Search

A promising way to mitigate the expensive process of obtaining a high-dimensional signal is to acquire a limited number of low-dimensional measurements and solve an under-determined inverse problem by utilizing the structural prior about the signal. In this paper, we focus on adaptive acquisition schemes to save further the number of measurements. To this end, we propose a reinforcement learning-based approach that sequentially collects measurements to better recover the underlying signal by acquiring fewer measurements. Our approach applies to general inverse problems with continuous action spaces and jointly learns the recovery algorithm. Using insights obtained from theoretical analysis, we also provide a probabilistic design for our methods using variational formulation. We evaluate our approach on multiple datasets and with two measurement spaces (Gaussian, Radon). Our results confirm the benefits of adaptive strategies in low-acquisition horizon settings.

Compressed sensing combines the power of convex optimization techniques with a sparsity-inducing prior on the signal space to solve an underdetermined system of equations. For many problems, the sparsifying dictionary is not directly given, nor its existence can be assumed. Besides, the sensing matrix can change across different scenarios. Addressing these issues requires solving a sparse representation learning problem, namely dictionary learning, taking into account the epistemic uncertainty of the learned dictionaries and, finally, jointly learning sparse representations and reconstructions under varying sensing matrix conditions. We address both concerns by proposing a variant of the LISTA architecture. First, we introduce Augmented Dictionary Learning ISTA (A-DLISTA), which incorporates an augmentation module to adapt parameters to the current measurement setup. Then, we propose to learn a distribution over dictionaries via a variational approach, dubbed Variational Learning ISTA (VLISTA). VLISTA exploits A-DLISTA as the likelihood model and approximates a posterior distribution over the dictionaries as part of an unfolded LISTA-based recovery algorithm. As a result, VLISTA provides a probabilistic way to jointly learn the dictionary distribution and the reconstruction algorithm with varying sensing matrices. We provide theoretical and experimental support for our architecture and show that our model learns calibrated uncertainties.

In recent years, solving optimization problems involving black-box simulators has become a point of focus for the machine learning community due to their ubiquity in science and engineering. The simulators describe a forward process $f_{\mathrm{sim}}: (\psi, x) \rightarrow y$ from simulation parameters $\psi$ and input data $x$ to observations $y$, and the goal of the optimization problem is to find parameters $\psi$ that minimize a desired loss function. Sophisticated optimization algorithms typically require gradient information regarding the forward process, $f_{\mathrm{sim}}$, with respect to the parameters $\psi$. However, obtaining gradients from black-box simulators can often be prohibitively expensive or, in some cases, impossible. Furthermore, in many applications, practitioners aim to solve a set of related problems. Thus, starting the optimization ``ab initio", i.e. from scratch, each time might be inefficient if the forward model is expensive to evaluate. To address those challenges, this paper introduces a novel method for solving classes of similar black-box optimization problems by learning an active learning policy that guides a differentiable surrogate's training and uses the surrogate's gradients to optimize the simulation parameters with gradient descent. After training the policy, downstream optimization of problems involving black-box simulators requires up to $\sim$90\% fewer expensive simulator calls compared to baselines such as local surrogate-based approaches, numerical optimization, and Bayesian methods.

Conformal Prediction (CP) is a distribution-free uncertainty estimation framework that constructs prediction sets guaranteed to contain the true answer with a user-specified probability. Intuitively, the size of the prediction set encodes a general notion of uncertainty, with larger sets associated with higher degrees of uncertainty. In this work, we leverage information theory to connect conformal prediction to other notions of uncertainty. More precisely, we prove three different ways to upper bound the intrinsic uncertainty, as described by the conditional entropy of the target variable given the inputs, by combining CP with information theoretical inequalities. Moreover, we demonstrate two direct and useful applications of such connection between conformal prediction and information theory: (i) more principled and effective conformal training objectives that generalize previous approaches and enable end-to-end training of machine learning models from scratch, and (ii) a natural mechanism to incorporate side information into conformal prediction. We empirically validate both applications in centralized and federated learning settings, showing our theoretical results translate to lower inefficiency (average prediction set size) for popular CP methods.

