Search

Heatmaps generated on inputs of image classification networks via explainable AI methods like Grad-CAM and LRP have been observed to resemble segmentations of input images in many cases. Consequently, heatmaps have also been leveraged for achieving weakly supervised segmentation with image-level supervision. On the other hand, losses can be imposed on differentiable heatmaps, which has been shown to serve for (1)~improving heatmaps to be more human-interpretable, (2)~regularization of networks towards better generalization, (3)~training diverse ensembles of networks, and (4)~for explicitly ignoring confounding input features. Due to the latter use case, the paradigm of imposing losses on heatmaps is often referred to as "Right for the right reasons". We unify these two lines of research by investigating semi-supervised segmentation as a novel use case for the Right for the Right Reasons paradigm. First, we show formal parallels between differentiable heatmap architectures and standard encoder-decoder architectures for image segmentation. Second, we show that such differentiable heatmap architectures yield competitive results when trained with standard segmentation losses. Third, we show that such architectures allow for training with weak supervision in the form of image-level labels and small numbers of pixel-level labels, outperforming comparable encoder-decoder models. Code is available at \url{https://github.com/Kainmueller-Lab/TW-autoencoder}.

A crucial aspect of understanding the complex nature of Deep Neural Networks (DNNs) is the ability to explain learned concepts within their latent representations. While various methods exist to connect neurons to textual descriptions of human-understandable concepts, evaluating the quality of these explanation methods presents a major challenge in the field due to a lack of unified, general-purpose quantitative evaluation. In this work, we introduce CoSy (Concept Synthesis) -- a novel, architecture-agnostic framework to evaluate the quality of textual explanations for latent neurons. Given textual explanations, our proposed framework leverages a generative model conditioned on textual input to create data points representing the textual explanation. Then, the neuron's response to these explanation data points is compared with the response to control data points, providing a quality estimate of the given explanation. We ensure the reliability of our proposed framework in a series of meta-evaluation experiments and demonstrate practical value through insights from benchmarking various concept-based textual explanation methods for Computer Vision tasks, showing that tested explanation methods significantly differ in quality., Comment: 10 pages, 5 figures

The Model Parameter Randomisation Test (MPRT) is highly recognised in the eXplainable Artificial Intelligence (XAI) community due to its fundamental evaluative criterion: explanations should be sensitive to the parameters of the model they seek to explain. However, recent studies have raised several methodological concerns for the empirical interpretation of MPRT. In response, we propose two modifications to the original test: Smooth MPRT and Efficient MPRT. The former reduces the impact of noise on evaluation outcomes via sampling, while the latter avoids the need for biased similarity measurements by re-interpreting the test through the increase in explanation complexity after full model randomisation. Our experiments show that these modifications enhance the metric reliability, facilitating a more trustworthy deployment of explanation methods., Comment: arXiv admin note: text overlap with arXiv:2401.06465

We revisit the stability of the Standard Model vacuum, and investigate its quantum effective potential using the highest available orders in perturbation theory and the most accurate determination of input parameters to date. We observe that the stability of the electroweak vacuum centrally depends on the values of the top mass and the strong coupling constant. We estimate that reducing their uncertainties by a factor of two to three is sufficient to establish or refute SM vacuum stability at the $5\sigma$ level. We further investigate vacuum stability for a variety of singlet scalar field extensions with and without flavor using the Higgs portal mechanism. We identify the BSM parameter spaces for stability and find sizable room for new physics. We further study the phenomenology of Planck-safe models at colliders, and determine the impact on the Higgs trilinear, the Higgs-to-electroweak-boson, and the Higgs quartic couplings, some of which can be significant. The former two can be probed at the HL-LHC, the latter requires a future collider with sufficient energy and precision such as the FCC-hh., Comment: v2: Numerical typo in SM input fixed and Tab. 1, Figs. 1, 2 and related text corrected, clarifications to App A. added

The Model Parameter Randomisation Test (MPRT) is widely acknowledged in the eXplainable Artificial Intelligence (XAI) community for its well-motivated evaluative principle: that the explanation function should be sensitive to changes in the parameters of the model function. However, recent works have identified several methodological caveats for the empirical interpretation of MPRT. To address these caveats, we introduce two adaptations to the original MPRT -- Smooth MPRT and Efficient MPRT, where the former minimises the impact that noise has on the evaluation results through sampling and the latter circumvents the need for biased similarity measurements by re-interpreting the test through the explanation's rise in complexity, after full parameter randomisation. Our experimental results demonstrate that these proposed variants lead to improved metric reliability, thus enabling a more trustworthy application of XAI methods., Comment: 19 pages, 12 figures, NeurIPS XAIA 2023

Deep Neural Networks (DNNs) are capable of learning complex and versatile representations, however, the semantic nature of the learned concepts remains unknown. A common method used to explain the concepts learned by DNNs is Feature Visualization (FV), which generates a synthetic input signal that maximally activates a particular neuron in the network. In this paper, we investigate the vulnerability of this approach to adversarial model manipulations and introduce a novel method for manipulating FV without significantly impacting the model's decision-making process. The key distinction of our proposed approach is that it does not alter the model architecture. We evaluate the effectiveness of our method on several neural network models and demonstrate its capabilities to hide the functionality of arbitrarily chosen neurons by masking the original explanations of neurons with chosen target explanations during model auditing.

