Search

DarkSide-20k (DS-20k) is a dark matter detection experiment under construction at the Laboratori Nazionali del Gran Sasso (LNGS) in Italy. It utilises ~100 t of low radioactivity argon from an underground source (UAr) in its inner detector, with half serving as target in a dual-phase time projection chamber (TPC). The UAr cryogenics system must maintain stable thermodynamic conditions throughout the experiment's lifetime of >10 years. Continuous removal of impurities and radon from the UAr is essential for maximising signal yield and mitigating background. We are developing an efficient and powerful cryogenics system with a gas purification loop with a target circulation rate of 1000 slpm. Central to its design is a condenser operated with liquid nitrogen which is paired with a gas heat exchanger cascade, delivering a combined cooling power of >8 kW. Here we present the design choices in view of the DS-20k requirements, in particular the condenser's working principle and the cooling control, and we show test results obtained with a dedicated benchmarking platform at CERN and LNGS. We find that the thermal efficiency of the recirculation loop, defined in terms of nitrogen consumption per argon flow rate, is 95 % and the pressure in the test cryostat can be maintained within $\pm$(0.1-0.2) mbar. We further detail a 5-day cool-down procedure of the test cryostat, maintaining a cooling rate typically within -2 K/h, as required for the DS-20k inner detector. Additionally, we assess the circuit's flow resistance, and the heat transfer capabilities of two heat exchanger geometries for argon phase change, used to provide gas for recirculation. We conclude by discussing how our findings influence the finalisation of the system design, including necessary modifications to meet requirements and ongoing testing activities., Comment: 45 pages, 24 figures

The dual-phase liquid argon time projection chamber is presently one of the leading technologies to search for dark matter particles with masses below 10 GeV/c$^2$. This was demonstrated by the DarkSide-50 experiment with approximately 50 kg of low-radioactivity liquid argon as target material. The next generation experiment DarkSide-20k, currently under construction, will use 1,000 times more argon and is expected to start operation in 2027. Based on the DarkSide-50 experience, here we assess the DarkSide-20k sensitivity to models predicting light dark matter particles, including Weakly Interacting Massive Particles (WIMPs) and sub-GeV/c$^2$ particles interacting with electrons in argon atoms. With one year of data, a sensitivity improvement to dark matter interaction cross-sections by at least one order of magnitude with respect to DarkSide-50 is expected for all these models. A sensitivity to WIMP--nucleon interaction cross-sections below $1\times10^{-42}$ cm$^2$ is achievable for WIMP masses above 800 MeV/c$^2$. With 10 years exposure, the neutrino fog can be reached for WIMP masses around 5 GeV/c$^2$., Comment: submitted to Nature Communications

Experiments aimed at direct searches for WIMP dark matter require highly effective reduction of backgrounds and control of any residual radioactive contamination. In particular, neutrons interacting with atomic nuclei represent an important class of backgrounds due to the expected similarity of a WIMP-nucleon interaction, so that such experiments often feature a dedicated neutron detector surrounding the active target volume. In the context of the development of DarkSide-20k detector at INFN Gran Sasso National Laboratory (LNGS), several R&D projects were conceived and developed for the creation of a new hybrid material rich in both hydrogen and gadolinium nuclei to be employed as an essential element of the neutron detector. Thanks to its very high cross-section for neutron capture, gadolinium is one of the most widely used elements in neutron detectors, while the hydrogen-rich material is instrumental in efficiently moderating the neutrons. In this paper results from one of the R&Ds are presented. In this effort the new hybrid material was obtained as a poly(methyl methacrylate) (PMMA) matrix, loaded with gadolinium oxide in the form of nanoparticles. We describe its realization, including all phases of design, purification, construction, characterization, and determination of mechanical properties of the new material.

Weakly interacting massive particles (WIMPs) are well-motivated candidates for dark matter. One signature of galactic WIMPs is the annual modulation expected in a detector's interaction rate, which arises from Earth's revolution around the Sun. Over two decades, the DAMA/LIBRA experiment has observed such modulation with 250 kg of NaI(Tl) scintillators, in accordance with WIMP expectations but inconsistent with the negative results of other experiments. The signal depends on the target material, so to validate or refute the DAMA result, the experiment must be replicated using the same material. This is the goal of the ANAIS-112 experiment, currently underway since August 2017 with 112.5 kg of NaI(Tl). In this work, we present a reanalysis of three years of data employing an improved analysis chain to enhance the experimental sensitivity. The results presented here are consistent with the absence of modulation and inconsistent with DAMA's observation at nearly 3$\sigma$ confidence level, with the potential to reach a 5$\sigma$ level within 8 years from the beginning of the data collection. Additionally, we explore the impact of different scintillation quenching factors in the comparison between ANAIS-112 and DAMA/LIBRA., Comment: 16 pages, 10 figures, 1 table