In the context of communication networks, digital twin technology provides a means to replicate the radio frequency (RF) propagation environment as well as the system behaviour, allowing for a way to optimize the performance of a deployed system based on simulations. One of the key challenges in the application of Digital Twin technology to mmWave systems is the prevalent channel simulators' stringent requirements on the accuracy of the 3D Digital Twin, reducing the feasibility of the technology in real applications. We propose a practical Digital Twin creation pipeline and a channel simulator, that relies only on a single mounted camera and position information. We demonstrate the performance benefits compared to methods that do not explicitly model the 3D environment, on downstream sub-tasks in beam acquisition, using the real-world dataset of the DeepSense6G challenge, Comment: ICC2024 accepted paper. Copyright IEEE

Analog beamforming is the predominant approach for millimeter wave (mmWave) communication given its favorable characteristics for limited-resource devices. In this work, we aim at reducing the spectral efficiency gap between analog and digital beamforming methods. We propose a method for refined beam selection based on the estimated raw channel. The channel estimation, an underdetermined problem, is solved using compressed sensing (CS) methods leveraging angular domain sparsity of the channel. To reduce the complexity of CS methods, we propose dictionary learning iterative soft-thresholding algorithm, which jointly learns the sparsifying dictionary and signal reconstruction. We evaluate the proposed method on a realistic mmWave setup and show considerable performance improvement with respect to code-book based analog beamforming approaches.

In compressed sensing, the goal is to reconstruct the signal from an underdetermined system of linear measurements. Thus, prior knowledge about the signal of interest and its structure is required. Additionally, in many scenarios, the signal has an unknown orientation prior to measurements. To address such recovery problems, we propose using equivariant generative models as a prior, which encapsulate orientation information in their latent space. Thereby, we show that signals with unknown orientations can be recovered with iterative gradient descent on the latent space of these models and provide additional theoretical recovery guarantees. We construct an equivariant variational autoencoder and use the decoder as generative prior for compressed sensing. We discuss additional potential gains of the proposed approach in terms of convergence and latency.

In recent years, Quantum Computing witnessed massive improvements in terms of available resources and algorithms development. The ability to harness quantum phenomena to solve computational problems is a long-standing dream that has drawn the scientific community's interest since the late 80s. In such a context, we propose our contribution. First, we introduce basic concepts related to quantum computations, and then we explain the core functionalities of technologies that implement the Gate Model and Adiabatic Quantum Computing paradigms. Finally, we gather, compare and analyze the current state-of-the-art concerning Quantum Perceptrons and Quantum Neural Networks implementations.

Emotions play a central role in the social life of every human being, and their study, which represents a multidisciplinary subject, embraces a great variety of research fields. Especially concerning the latter, the analysis of facial expressions represents a very active research area due to its relevance to human-computer interaction applications. In such a context, Facial Expression Recognition (FER) is the task of recognizing expressions on human faces. Typically, face images are acquired by cameras that have, by nature, different characteristics, such as the output resolution. It has been already shown in the literature that Deep Learning models applied to face recognition experience a degradation in their performance when tested against multi-resolution scenarios. Since the FER task involves analyzing face images that can be acquired with heterogeneous sources, thus involving images with different quality, it is plausible to expect that resolution plays an important role in such a case too. Stemming from such a hypothesis, we prove the benefits of multi-resolution training for models tasked with recognizing facial expressions. Hence, we propose a two-step learning procedure, named MAFER, to train DCNNs to empower them to generate robust predictions across a wide range of resolutions. A relevant feature of MAFER is that it is task-agnostic, i.e., it can be used complementarily to other objective-related techniques. To assess the effectiveness of the proposed approach, we performed an extensive experimental campaign on publicly available datasets: \fer{}, \raf{}, and \oulu{}. For a multi-resolution context, we observe that with our approach, learning models improve upon the current SotA while reporting comparable results in fix-resolution contexts. Finally, we analyze the performance of our models and observe the higher discrimination power of deep features generated from them.

Facial expressions play a fundamental role in human communication. Indeed, they typically reveal the real emotional status of people beyond the spoken language. Moreover, the comprehension of human affect based on visual patterns is a key ingredient for any human-machine interaction system and, for such reasons, the task of Facial Expression Recognition (FER) draws both scientific and industrial interest. In the recent years, Deep Learning techniques reached very high performance on FER by exploiting different architectures and learning paradigms. In such a context, we propose a multi-resolution approach to solve the FER task. We ground our intuition on the observation that often faces images are acquired at different resolutions. Thus, directly considering such property while training a model can help achieve higher performance on recognizing facial expressions. To our aim, we use a ResNet-like architecture, equipped with Squeeze-and-Excitation blocks, trained on the Affect-in-the-Wild 2 dataset. Not being available a test set, we conduct tests and models selection by employing the validation set only on which we achieve more than 90\% accuracy on classifying the seven expressions that the dataset comprises.