Grasslands are known for their high biodiversity and ability to provide multiple ecosystem services. Challenges in automating the identification of indicator plants are key obstacles to large-scale grassland monitoring. These challenges stem from the scarcity of extensive datasets, the distributional shifts between generic and grassland-specific datasets, and the inherent opacity of deep learning models. This paper delves into the latter two challenges, with a specific focus on transfer learning and eXplainable Artificial Intelligence (XAI) approaches to grassland monitoring, highlighting the novelty of XAI in this domain. We analyze various transfer learning methods to bridge the distributional gaps between generic and grassland-specific datasets. Additionally, we showcase how explainable AI techniques can unveil the model's domain adaptation capabilities, employing quantitative assessments to evaluate the model's proficiency in accurately centering relevant input features around the object of interest. This research contributes valuable insights for enhancing model performance through transfer learning and measuring domain adaptability with explainable AI, showing significant promise for broader applications within the agricultural community.

Explainable AI (XAI) has unfolded in two distinct research directions with, on the one hand, post-hoc methods that explain the predictions of a pre-trained black-box model and, on the other hand, self-explainable models (SEMs) which are trained directly to provide explanations alongside their predictions. While the latter is preferred in safety-critical scenarios, post-hoc approaches have received the majority of attention until now, owing to their simplicity and ability to explain base models without retraining. Current SEMs, instead, require complex architectures and heavily regularized loss functions, thus necessitating specific and costly training. To address this shortcoming and facilitate wider use of SEMs, we propose a simple yet efficient universal method called KMEx (K-Means Explainer), which can convert any existing pre-trained model into a prototypical SEM. The motivation behind KMEx is to enhance transparency in deep learning-based decision-making via class-prototype-based explanations that are diverse and trustworthy without retraining the base model. We compare models obtained from KMEx to state-of-the-art SEMs using an extensive qualitative evaluation to highlight the strengths and weaknesses of each model, further paving the way toward a more reliable and objective evaluation of SEMs (The code is available at https://github.com/SrishtiGautam/KMEx).

Deep Neural Networks (DNNs) demonstrate remarkable capabilities in learning complex hierarchical data representations, but the nature of these representations remains largely unknown. Existing global explainability methods, such as Network Dissection, face limitations such as reliance on segmentation masks, lack of statistical significance testing, and high computational demands. We propose Inverse Recognition (INVERT), a scalable approach for connecting learned representations with human-understandable concepts by leveraging their capacity to discriminate between these concepts. In contrast to prior work, INVERT is capable of handling diverse types of neurons, exhibits less computational complexity, and does not rely on the availability of segmentation masks. Moreover, INVERT provides an interpretable metric assessing the alignment between the representation and its corresponding explanation and delivering a measure of statistical significance. We demonstrate the applicability of INVERT in various scenarios, including the identification of representations affected by spurious correlations, and the interpretation of the hierarchical structure of decision-making within the models., Comment: 25 pages, 16 figures

Here we report the performance of a small scale 4 K pulse tube cryocooler operating with a low input power reaching a minimum temperature of 2.2 K, as well as a cooling capacity of over 240 mW at 4.2 K. The compressor is air cooled and can be supplied by single phase power sockets. With an input power of about 1.3 kW the coefficient of performance reaches values of up to 185 mW/kW, which is among the highest currently reported values for small to medium power pulse tubes. The combination of an oil-free Helium compressor and low maintenance pulse tube cryocooler provides a unique miniaturized, energy efficient and mobile cooling tool for applications at 4 K and below., Comment: 6 pages, 3 figures, conference proceedings of 25th joint CEC-ICMC, July 2023 in Honolulu, Hawaii, USA

Social tipping points are promising levers to achieve net-zero greenhouse gas emission targets. They describe how social, political, economic or technological systems can move rapidly into a new state if cascading positive feedback mechanisms are triggered. Analysing the potential of social tipping for rapid decarbonization requires considering the inherent complexity of social systems. Here, we identify that existing scientific literature is inclined to a narrative-based account of social tipping, lacks a broad empirical framework and a multi-systems view. We subsequently outline a dynamic systems approach that entails (i) a systems outlook involving interconnected feedback mechanisms alongside cross-system and cross-scale interactions, and including a socioeconomic and environmental injustice perspective (ii) directed data collection efforts to provide empirical evidence for and monitor social tipping dynamics, (iii) global, integrated, descriptive modelling to project future dynamics and provide ex-ante evidence for interventions. Research on social tipping must be accordingly solidified for climate policy relevance.