The amount of light produced by nuclear recoils in scintillating targets is strongly quenched compared to that produced by electrons. A precise understanding of the quenching factor is particularly interesting for WIMP searches and CE{\nu}NS measurements since both rely on nuclear recoils, whereas energy calibrations are more readily accessible from electron recoils. There is a wide variation among the current measurements of the quenching factor in sodium iodide (NaI) crystals, especially below 10 keV, the energy region of interest for dark matter and CE{\nu}NS studies. A better understanding of the quenching factor in NaI(Tl) is of particular interest for resolving the decades-old puzzle in the field of dark matter between the null results of most WIMP searches and the claim for dark matter detection by the DAMA/LIBRA collaboration. In this work, we measured sodium and iodine quenching factors for five small NaI(Tl) crystals grown with similar thallium concentrations and growth procedures. Unlike previous experiments, multiple crystals were tested, with measurements made in the same experimental setup to control systematic effects. The quenching factors agree in all crystals we investigated, and both sodium and iodine quenching factors are smaller than those reported by DAMA/LIBRA. The dominant systematic effect was due to the electron equivalent energy calibration originating from the non-proportional behavior of the NaI(Tl) light yield at lower energies, potentially the cause for the discrepancies among the previous measurements., Comment: 22 pages, 27 figures, 7 tables. arXiv admin note: text overlap with arXiv:2310.07339

The representations of the activation space of deep neural networks (DNNs) are widely utilized for tasks like natural language processing, anomaly detection and speech recognition. Due to the diverse nature of these tasks and the large size of DNNs, an efficient and task-independent representation of activations becomes crucial. Empirical p-values have been used to quantify the relative strength of an observed node activation compared to activations created by already-known inputs. Nonetheless, keeping raw data for these calculations increases memory resource consumption and raises privacy concerns. To this end, we propose a model-agnostic framework for creating representations of activations in DNNs using node-specific histograms to compute p-values of observed activations without retaining already-known inputs. Our proposed approach demonstrates promising potential when validated with multiple network architectures across various downstream tasks and compared with the kernel density estimates and brute-force empirical baselines. In addition, the framework reduces memory usage by 30% with up to 4 times faster p-value computing time while maintaining state of-the-art detection power in downstream tasks such as the detection of adversarial attacks and synthesized content. Moreover, as we do not persist raw data at inference time, we could potentially reduce susceptibility to attacks and privacy issues.

We propose an auditing method to identify whether a large language model (LLM) encodes patterns such as hallucinations in its internal states, which may propagate to downstream tasks. We introduce a weakly supervised auditing technique using a subset scanning approach to detect anomalous patterns in LLM activations from pre-trained models. Importantly, our method does not need knowledge of the type of patterns a-priori. Instead, it relies on a reference dataset devoid of anomalies during testing. Further, our approach enables the identification of pivotal nodes responsible for encoding these patterns, which may offer crucial insights for fine-tuning specific sub-networks for bias mitigation. We introduce two new scanning methods to handle LLM activations for anomalous sentences that may deviate from the expected distribution in either direction. Our results confirm prior findings of BERT's limited internal capacity for encoding hallucinations, while OPT appears capable of encoding hallucination information internally. Importantly, our scanning approach, without prior exposure to false statements, performs comparably to a fully supervised out-of-distribution classifier.

ANAIS is a direct dark matter detection experiment whose goal is to confirm or refute in a model independent way the positive annual modulation signal claimed by DAMA/LIBRA. Consisting of 112.5 kg of NaI(Tl) scintillators, ANAIS-112 is taking data at the Canfranc Underground Laboratory in Spain since August, 2017. Results corresponding to the analysis of three years of data are compatible with the absence of modulation and incompatible with DAMA/LIBRA. However, testing this signal relies on the knowledge of the scintillation quenching factors (QF), which measure the relative efficiency for the conversion into light of the nuclear recoil energy with respect to the same energy deposited by electrons. Previous measurements of the QF in NaI(Tl) show a large dispersion. Consequently, in order to better understand the response of the ANAIS-112 detectors to nuclear recoils, a specific neutron calibration program has been developed. This program combines two different approaches: on the one hand, QF measurements were carried out in a monoenergetic neutron beam; on the other hand, the study presented here aims at the evaluation of the QF by exposing directly the ANAIS-112 crystals to neutrons from low activity $^{252}$Cf sources, placed outside the lead shielding. Comparison between these onsite neutron measurements and detailed GEANT4 simulations will be presented, confirming that this approach allows testing different QF models., Comment: Contributed to the TAUP2023 Conference, August-September 2023. To be published in Proceeding of Science