Facial Expression Recognition(FER) is one of the most important topic in Human-Computer interactions(HCI). In this work we report details and experimental results about a facial expression recognition method based on state-of-the-art methods. We fine-tuned a SeNet deep learning architecture pre-trained on the well-known VGGFace2 dataset, on the AffWild2 facial expression recognition dataset. The main goal of this work is to define a baseline for a novel method we are going to propose in the near future. This paper is also required by the Affective Behavior Analysis in-the-wild (ABAW) competition in order to evaluate on the test set this approach. The results reported here are on the validation set and are related on the Expression Challenge part (seven basic emotion recognition) of the competition. We will update them as soon as the actual results on the test set will be published on the leaderboard.

Anomalies are ubiquitous in all scientific fields and can express an unexpected event due to incomplete knowledge about the data distribution or an unknown process that suddenly comes into play and distorts observations. Due to such events' rarity, to train deep learning models on the Anomaly Detection (AD) task, scientists only rely on "normal" data, i.e., non-anomalous samples. Thus, letting the neural network infer the distribution beneath the input data. In such a context, we propose a novel framework, named Multi-layer One-Class ClassificAtion (MOCCA),to train and test deep learning models on the AD task. Specifically, we applied it to autoencoders. A key novelty in our work stems from the explicit optimization of intermediate representations for the AD task. Indeed, differently from commonly used approaches that consider a neural network as a single computational block, i.e., using the output of the last layer only, MOCCA explicitly leverages the multi-layer structure of deep architectures. Each layer's feature space is optimized for AD during training, while in the test phase, the deep representations extracted from the trained layers are combined to detect anomalies. With MOCCA, we split the training process into two steps. First, the autoencoder is trained on the reconstruction task only. Then, we only retain the encoder tasked with minimizing the L_2 distance between the output representation and a reference point, the anomaly-free training data centroid, at each considered layer. Subsequently, we combine the deep features extracted at the various trained layers of the encoder model to detect anomalies at inference time. To assess the performance of the models trained with MOCCA, we conduct extensive experiments on publicly available datasets. We show that our proposed method reaches comparable or superior performance to state-of-the-art approaches available in the literature., Comment: The paper has been accepted for publication in the IEEE Transactions on Neural Networks and Learning Systems, Special Issue on Deep Learning for Anomaly Detection

Deep Learning methods have become state-of-the-art for solving tasks such as Face Recognition (FR). Unfortunately, despite their success, it has been pointed out that these learning models are exposed to adversarial inputs - images to which an imperceptible amount of noise for humans is added to maliciously fool a neural network - thus limiting their adoption in real-world applications. While it is true that an enormous effort has been spent in order to train robust models against this type of threat, adversarial detection techniques have recently started to draw attention within the scientific community. A detection approach has the advantage that it does not require to re-train any model, thus it can be added on top of any system. In this context, we present our work on adversarial samples detection in forensics mainly focused on detecting attacks against FR systems in which the learning model is typically used only as a features extractor. Thus, in these cases, train a more robust classifier might not be enough to defence a FR system. In this frame, the contribution of our work is four-fold: i) we tested our recently proposed adversarial detection approach against classifier attacks, i.e. adversarial samples crafted to fool a FR neural network acting as a classifier; ii) using a k-Nearest Neighbor (kNN) algorithm as a guidance, we generated deep features attacks against a FR system based on a DL model acting as features extractor, followed by a kNN which gives back the query identity based on features similarity; iii) we used the deep features attacks to fool a FR system on the 1:1 Face Verification task and we showed their superior effectiveness with respect to classifier attacks in fooling such type of system; iv) we used the detectors trained on classifier attacks to detect deep features attacks, thus showing that such approach is generalizable to different types of offensives.