Inclusive e$^+$e$^-$ production has been studied with HADES in $\pi^-$ + p, $\pi^-$ + C and $\pi^- + \mathrm{CH}_2$ reactions, using the GSI pion beam at $\sqrt{s_{\pi p}}$ = 1.49 GeV. Invariant mass and transverse momentum distributions have been measured and reveal contributions from Dalitz decays of $\pi^0$, $\eta$ mesons and baryon resonances. The transverse momentum distributions are very sensitive to the underlying kinematics of the various processes. The baryon contribution exhibits a deviation up to a factor seven from the QED reference expected for the dielectron decay of a hypothetical point-like baryon with the production cross section constrained from the inverse $\gamma$ n$\rightarrow \pi^-$ p reaction. The enhancement is attributed to a strong four-momentum squared dependence of the time-like electromagnetic transition form factors as suggested by Vector Meson Dominance (VMD). Two versions of the VMD, that differ in the photon-baryon coupling, have been applied in simulations and compared to data. VMD1 (or two-component VMD) assumes a coupling via the $\rho$ meson and a direct coupling of the photon, while in VMD2 (or strict VMD) the coupling is only mediated via the $\rho$ meson. The VMD2 model, frequently used in transport calculations for dilepton decays, is found to overestimate the measured dielectron yields, while a good description of the data can be obtained with the VMD1 model assuming no phase difference between the two amplitudes. Similar descriptions have also been obtained using a time-like baryon transition form factor model where the pion cloud plays the major role., Comment: (HADES collaboration)

The recent LHCb determination of the direct CP asymmetries in the decays $D^0 \to K^+ K^-, \pi^+ \pi^-$ hints at a sizeable breaking of two approximate symmetries of the SM: CP and U-spin. We aim at explaining the data with BSM physics and use the framework of flavorful $Z^\prime$ models. Interestingly, experimental and theoretical constraints very much narrow down the shape of viable models: Viable, anomaly-free models are electron- and muon-phobic and feature a light $Z^\prime$ of 10-20 GeV coupling only to right-handed fermions. The $Z^\prime$ can be searched for in low mass dijets or at the LHC as well as dark photon searches. A light $Z^\prime$ of $\sim$ 3 GeV or $\sim$ 5-7 GeV can moreover resolve the longstanding discrepancy in the $J/\psi, \psi^\prime$ branching ratios with pion form factors from fits to $e^+ e^- \to \pi^+ \pi^-$ data, and simultaneously explain the charm CP asymmetries. Smoking gun signatures for this scenario are $\Upsilon$ and charmonium decays into pions, taus or invisbles., Comment: 2 pages, 4 figures

The classification of gigapixel histopathology images with deep multiple instance learning models has become a critical task in digital pathology and precision medicine. In this work, we propose a Transformer-based multiple instance learning approach that replaces the traditional learned attention mechanism with a regional, Vision Transformer inspired self-attention mechanism. We present a method that fuses regional patch information to derive slide-level predictions and show how this regional aggregation can be stacked to hierarchically process features on different distance levels. To increase predictive accuracy, especially for datasets with small, local morphological features, we introduce a method to focus the image processing on high attention regions during inference. Our approach is able to significantly improve performance over the baseline on two histopathology datasets and points towards promising directions for further research., Comment: 8 pages, LaTeX; header update after published, fixed typos

Autonomous flying robots, such as multirotors, often rely on deep learning models that make predictions based on a camera image, e.g. for pose estimation. These models can predict surprising results if applied to input images outside the training domain. This fault can be exploited by adversarial attacks, for example, by computing small images, so-called adversarial patches, that can be placed in the environment to manipulate the neural network's prediction. We introduce flying adversarial patches, where multiple images are mounted on at least one other flying robot and therefore can be placed anywhere in the field of view of a victim multirotor. By introducing the attacker robots, the system is extended to an adversarial multi-robot system. For an effective attack, we compare three methods that simultaneously optimize multiple adversarial patches and their position in the input image. We show that our methods scale well with the number of adversarial patches. Moreover, we demonstrate physical flights with two robots, where we employ a novel attack policy that uses the computed adversarial patches to kidnap a robot that was supposed to follow a human., Comment: Accepted at MRS 2023, 7 pages, 5 figures. arXiv admin note: substantial text overlap with arXiv:2305.12859