The ANAIS experiment is intended to search for dark matter annual modulation with ultrapure NaI(Tl) scintillators in order to provide a model independent confirmation or refutation of the long-standing DAMA/LIBRA positive annual modulation signal in the low energy detection rate, using the same target and technique. Other experiments exclude the region of parameters singled out by DAMA/LIBRA. However, these experiments use different target materials, so the comparison of their results depends on the models assumed for the dark matter particle and its distribution in the galactic halo. ANAIS-112, consisting of nine 12.5 kg NaI(Tl) modules produced by Alpha Spectra Inc., disposed in a 3$\times$3 matrix configuration, is taking data smoothly with excellent performance at the Canfranc Underground Laboratory, Spain, since August, 2017. Last published results corresponding to three-year exposure were compatible with the absence of modulation and incompatible with DAMA/LIBRA for a sensitivity above 2.5$\sigma$ C.L. Present status of the experiment and a reanalysis of the first 3 years data using new filtering protocols based on machine-learning techniques are reported. This reanalysis allows to improve the sensitivity previously achieved for the DAMA/LIBRA signal. Updated sensitivity prospects are also presented: with the improved filtering, testing the DAMA/LIBRA signal at 5$\sigma$ will be within reach in 2025., Comment: Contributed to the TAUP2023 Conference, August-September 2023. To be published in Proceeding of Science. arXiv admin note: substantial text overlap with arXiv:2110.10649, arXiv:2209.14113

The goal of the ANAIS-112 experiment, which operates at LSC, is to test in a model independent way the DAMA/LIBRA positive signal using the same target (NaI(Tl) crystals) and technique. Several strategies have been followed in this work to improve the sensitivity of the ANAIS-112 experiment. The Quenching Factor (QF) is a key factor in the comparison between different experiments. It has been determined both for iodine and sodium nuclear recoils in 5 crystals, for the first time in the same setup and using the same analysis protocol. Compatible values have been obtained for all of them. The calibration method in electron equivalent energy has proved to be critical in the QF estimation and allows understanding the discrepancies between previous measurements, but not concluding if the sodium QF is constant with energy or it increases up to about 80 keVnr. Therefore, further work is required to understand the conversion of the energy deposited in NaI(Tl) into light by different particles, as it is essential to develop a model for the QF. As a first step in an effort to understand the origin of some of the spurious event populations observed in the experiment, a GEANT4 simulation of an ANAIS-112 module has been developed including optical light emission and propagation after an energy deposit. Finally, the possibility of improving the sensitivity of NaI scintillators by replacing Photomultiplier Tubes with Silicon Photomultipliers as light sensors has been studied. These detectors offer very interesting opportunities within the new project ANAIS+. In this work we have characterized SiPMs from different manufacturers, established measurement protocols for the most relevant parameters, studied the properties of NaI and NaI(Tl) crystals at different temperatures and designed NaI+SiPM prototypes that will allow us to evaluate the limitations and opportunities of this new technology.

Addressing fairness in lesion classification from dermatological images is crucial due to variations in how skin diseases manifest across skin tones. However, the absence of skin tone labels in public datasets hinders building a fair classifier. To date, such skin tone labels have been estimated prior to fairness analysis in independent studies using the Individual Typology Angle (ITA). Briefly, ITA calculates an angle based on pixels extracted from skin images taking into account the lightness and yellow-blue tints. These angles are then categorised into skin tones that are subsequently used to analyse fairness in skin cancer classification. In this work, we review and compare four ITA-based approaches of skin tone classification on the ISIC18 dataset, a common benchmark for assessing skin cancer classification fairness in the literature. Our analyses reveal a high disagreement among previously published studies demonstrating the risks of ITA-based skin tone estimation methods. Moreover, we investigate the causes of such large discrepancy among these approaches and find that the lack of diversity in the ISIC18 dataset limits its use as a testbed for fairness analysis. Finally, we recommend further research on robust ITA estimation and diverse dataset acquisition with skin tone annotation to facilitate conclusive fairness assessments of artificial intelligence tools in dermatology. Our code is available at https://github.com/tkalbl/RevisitingSkinToneFairness., Comment: Accepted at 2023 MICCAI FAIMI Workshop