Convolutional Neural Networks have reached extremely high performances on the Face Recognition task. Largely used datasets, such as VGGFace2, focus on gender, pose and age variations trying to balance them to achieve better results. However, the fact that images have different resolutions is not usually discussed and resize to 256 pixels before cropping is used. While specific datasets for very low resolution faces have been proposed, less attention has been payed on the task of cross-resolution matching. Such scenarios are of particular interest for forensic and surveillance systems in which it usually happens that a low-resolution probe has to be matched with higher-resolution galleries. While it is always possible to either increase the resolution of the probe image or to reduce the size of the gallery images, to the best of our knowledge an extensive experimentation of cross-resolution matching was missing in the recent deep learning based literature. In the context of low- and cross-resolution Face Recognition, the contributions of our work are: i) we proposed a training method to fine-tune a state-of-the-art model in order to make it able to extract resolution-robust deep features; ii) we tested our models on the benchmark datasets IJB-B/C considering images at both full and low resolutions in order to show the effectiveness of the proposed training algorithm. To the best of our knowledge, this is the first work testing extensively the performance of a FR model in a cross-resolution scenario; iii) we tested our models on the low resolution and low quality datasets QMUL-SurvFace and TinyFace and showed their superior performances, even though we did not train our model on low-resolution faces only and our main focus was cross-resolution; iv) we showed that our approach can be more effective with respect to preprocessing faces with super resolution techniques.

We report on the response of liquid xenon to low energy electronic recoils below 15 keV from beta decays of tritium at drift fields of 92 V/cm, 154 V/cm and 366 V/cm using the XENON100 detector. A data-to-simulation fitting method based on Markov Chain Monte Carlo is used to extract the photon yields and recombination fluctuations from the experimental data. The photon yields measured at the two lower fields are in agreement with those from literature; additional measurements at a higher field of 366 V/cm are presented. The electronic and nuclear recoil discrimination as well as its dependence on the drift field and photon detection efficiency are investigated at these low energies. The results provide new measurements in the energy region of interest for dark matter searches using liquid xenon.

We present results of searches for vector and pseudo-scalar bosonic super-WIMPs, which are dark matter candidates with masses at the keV-scale, with the XENON100 experiment. XENON100 is a dual-phase xenon time projection chamber operated at the Laboratori Nazionali del Gran Sasso. A profile likelihood analysis of data with an exposure of 224.6 live days $\times$ 34\,kg showed no evidence for a signal above the expected background. We thus obtain new and stringent upper limits in the $(8-125)$\,keV/c$^2$ mass range, excluding couplings to electrons with coupling constants of $g_{ae} > 3\times10^{-13}$ for pseudo-scalar and $\alpha'/\alpha > 2\times10^{-28}$ for vector super-WIMPs, respectively. These limits are derived under the assumption that super-WIMPs constitute all of the dark matter in our galaxy., Comment: 6 pages, 6 figures

The XENON1T experiment at the Laboratori Nazionali del Gran Sasso (LNGS) is the first WIMP dark matter detector operating with a liquid xenon target mass above the ton-scale. Out of its 3.2t liquid xenon inventory, 2.0t constitute the active target of the dual-phase time projection chamber. The scintillation and ionization signals from particle interactions are detected with low-background photomultipliers. This article describes the XENON1T instrument and its subsystems as well as strategies to achieve an unprecedented low background level. First results on the detector response and the performance of the subsystems are also presented., Comment: 22 pages, 25 figures

In this paper, we describe the XENON100 data analyses used to assess the target-intrinsic background sources radon ($^{222}$Rn), thoron ($^{220}$Rn) and krypton ($^{85}$Kr). We detail the event selections of high-energy alpha particles and decay-specific delayed coincidences. We derive distributions of the individual radionuclides inside the detector and quantify their abundances during the main three science runs of the experiment over a period of $\sim$ 4 years, from January 2010 to January 2014. We compare our results to external measurements of radon emanation and krypton concentrations where we find good agreement. We report an observed reduction in concentrations of radon daughters that we attribute to the plating-out of charged ions on the negatively biased cathode., Comment: v1: 11 pages, 5 figures, 4 tables; v2: 12 pages, 5 figures, 4 tables, article updated after referee process

