in real-time on resource-constrained devices. We introduce Re-ReND, a method enabling Real-time Rendering of NeRFs across Devices. Re-ReND is designed to achieve real-time performance by converting the NeRF into a representation that can be efficiently processed by standard graphics pipelines. The proposed method distills the NeRF by extracting the learned density into a mesh, while the learned color information is factorized into a set of matrices that represent the scene's light field. Factorization implies the field is queried via inexpensive MLP-free matrix multiplications, while using a light field allows rendering a pixel by querying the field a single time-as opposed to hundreds of queries when employing a radiance field. Since the proposed representation can be implemented using a fragment shader, it can be directly integrated with standard rasterization frameworks. Our flexible implementation can render a NeRF in real-time with low memory requirements and on a wide range of resource-constrained devices, including mobiles and AR/VR headsets. Notably, we find that Re-ReND can achieve over a 2.6-fold increase in rendering speed versus the state-of-the-art without perceptible losses in quality.

For more than a century, Cantor's theory of transfinite numbers has played a pivotal role in set theory, with ramifications that extend to many areas of mathematics. This article extends earlier findings with a fresh look at the critical facts of Cantor's theory: i) Cantor's widely renowned Diagonalization Argument (CDA) is fully refuted by a set of counter-examples that expose the fallacy of this proof. ii) The logical inconsistencies of CDA are revisited, exposing the short-comings of CDA's implementation of the method of proof by contradiction. iii) The denumerability of the power set of the set of the natural numbers, P(N), is substantiated by a proof that takes full account of all the infinite subsets of N. Such a result confirms the denumerability of the set of the real numbers, R, and with it the countable nature of the continuum. iv) Given that the denumerable character of (probably) all infinite sets makes their comparison in terms of one-to-one correspondences a rather pointless exercise, a new concept of relative cardinality is introduced which facilitates a quantitative evaluation of their different magnitudes., Comment: 22 pages, 2 figures, 1 table

A magnetic insulator is an ideal platform to propagate spin information by exploiting magnon currents. However, until now, most studies have focused on Y$_3$Fe$_5$O$_{12}$ (YIG) and a few other ferri- and antiferromagnetic insulators, but not on pure ferromagnets. In this study, we demonstrate for the first time that magnon currents can propagate in ferromagnetic insulating thin films of EuS. By performing both local and non-local transport measurements in 18-nm-thick films of EuS using Pt electrodes, we detect magnon currents arising from thermal generation by the spin Seebeck effect. By comparing the dependence of the local and non-local signals with the temperature (< 30 K) and magnetic field (< 9 T), we confirm the magnon transport origin of the non-local signal. Finally, we extract the magnon diffusion length in the EuS film (~140 nm), a short value in good correspondence with the large Gilbert damping measured in the same film., Comment: 17 pages, 5 figures, and Supplemental Material

In recent years, hate speech has gained great relevance in social networks and other virtual media because of its intensity and its relationship with violent acts against members of protected groups. Due to the great amount of content generated by users, great effort has been made in the research and development of automatic tools to aid the analysis and moderation of this speech, at least in its most threatening forms. One of the limitations of current approaches to automatic hate speech detection is the lack of context. Most studies and resources are performed on data without context; that is, isolated messages without any type of conversational context or the topic being discussed. This restricts the available information to define if a post on a social network is hateful or not. In this work, we provide a novel corpus for contextualized hate speech detection based on user responses to news posts from media outlets on Twitter. This corpus was collected in the Rioplatense dialectal variety of Spanish and focuses on hate speech associated with the COVID-19 pandemic. Classification experiments using state-of-the-art techniques show evidence that adding contextual information improves hate speech detection performance for two proposed tasks (binary and multi-label prediction). We make our code, models, and corpus available for further research.