Despite major advances in artificial intelligence (AI) for medicine and healthcare, the deployment and adoption of AI technologies remain limited in real-world clinical practice. In recent years, concerns have been raised about the technical, clinical, ethical and legal risks associated with medical AI. To increase real world adoption, it is essential that medical AI tools are trusted and accepted by patients, clinicians, health organisations and authorities. This work describes the FUTURE-AI guideline as the first international consensus framework for guiding the development and deployment of trustworthy AI tools in healthcare. The FUTURE-AI consortium was founded in 2021 and currently comprises 118 inter-disciplinary experts from 51 countries representing all continents, including AI scientists, clinicians, ethicists, and social scientists. Over a two-year period, the consortium defined guiding principles and best practices for trustworthy AI through an iterative process comprising an in-depth literature review, a modified Delphi survey, and online consensus meetings. The FUTURE-AI framework was established based on 6 guiding principles for trustworthy AI in healthcare, i.e. Fairness, Universality, Traceability, Usability, Robustness and Explainability. Through consensus, a set of 28 best practices were defined, addressing technical, clinical, legal and socio-ethical dimensions. The recommendations cover the entire lifecycle of medical AI, from design, development and validation to regulation, deployment, and monitoring. FUTURE-AI is a risk-informed, assumption-free guideline which provides a structured approach for constructing medical AI tools that will be trusted, deployed and adopted in real-world practice. Researchers are encouraged to take the recommendations into account in proof-of-concept stages to facilitate future translation towards clinical practice of medical AI.

The direct search for dark matter in the form of weakly interacting massive particles (WIMP) is performed by detecting nuclear recoils (NR) produced in a target material from the WIMP elastic scattering. A promising experimental strategy for direct dark matter search employs argon dual-phase time projection chambers (TPC). One of the advantages of the TPC is the capability to detect both the scintillation and charge signals produced by NRs. Furthermore, the existence of a drift electric field in the TPC breaks the rotational symmetry: the angle between the drift field and the momentum of the recoiling nucleus can potentially affect the charge recombination probability in liquid argon and then the relative balance between the two signal channels. This fact could make the detector sensitive to the directionality of the WIMP-induced signal, enabling unmistakable annual and daily modulation signatures for future searches aiming for discovery. The Recoil Directionality (ReD) experiment was designed to probe for such directional sensitivity. The TPC of ReD was irradiated with neutrons at the INFN Laboratori Nazionali del Sud, and data were taken with 72 keV NRs of known recoil directions. The direction-dependent liquid argon charge recombination model by Cataudella et al. was adopted and a likelihood statistical analysis was performed, which gave no indications of significant dependence of the detector response to the recoil direction. The aspect ratio R of the initial ionization cloud is estimated to be 1.037 +/- 0.027 and the upper limit is R < 1.072 with 90% confidence level, Comment: 20 pages, 10 figures, submitted to Eur. Phys. J. C

This study, to the best of our knowledge, is the first to explore the intersection between lightweight cryptography (LWC) and advanced artificial intelligence (AI) language models. LWC, in particular the ASCON algorithm which has been selected as the LWC standard by the National Institute of Standards and Technology (NIST) in Feb. 2023, has become increasingly significant for preserving data security given the quick expansion and resource limitations of Internet of Things (IoT) devices. On the other hand, OpenAI's large language model (LLM) ChatGPT has demonstrated significant potential in producing complex, human-like text. This paper offers a novel method for implementing the NIST LWC standard, ASCON, using the GPT-4 model. Moreover, this paper details the design and functionality of ASCON, the procedures and actual Python implementation of ASCON using ChatGPT, and a discussion of the outcomes. The results contribute valuable insights into the efficient application of advanced AI language models in cryptography, particularly in constrained environments. Source code can be found at: https://github.com/DrCintas/ASCON-with-ChatGPT.

This survey is on forward-looking, emerging security concerns in post-quantum era, i.e., the implementation attacks for 2022 winners of NIST post-quantum cryptography (PQC) competition and thus the visions, insights, and discussions can be used as a step forward towards scrutinizing the new standards for applications ranging from Metaverse, Web 3.0 to deeply-embedded systems. The rapid advances in quantum computing have brought immense opportunities for scientific discovery and technological progress; however, it poses a major risk to today's security since advanced quantum computers are believed to break all traditional public-key cryptographic algorithms. This has led to active research on PQC algorithms that are believed to be secure against classical and powerful quantum computers. However, algorithmic security is unfortunately insufficient, and many cryptographic algorithms are vulnerable to side-channel attacks (SCA), where an attacker passively or actively gets side-channel data to compromise the security properties that are assumed to be safe theoretically. In this survey, we explore such imminent threats and their countermeasures with respect to PQC. We provide the respective, latest advancements in PQC research, as well as assessments and providing visions on the different types of SCAs.