We report the first dark matter search results from XENON1T, a $\sim$2000-kg-target-mass dual-phase (liquid-gas) xenon time projection chamber in operation at the Laboratori Nazionali del Gran Sasso in Italy and the first ton-scale detector of this kind. The blinded search used 34.2 live days of data acquired between November 2016 and January 2017. Inside the (1042$\pm$12) kg fiducial mass and in the [5, 40] $\mathrm{keV}_{\mathrm{nr}}$ energy range of interest for WIMP dark matter searches, the electronic recoil background was $(1.93 \pm 0.25) \times 10^{-4}$ events/(kg $\times$ day $\times \mathrm{keV}_{\mathrm{ee}}$), the lowest ever achieved in a dark matter detector. A profile likelihood analysis shows that the data is consistent with the background-only hypothesis. We derive the most stringent exclusion limits on the spin-independent WIMP-nucleon interaction cross section for WIMP masses above 10 GeV/c${}^2$, with a minimum of 7.7 $\times 10^{-47}$ cm${}^2$ for 35-GeV/c${}^2$ WIMPs at 90% confidence level., Comment: 6 pages, 4 figures, v2 data upload Fig 2/4, v3 as accepted by PRL with textual clarifications, v4 data upload Fig 1

We report on WIMP search results in the XENON100 detector using a non-relativistic effective field theory approach. The data from science run II (34 kg $\times$ 224.6 live days) was re-analyzed, with an increased recoil energy interval compared to previous analyses, ranging from $(6.6 - 240)~\mathrm{keV_\mathrm{nr}}$. The data is found to be compatible with the background-only hypothesis. We present 90% confidence level exclusion limits on the coupling constants of WIMP-nucleon effective operators using a binned profile likelihood method. We also consider the case of inelastic WIMP scattering, where incident WIMPs may up-scatter to a higher mass state, and set exclusion limits on this model as well.

The XENON1T dark matter experiment aims to detect Weakly Interacting Massive Particles (WIMPs) through low-energy interactions with xenon atoms. To detect such a rare event necessitates the use of radiopure materials to minimize the number of background events within the expected WIMP signal region. In this paper we report the results of an extensive material radioassay campaign for the XENON1T experiment. Using gamma-ray spectroscopy and mass spectrometry techniques, systematic measurements of trace radioactive impurities in over one hundred samples within a wide range of materials were performed. The measured activities allowed for stringent selection and placement of materials during the detector construction phase and provided the input for XENON1T detection sensitivity estimates through Monte Carlo simulations.

We present the first constraints on the spin-dependent, inelastic scattering cross section of Weakly Interacting Massive Particles (WIMPs) on nucleons from XENON100 data with an exposure of 7.64$\times$10$^3$\,kg\,day. XENON100 is a dual-phase xenon time projection chamber with 62\,kg of active mass, operated at the Laboratori Nazionali del Gran Sasso (LNGS) in Italy and designed to search for nuclear recoils from WIMP-nucleus interactions. Here we explore inelastic scattering, where a transition to a low-lying excited nuclear state of $^{129}$Xe is induced. The experimental signature is a nuclear recoil observed together with the prompt de-excitation photon. We see no evidence for such inelastic WIMP-$^{129}$Xe interactions. A profile likelihood analysis allows us to set a 90\% C.L. upper limit on the inelastic, spin-dependent WIMP-nucleon cross section of $3.3 \times 10^{-38}$\,cm$^{2}$ at 100\,GeV/c$^2$. This is the most constraining result to date, and sets the pathway for an analysis of this interaction channel in upcoming, larger dual-phase xenon detectors.

We present the first search for dark matter-induced delayed coincidence signals in a dual-phase xenon time projection chamber, using the 224.6 live days of the XENON100 science run II. This very distinct signature is predicted in the framework of magnetic inelastic dark matter which has been proposed to reconcile the modulation signal reported by the DAMA/LIBRA collaboration with the null results from other direct detection experiments. No candidate event has been found in the region of interest and upper limits on the WIMP's magnetic dipole moment are derived. The scenarios proposed to explain the DAMA/LIBRA modulation signal by magnetic inelastic dark matter interactions of WIMPs with masses of 58.0 GeV/c$^2$ and 122.7 GeV/c$^2$ are excluded at 3.3 $\sigma$ and 9.3 $\sigma$, respectively., Comment: 10 pages, 6 figures

We describe the purification of xenon from traces of the radioactive noble gas radon using a cryogenic distillation column. The distillation column is integrated into the gas purification loop of the XENON100 detector for online radon removal. This enabled us to significantly reduce the constant $^{222}$Rn background originating from radon emanation. After inserting an auxiliary $^{222}$Rn emanation source in the gas loop, we determined a radon reduction factor of R > 27 (95% C.L.) for the distillation column by monitoring the $^{222}$Rn activity concentration inside the XENON100 detector.

We report on a search for electronic recoil event rate modulation signatures in the XENON100 data accumulated over a period of 4 years, from January 2010 to January 2014. A profile likelihood method, which incorporates the stability of the XENON100 detector and the known electronic recoil background model, is used to quantify the significance of periodicity in the time distribution of events. There is a weak modulation signature at a period of $431^{+16}_{-14}$ days in the low energy region of $(2.0-5.8)$ keV in the single scatter event sample, with a global significance of $1.9\,\sigma$, however no other more significant modulation is observed. The expected annual modulation of a dark matter signal is not compatible with this result. Single scatter events in the low energy region are thus used to exclude the DAMA/LIBRA annual modulation as being due to dark matter electron interactions via axial vector coupling at $5.7\,\sigma$., Comment: 6 pages, 5 figures

The XENON1T experiment aims for the direct detection of dark matter in a cryostat filled with 3.3 tons of liquid xenon. In order to achieve the desired sensitivity, the background induced by radioactive decays inside the detector has to be sufficiently low. One major contributor is the $\beta$-emitter $^{85}$Kr which is an intrinsic contamination of the xenon. For the XENON1T experiment a concentration of natural krypton in xenon $\rm{^{nat}}$Kr/Xe < 200 ppq (parts per quadrillion, 1 ppq = 10$^{-15}$ mol/mol) is required. In this work, the design of a novel cryogenic distillation column using the common McCabe-Thiele approach is described. The system demonstrated a krypton reduction factor of 6.4$\cdot$10$^5$ with thermodynamic stability at process speeds above 3 kg/h. The resulting concentration of $\rm{^{nat}}$Kr/Xe < 26 ppq is the lowest ever achieved, almost one order of magnitude below the requirements for XENON1T and even sufficient for future dark matter experiments using liquid xenon, such as XENONnT and DARWIN.

A Rn-220 source is deployed on the XENON100 dark matter detector in order to address the challenges in calibration of tonne-scale liquid noble element detectors. We show that the Pb-212 beta emission can be used for low-energy electronic recoil calibration in searches for dark matter. The isotope spreads throughout the entire active region of the detector, and its activity naturally decays below background level within a week after the source is closed. We find no increase in the activity of the troublesome Rn-222 background after calibration. Alpha emitters are also distributed throughout the detector and facilitate calibration of its response to Rn-222. Using the delayed coincidence of Rn-220/Po-216, we map for the first time the convective motion of particles in the XENON100 detector. Additionally, we make a competitive measurement of the half-life of Po-212, t = 293.9+-(1.0)+-(0.6) ns.

We report on WIMP search results of the XENON100 experiment, combining three runs summing up to 477 live days from January 2010 to January 2014. Data from the first two runs were already published. A blind analysis was applied to the last run recorded between April 2013 and January 2014 prior to combining the results. The ultra-low electromagnetic background of the experiment, ~$5 \times 10^{-3}$ events/(keV$_{\mathrm{ee}}\times$kg$\times$day) before electronic recoil rejection, together with the increased exposure of 48 kg $\times$ yr improves the sensitivity. A profile likelihood analysis using an energy range of (6.6 - 43.3) keV$_{\mathrm{nr}}$ sets a limit on the elastic, spin-independent WIMP-nucleon scattering cross section for WIMP masses above 8 GeV/$c^2$, with a minimum of 1.1 $\times 10^{-45}$ cm$^2$ at 50 GeV/$c^2$ and 90% confidence level. We also report updated constraints on the elastic, spin-dependent WIMP-nucleon cross sections obtained with the same data. We set upper limits on the WIMP-neutron (proton) cross section with a minimum of 2.0 $\times 10^{-40}$ cm$^2$ (52$\times 10^{-40}$ cm$^2$) at a WIMP mass of 50 GeV/$c^2$, at 90% confidence level., Comment: 12 pages, 13 figures, 2 tables, Limit data points in TeX

Two-neutrino double electron capture is a rare nuclear decay where two electrons are simultaneously captured from the atomic shell. For $^{124}$Xe this process has not yet been observed and its detection would provide a new reference for nuclear matrix element calculations. We have conducted a search for two-neutrino double electron capture from the K-shell of $^{124}$Xe using 7636 kg$\cdot$d of data from the XENON100 dark matter detector. Using a Bayesian analysis we observed no significant excess above background, leading to a lower 90 % credibility limit on the half-life $T_{1/2}>6.5\times10^{20}$ yr. We also evaluated the sensitivity of the XENON1T experiment, which is currently being commissioned, and find a sensitivity of $T_{1/2}>6.1\times10^{22}$ yr after an exposure of 2 t$\cdot$yr., Comment: 6 pages, 4 figures