This survey is the first work on the current standard for lightweight cryptography, standardized in 2023. Lightweight cryptography plays a vital role in securing resource-constrained embedded systems such as deeply-embedded systems (implantable and wearable medical devices, smart fabrics, smart homes, and the like), radio frequency identification (RFID) tags, sensor networks, and privacy-constrained usage models. National Institute of Standards and Technology (NIST) initiated a standardization process for lightweight cryptography and after a relatively-long multi-year effort, eventually, in Feb. 2023, the competition ended with ASCON as the winner. This lightweight cryptographic standard will be used in deeply-embedded architectures to provide security through confidentiality and integrity/authentication (the dual of the legacy AES-GCM block cipher which is the NIST standard for symmetric key cryptography). ASCON's lightweight design utilizes a 320-bit permutation which is bit-sliced into five 64-bit register words, providing 128-bit level security. This work summarizes the different implementations of ASCON on field-programmable gate array (FPGA) and ASIC hardware platforms on the basis of area, power, throughput, energy, and efficiency overheads. The presented work also reviews various differential and side-channel analysis attacks (SCAs) performed across variants of ASCON cipher suite in terms of algebraic, cube/cube-like, forgery, fault injection, and power analysis attacks as well as the countermeasures for these attacks. We also provide our insights and visions throughout this survey to provide new future directions in different domains. This survey is the first one in its kind and a step forward towards scrutinizing the advantages and future directions of the NIST lightweight cryptography standard introduced in 2023.

While the capabilities of generative models heavily improved in different domains (images, text, graphs, molecules, etc.), their evaluation metrics largely remain based on simplified quantities or manual inspection with limited practicality. To this end, we propose a framework for Multi-level Performance Evaluation of Generative mOdels (MPEGO), which could be employed across different domains. MPEGO aims to quantify generation performance hierarchically, starting from a sub-feature-based low-level evaluation to a global features-based high-level evaluation. MPEGO offers great customizability as the employed features are entirely user-driven and can thus be highly domain/problem-specific while being arbitrarily complex (e.g., outcomes of experimental procedures). We validate MPEGO using multiple generative models across several datasets from the material discovery domain. An ablation study is conducted to study the plausibility of intermediate steps in MPEGO. Results demonstrate that MPEGO provides a flexible, user-driven, and multi-level evaluation framework, with practical insights on the generation quality. The framework, source code, and experiments will be available at https://github.com/GT4SD/mpego.

The DAMA/LIBRA observation of an annual modulation in the detection rate compatible with that expected for dark matter particles from the galactic halo has accumulated evidence for more than twenty years. It is the only hint of a direct detection of the elusive dark matter, but it is in strong tension with the negative results of other very sensitive experiments, requiring ad-hoc scenarios to reconcile all the present experimental results. Testing the DAMA/LIBRA result using the same target material, NaI(Tl), removes the dependence on the particle and halo models and is the goal of the ANAIS-112 experiment, taking data at the Canfranc Underground Laboratory in Spain since August 2017 with 112.5 kg of NaI(Tl). At very low energies, the detection rate is dominated by non-bulk scintillation events and careful event selection is mandatory. This article summarizes the efforts devoted to better characterize and filter this contribution in ANAIS-112 data using a boosted decision tree (BDT), trained for this goal with high efficiency. We report on the selection of the training populations, the procedure to determine the optimal cut on the BDT parameter, the estimate of the efficiencies for the selection of bulk scintillation in the region of interest (ROI), and the evaluation of the performance of this analysis with respect to the previous filtering. The improvement achieved in background rejection in the ROI, but moreover, the increase in detection efficiency, push the ANAIS-112 sensitivity to test the DAMA/LIBRA annual modulation result around 3$\sigma$ with three-year exposure, being possible to reach 5$\sigma$ by extending the data taking for a few more years than the scheduled 5 years which were due in August 2022.

Dark matter lighter than 10 GeV/c$^2$ encompasses a promising range of candidates. A conceptual design for a new detector, DarkSide-LowMass, is presented, based on the DarkSide-50 detector and progress toward DarkSide-20k, optimized for a low-threshold electron-counting measurement. Sensitivity to light dark matter is explored for various potential energy thresholds and background rates. These studies show that DarkSide-LowMass can achieve sensitivity to light dark matter down to the solar neutrino floor for GeV-scale masses and significant sensitivity down to 10 MeV/c$^2$ considering the Migdal effect or interactions with electrons. Requirements for optimizing the detector's sensitivity are explored, as are potential sensitivity gains from modeling and mitigating spurious electron backgrounds that may dominate the signal at the lowest energies.

We report the results of six years (2013-2018) of measurements of $^{222}$Rn air concentration, relative humidity, atmospheric pressure and temperature in the halls A, B and C of the Canfranc Underground Laboratory (LSC). We have calculated all the Pearson correlation coefficients among these parameters and we have found a positive correlation between the $^{222}$Rn concentration and the relative humidity. Both correlated variables show a seasonal periodicity. The joint analysis of laboratory data and four years (2015-2018) of the meteorological variables outside the laboratory shows the correlation between the $^{222}$Rn concentration and the outside temperature. The collected information stresses the relevance of designing good Rn-mitigation strategies in current and future experiments at LSC; in particular, we have checked for two years (2017-2018) the good performance of the mitigation procedure of the ANAIS--112 experiment. Finally, in another measurement (2019-2021) for two years of live time, we report an upper limit to the residual $^{222}$Rn content of the radon-free air provided by the radon abatement system installed in the laboratory.

Data-centric AI encourages the need of cleaning and understanding of data in order to achieve trustworthy AI. Existing technologies, such as AutoML, make it easier to design and train models automatically, but there is a lack of a similar level of capabilities to extract data-centric insights. Manual stratification of tabular data per a feature (e.g., gender) is limited to scale up for higher feature dimension, which could be addressed using automatic discovery of divergent subgroups. Nonetheless, these automatic discovery techniques often search across potentially exponential combinations of features that could be simplified using a preceding feature selection step. Existing feature selection techniques for tabular data often involve fitting a particular model in order to select important features. However, such model-based selection is prone to model-bias and spurious correlations in addition to requiring extra resource to design, fine-tune and train a model. In this paper, we propose a model-free and sparsity-based automatic feature selection (SAFS) framework to facilitate automatic discovery of divergent subgroups. Different from filter-based selection techniques, we exploit the sparsity of objective measures among feature values to rank and select features. We validated SAFS across two publicly available datasets (MIMIC-III and Allstate Claims) and compared it with six existing feature selection methods. SAFS achieves a reduction of feature selection time by a factor of 81x and 104x, averaged cross the existing methods in the MIMIC-III and Claims datasets respectively. SAFS-selected features are also shown to achieve competitive detection performance, e.g., 18.3% of features selected by SAFS in the Claims dataset detected divergent samples similar to those detected by using the whole features with a Jaccard similarity of 0.95 but with a 16x reduction in detection time.

Deep generative models, such as Variational Autoencoders (VAEs) and Generative Adversarial Networks (GANs), have been employed widely in computational creativity research. However, such models discourage out-of-distribution generation to avoid spurious sample generation, thereby limiting their creativity. Thus, incorporating research on human creativity into generative deep learning techniques presents an opportunity to make their outputs more compelling and human-like. As we see the emergence of generative models directed toward creativity research, a need for machine learning-based surrogate metrics to characterize creative output from these models is imperative. We propose group-based subset scanning to identify, quantify, and characterize creative processes by detecting a subset of anomalous node-activations in the hidden layers of the generative models. Our experiments on the standard image benchmarks, and their "creatively generated" variants, reveal that the proposed subset scores distribution is more useful for detecting creative processes in the activation space rather than the pixel space. Further, we found that creative samples generate larger subsets of anomalies than normal or non-creative samples across datasets. The node activations highlighted during the creative decoding process are different from those responsible for the normal sample generation. Lastly, we assess if the images from the subsets selected by our method were also found creative by human evaluators, presenting a link between creativity perception in humans and node activations within deep neural nets., Comment: Accepted to IJCAI 2022 - Creativity Track - Extended version from Synthetic Data Generation Workshop at ICLR'21 submission (arXiv:2104.00479). arXiv admin note: text overlap with arXiv:2105.12479

Testing the DAMA/LIBRA annual modulation result independently of dark matter particle and halo models has been a challenge for twenty years. Using the same target material, NaI(Tl), is required and presently two experiments, ANAIS-112 and COSINE-100, are running for such a goal. A precise knowledge of the detector response to nuclear recoils is mandatory because this is the most likely channel to find the dark matter signal. The light produced by nuclear recoils is quenched with respect to that produced by electrons by a factor that has to be measured experimentally. However, current quenching factor measurements in NaI(Tl) crystals disagree within the energy region of interest for dark matter searches. To disentangle whether this discrepancy is due to intrinsic differences in the light response among different NaI(Tl) crystals, or has its origin in unaccounted for systematic effects will be key in the comparison among the different experiments. We present measurements of the quenching factors for five small NaI(Tl) crystals performed in the same experimental setup to control systematics. Quenching factor results are compatible between crystals and no clear dependence with energy is observed from 10 to 80 keVnr., Comment: Proceeding of the TAUP 2021 Conference