DARk matter WImp search with liquid xenoN (DARWIN) will be an experiment for the direct detection of dark matter using a multi-ton liquid xenon time projection chamber at its core. Its primary goal will be to explore the experimentally accessible parameter space for Weakly Interacting Massive Particles (WIMPs) in a wide mass-range, until neutrino interactions with the target become an irreducible background. The prompt scintillation light and the charge signals induced by particle interactions in the xenon will be observed by VUV sensitive, ultra-low background photosensors. Besides its excellent sensitivity to WIMPs above a mass of 5 GeV/c2, such a detector with its large mass, low-energy threshold and ultra-low background level will also be sensitive to other rare interactions. It will search for solar axions, galactic axion-like particles and the neutrinoless double-beta decay of 136-Xe, as well as measure the low-energy solar neutrino flux with <1% precision, observe coherent neutrino-nucleus interactions, and detect galactic supernovae. We present the concept of the DARWIN detector and discuss its physics reach, the main sources of backgrounds and the ongoing detector design and R&D efforts., Comment: 36 pages, 11 figures

We perform a low-mass dark matter search using an exposure of 30\,kg$\times$yr with the XENON100 detector. By dropping the requirement of a scintillation signal and using only the ionization signal to determine the interaction energy, we lowered the energy threshold for detection to 0.7\,keV for nuclear recoils. No dark matter detection can be claimed because a complete background model cannot be constructed without a primary scintillation signal. Instead, we compute an upper limit on the WIMP-nucleon scattering cross section under the assumption that every event passing our selection criteria could be a signal event. Using an energy interval from 0.7\,keV to 9.1\,keV, we derive a limit on the spin-independent WIMP-nucleon cross section that excludes WIMPs with a mass of 6\,GeV/$c^2$ above $1.4 \times 10^{-41}$\,cm$^2$ at 90\% confidence level., Comment: 6 pages; 7 figures; PRD. Additional file in source material, s2stot, contains the full list of events passing all selection cuts. Limit data points in TeX; Corrected LUX points used for comparison and respective reference in figure 5

The XENON1T experiment is currently in the commissioning phase at the Laboratori Nazionali del Gran Sasso, Italy. In this article we study the experiment's expected sensitivity to the spin-independent WIMP-nucleon interaction cross section, based on Monte Carlo predictions of the electronic and nuclear recoil backgrounds. The total electronic recoil background in $1$ tonne fiducial volume and ($1$, $12$) keV electronic recoil equivalent energy region, before applying any selection to discriminate between electronic and nuclear recoils, is $(1.80 \pm 0.15) \cdot 10^{-4}$ ($\rm{kg} \cdot day \cdot keV)^{-1}$, mainly due to the decay of $^{222}\rm{Rn}$ daughters inside the xenon target. The nuclear recoil background in the corresponding nuclear recoil equivalent energy region ($4$, $50$) keV, is composed of $(0.6 \pm 0.1)$ ($\rm{t} \cdot y)^{-1}$ from radiogenic neutrons, $(1.8 \pm 0.3) \cdot 10^{-2}$ ($\rm{t} \cdot y)^{-1}$ from coherent scattering of neutrinos, and less than $0.01$ ($\rm{t} \cdot y)^{-1}$ from muon-induced neutrons. The sensitivity of XENON1T is calculated with the Profile Likelihood Ratio method, after converting the deposited energy of electronic and nuclear recoils into the scintillation and ionization signals seen in the detector. We take into account the systematic uncertainties on the photon and electron emission model, and on the estimation of the backgrounds, treated as nuisance parameters. The main contribution comes from the relative scintillation efficiency $\mathcal{L}_\mathrm{eff}$, which affects both the signal from WIMPs and the nuclear recoil backgrounds. After a $2$ y measurement in $1$ t fiducial volume, the sensitivity reaches a minimum cross section of $1.6 \cdot 10^{-47}$ cm$^2$ at m$_\chi$=$50$ GeV/$c^2$., Comment: 36 pages, 18 figures, published by JCAP