ANAIS is a direct dark matter detection experiment aiming at the confirmation or refutation of the DAMA/LIBRA positive annual modulation signal in the low energy detection rate, using the same target and technique. ANAIS-112, located at the Canfranc Underground Laboratory in Spain, is operating an array of 3$\times$3 ultrapure NaI(Tl) crystals with a total mass of 112.5 kg since August 2017. The trigger rate in the region of interest (1-6 keV) is dominated by non-bulk scintillation events. In order to discriminate these noise events from bulk scintillation events, robust filtering protocols have been developed. Although this filtering procedure works very well above 2 keV, the measured rate from 1 to 2 keV is about 50% higher than expected according to our background model, and we cannot discard non-bulk scintillation events as responsible of that excess. In order to improve the rejection of noise events, a Boosted Decision Tree has been developed and applied. With this new PMT-related noise rejection algorithm, the ANAIS-112 background between 1 and 2 keV is reduced by almost 30%, leading to an increase in sensitivity to the annual modulation signal. The reanalysis of the three years of ANAIS-112 data with this technique is also presented., Comment: Contributed to the TAUP2021 Conference, August 2021. To be published in Journal of Physics: Conference Series

The ANAIS (Annual modulation with NaI(Tl) Scintillators) experiment aims at the confirmation or refutation of the DAMA/LIBRA positive annual modulation signal in the low energy detection rate. ANAIS-112, consisting of nine 12.5 kg NaI(Tl) modules, is taking data since August, 2017 at the Canfranc Underground Laboratory (LSC) in Spain. Results from the analysis of three years of data are compatible with the absence of modulation. The background model developed for all nine ANAIS-112 detectors was established from commissioning data and non-blinded events in the first year of data taking. Now, background characterization is being improved profiting from the larger accumulated exposure available. Here, the background model is described and comparisons of model and measurements for energy spectra and counting rate time evolution for three-year exposure (considering different analysis conditions) are presented., Comment: Poster presented at TAUP2021 conference; to appear at Journal of Physics: Conference Series

Generative Adversarial Networks (GANs) have recently achieved unprecedented success in photo-realistic image synthesis from low-dimensional random noise. The ability to synthesize high-quality content at a large scale brings potential risks as the generated samples may lead to misinformation that can create severe social, political, health, and business hazards. We propose SubsetGAN to identify generated content by detecting a subset of anomalous node-activations in the inner layers of pre-trained neural networks. These nodes, as a group, maximize a non-parametric measure of divergence away from the expected distribution of activations created from real data. This enable us to identify synthesised images without prior knowledge of their distribution. SubsetGAN efficiently scores subsets of nodes and returns the group of nodes within the pre-trained classifier that contributed to the maximum score. The classifier can be a general fake classifier trained over samples from multiple sources or the discriminator network from different GANs. Our approach shows consistently higher detection power than existing detection methods across several state-of-the-art GANs (PGGAN, StarGAN, and CycleGAN) and over different proportions of generated content., Comment: The paper is under consideration at Pattern Recognition Letters

Recent advances in deep learning have led to breakthroughs in the development of automated skin disease classification. As we observe an increasing interest in these models in the dermatology space, it is crucial to address aspects such as the robustness towards input data distribution shifts. Current skin disease models could make incorrect inferences for test samples from different hardware devices and clinical settings or unknown disease samples, which are out-of-distribution (OOD) from the training samples. To this end, we propose a simple yet effective approach that detect these OOD samples prior to making any decision. The detection is performed via scanning in the latent space representation (e.g., activations of the inner layers of any pre-trained skin disease classifier). The input samples could also perturbed to maximise divergence of OOD samples. We validate our ODD detection approach in two use cases: 1) identify samples collected from different protocols, and 2) detect samples from unknown disease classes. Additionally, we evaluate the performance of the proposed approach and compare it with other state-of-the-art methods. Furthermore, data-driven dermatology applications may deepen the disparity in clinical care across racial and ethnic groups since most datasets are reported to suffer from bias in skin tone distribution. Therefore, we also evaluate the fairness of these OOD detection methods across different skin tones. Our experiments resulted in competitive performance across multiple datasets in detecting OOD samples, which could be used (in the future) to design more effective transfer learning techniques prior to inferring on these samples., Comment: Under review for 6th Outlier Detection & Description Workshop

Deep generative models, such as Variational Autoencoders (VAEs), have been employed widely in computational creativity research. However, such models discourage out-of-distribution generation to avoid spurious sample generation, limiting their creativity. Thus, incorporating research on human creativity into generative deep learning techniques presents an opportunity to make their outputs more compelling and human-like. As we see the emergence of generative models directed to creativity research, a need for machine learning-based surrogate metrics to characterize creative output from these models is imperative. We propose group-based subset scanning to quantify, detect, and characterize creative processes by detecting a subset of anomalous node-activations in the hidden layers of generative models. Our experiments on original, typically decoded, and "creatively decoded" (Das et al 2020) image datasets reveal that the proposed subset scores distribution is more useful for detecting creative processes in the activation space rather than the pixel space. Further, we found that creative samples generate larger subsets of anomalies than normal or non-creative samples across datasets. The node activations highlighted during the creative decoding process are different from those responsible for normal sample generation., Comment: Synthetic Data Generation Workshop at ICLR'21

ANAIS (Annual modulation with NaI Scintillators) is a dark matter direct detection experiment consisting of 112.5 kg of NaI(Tl) detectors in operation at the Canfranc Underground Laboratory (LSC), in Spain, since August 2017. ANAIS' goal is to confirm or refute in a model independent way the DAMA/LIBRA positive result: an annual modulation in the low-energy detection rate having all the features expected for the signal induced by dark matter particles in a standard galactic halo. This modulation, observed for about 20 years, is in strong tension with the negative results of other very sensitive experiments, but a model-independent comparison is still lacking. By using the same target material, NaI(Tl), such comparison is more direct and almost independent on dark matter particle and halo models. Here, we present the annual modulation analysis corresponding to three years of ANAIS data (for an effective exposure of 313.95 kg$\times$y), applying a blind procedure which updates that developed for the 1.5 years analysis, and later applied to 2 years. The analysis also improves the background modelling in the fitting of the region of interest rates. We obtain for the best fit in the [1-6] keV ([2-6] keV) energy region a modulation amplitude of -0.0034$\pm$0.0042 cpd/kg/keV (0.0003$\pm$0.0037 cpd/kg/keV), supporting the absence of modulation in our data, and incompatible with DAMA/LIBRA result at 3.3 (2.6) $\sigma$, for a sensitivity of 2.5 (2.7) $\sigma$. Moreover, we include two complementary analyses: a phase-free annual modulation search and the exploration of the possible presence of a periodic signal at other frequencies. Finally, we carry out several consistency checks of our result, and we update the ANAIS-112 projected sensitivity for the scheduled 5 years of operation., Comment: 20 pages, 20 figures

The Aria project consists of a plant, hosting a 350 m cryogenic isotopic distillation column, the tallest ever built, which is currently in the installation phase in a mine shaft at Carbosulcis S.p.A., Nuraxi-Figus (SU), Italy. Aria is one of the pillars of the argon dark-matter search experimental program, lead by the Global Argon Dark Matter Collaboration. Aria was designed to reduce the isotopic abundance of $^{39}$Ar, a $\beta$-emitter of cosmogenic origin, whose activity poses background and pile-up concerns in the detectors, in the argon used for the dark-matter searches, the so-called Underground Argon (UAr). In this paper, we discuss the requirements, design, construction, tests, and projected performance of the plant for the isotopic cryogenic distillation of argon. We also present the successful results of isotopic cryogenic distillation of nitrogen with a prototype plant, operating the column at total reflux.

Existing datasets available to address crucial problems, such as child mortality and family planning discontinuation in developing countries, are not ample for data-driven approaches. This is partly due to disjoint data collection efforts employed across locations, times, and variations of modalities. On the other hand, state-of-the-art methods for small data problem are confined to image modalities. In this work, we proposed a data-level linkage of disjoint surveys across Sub-Saharan African countries to improve prediction performance of neonatal death and provide cross-domain explainability., Comment: 3 pages

Future liquid-argon DarkSide-20k and ARGO detectors, designed for direct dark matter search, will be sensitive also to core-collapse supernova neutrinos, via coherent elastic neutrino-nucleus scattering. This interaction channel is flavor-insensitive with a high-cross section, enabling for a high-statistics neutrino detection with target masses of $\sim$50~t and $\sim$360~t for DarkSide-20k and ARGO, respectively. Thanks to the low-energy threshold of $\sim$0.5~keV$_{nr}$ achievable by exploiting the ionization channel, DarkSide-20k and ARGO have the potential to discover supernova bursts throughout our galaxy and up to the Small Magellanic Cloud, respectively, assuming a 11-M$_{\odot}$ progenitor star. We report also on the sensitivity to the neutronization burst, whose electron neutrino flux is suppressed by oscillations when detected via charged current and elastic scattering. Finally, the accuracies in the reconstruction of the average and total neutrino energy in the different phases of the supernova burst, as well as its time profile, are also discussed, taking into account the expected background and the detector response., Comment: 21 pages, 8 figures

Audio processing models based on deep neural networks are susceptible to adversarial attacks even when the adversarial audio waveform is 99.9% similar to a benign sample. Given the wide application of DNN-based audio recognition systems, detecting the presence of adversarial examples is of high practical relevance. By applying anomalous pattern detection techniques in the activation space of these models, we show that 2 of the recent and current state-of-the-art adversarial attacks on audio processing systems systematically lead to higher-than-expected activation at some subset of nodes and we can detect these with up to an AUC of 0.98 with no degradation in performance on benign samples.