Laboratory experiments searching for galactic dark matter particles scattering off nuclei have so far not been able to establish a discovery. We use data from the XENON100 experiment to search for dark matter interacting with electrons. With no evidence for a signal above the low background of our experiment, we exclude a variety of representative dark matter models that would induce electronic recoils. For axial-vector couplings to electrons, we exclude cross-sections above 6x10^(-35) cm^2 for particle masses of m_chi = 2 GeV/c^2. Independent of the dark matter halo, we exclude leptophilic models as explanation for the long-standing DAMA/LIBRA signal, such as couplings to electrons through axial-vector interactions at a 4.4 sigma confidence level, mirror dark matter at 3.6 sigma, and luminous dark matter at 4.6 sigma., Comment: 4 pages, 4 figures, with supporting online material

We have searched for periodic variations of the electronic recoil event rate in the (2-6) keV energy range recorded between February 2011 and March 2012 with the XENON100 detector, adding up to 224.6 live days in total. Following a detailed study to establish the stability of the detector and its background contributions during this run, we performed an un-binned profile likelihood analysis to identify any periodicity up to 500 days. We find a global significance of less than 1 sigma for all periods suggesting no statistically significant modulation in the data. While the local significance for an annual modulation is 2.8 sigma, the analysis of a multiple-scatter control sample and the phase of the modulation disfavor a dark matter interpretation. The DAMA/LIBRA annual modulation interpreted as a dark matter signature with axial-vector coupling of WIMPs to electrons is excluded at 4.8 sigma., Comment: 6 pages, 4 figures

The low-background, VUV-sensitive 3-inch diameter photomultiplier tube R11410 has been developed by Hamamatsu for dark matter direct detection experiments using liquid xenon as the target material. We present the results from the joint effort between the XENON collaboration and the Hamamatsu company to produce a highly radio-pure photosensor (version R11410-21) for the XENON1T dark matter experiment. After introducing the photosensor and its components, we show the methods and results of the radioactive contamination measurements of the individual materials employed in the photomultiplier production. We then discuss the adopted strategies to reduce the radioactivity of the various PMT versions. Finally, we detail the results from screening 216 tubes with ultra-low background germanium detectors, as well as their implications for the expected electronic and nuclear recoil background of the XENON1T experiment., Comment: 10 pages, 5 figures

XENON is a dark matter direct detection project, consisting of a time projection chamber (TPC) filled with liquid xenon as detection medium. The construction of the next generation detector, XENON1T, is presently taking place at the Laboratori Nazionali del Gran Sasso (LNGS) in Italy. It aims at a sensitivity to spin-independent cross sections of $2 \cdot 10^{-47} ~ \mathrm{cm}^{\mathrm{2}}$ for WIMP masses around 50 GeV/c$^{2}$, which requires a background reduction by two orders of magnitude compared to XENON100, the current generation detector. An active system that is able to tag muons and muon-induced backgrounds is critical for this goal. A water Cherenkov detector of $\sim$10 m height and diameter has been therefore developed, equipped with 8 inch photomultipliers and cladded by a reflective foil. We present the design and optimization study for this detector, which has been carried out with a series of Monte Carlo simulations. The muon veto will reach very high detection efficiencies for muons ($>99.5%$) and showers of secondary particles from muon interactions in the rock ($>70%$). Similar efficiencies will be obtained for XENONnT, the upgrade of XENON1T, which will later improve the WIMP sensitivity by another order of magnitude. With the Cherenkov water shield studied here, the background from muon-induced neutrons in XENON1T is negligible.

We present the first results of searches for axions and axion-like-particles with the XENON100 experiment. The axion-electron coupling constant, $g_{Ae}$, has been probed by exploiting the axio-electric effect in liquid xenon. A profile likelihood analysis of 224.6 live days $\times$ 34 kg exposure has shown no evidence for a signal. By rejecting $g_{Ae}$, larger than $7.7 \times 10^{-12}$ (90\% CL) in the solar axion search, we set the best limit to date on this coupling. In the frame of the DFSZ and KSVZ models, we exclude QCD axions heavier than 0.3 eV/c$^2$ and 80 eV/c$^2$, respectively. For axion-like-particles, under the assumption that they constitute the whole abundance of dark matter in our galaxy, we constrain $g_{Ae}$, to be lower than $1 \times 10^{-12}$ (90\% CL) for mass range from 1 to 40 keV/c$^2$, and set the best limit to date as well.