We explain why vector-like fermions are natural candidates to lift the Standard Model vacuum instability. Results are further discussed from the viewpoint of criticality. Several models allow for vector-like quarks and leptons in the TeV-range which can be searched for at the LHC., Comment: Proceedings Moriond EW 2023, 6 pages, 5 figures

Autonomous flying robots, e.g. multirotors, often rely on a neural network that makes predictions based on a camera image. These deep learning (DL) models can compute surprising results if applied to input images outside the training domain. Adversarial attacks exploit this fault, for example, by computing small images, so-called adversarial patches, that can be placed in the environment to manipulate the neural network's prediction. We introduce flying adversarial patches, where an image is mounted on another flying robot and therefore can be placed anywhere in the field of view of a victim multirotor. For an effective attack, we compare three methods that simultaneously optimize the adversarial patch and its position in the input image. We perform an empirical validation on a publicly available DL model and dataset for autonomous multirotors. Ultimately, our attacking multirotor would be able to gain full control over the motions of the victim multirotor., Comment: 6 pages, 5 figures, Workshop on Multi-Robot Learning, International Conference on Robotics and Automation (ICRA)

The utilization of pre-trained networks, especially those trained on ImageNet, has become a common practice in Computer Vision. However, prior research has indicated that a significant number of images in the ImageNet dataset contain watermarks, making pre-trained networks susceptible to learning artifacts such as watermark patterns within their latent spaces. In this paper, we aim to assess the extent to which popular pre-trained architectures display such behavior and to determine which classes are most affected. Additionally, we examine the impact of watermarks on the extracted features. Contrary to the popular belief that the Chinese logographic watermarks impact the "carton" class only, our analysis reveals that a variety of ImageNet classes, such as "monitor", "broom", "apron" and "safe" rely on spurious correlations. Finally, we propose a simple approach to mitigate this issue in fine-tuned networks by ignoring the encodings from the feature-extractor layer of ImageNet pre-trained networks that are most susceptible to watermark imprints., Comment: 5 pages, 4 figures, Accepted to the ICLR 2023 TrustML-(un)Limited workshop

Explainable artificial intelligence (XAI) methods shed light on the predictions of machine learning algorithms. Several different approaches exist and have already been applied in climate science. However, usually missing ground truth explanations complicate their evaluation and comparison, subsequently impeding the choice of the XAI method. Therefore, in this work, we introduce XAI evaluation in the climate context and discuss different desired explanation properties, namely robustness, faithfulness, randomization, complexity, and localization. To this end, we chose previous work as a case study where the decade of annual-mean temperature maps is predicted. After training both a multi-layer perceptron (MLP) and a convolutional neural network (CNN), multiple XAI methods are applied and their skill scores in reference to a random uniform explanation are calculated for each property. Independent of the network, we find that XAI methods Integrated Gradients, layer-wise relevance propagation, and input times gradients exhibit considerable robustness, faithfulness, and complexity while sacrificing randomization performance. Sensitivity methods -- gradient, SmoothGrad, NoiseGrad, and FusionGrad, match the robustness skill but sacrifice faithfulness and complexity for randomization skill. We find architecture-dependent performance differences regarding robustness, complexity and localization skills of different XAI methods, highlighting the necessity for research task-specific evaluation. Overall, our work offers an overview of different evaluation properties in the climate science context and shows how to compare and benchmark different explanation methods, assessing their suitability based on strengths and weaknesses, for the specific research problem at hand. By that, we aim to support climate researchers in the selection of a suitable XAI method., Comment: 19 pages, 10 figure, accepted at AIES journal by AMS

One of the unsolved challenges in the field of Explainable AI (XAI) is determining how to most reliably estimate the quality of an explanation method in the absence of ground truth explanation labels. Resolving this issue is of utmost importance as the evaluation outcomes generated by competing evaluation methods (or ''quality estimators''), which aim at measuring the same property of an explanation method, frequently present conflicting rankings. Such disagreements can be challenging for practitioners to interpret, thereby complicating their ability to select the best-performing explanation method. We address this problem through a meta-evaluation of different quality estimators in XAI, which we define as ''the process of evaluating the evaluation method''. Our novel framework, MetaQuantus, analyses two complementary performance characteristics of a quality estimator: its resilience to noise and reactivity to randomness, thus circumventing the need for ground truth labels. We demonstrate the effectiveness of our framework through a series of experiments, targeting various open questions in XAI such as the selection and hyperparameter optimisation of quality estimators. Our work is released under an open-source license (https://github.com/annahedstroem/MetaQuantus) to serve as a development tool for XAI- and Machine Learning (ML) practitioners to verify and benchmark newly constructed quality estimators in a given explainability context. With this work, we provide the community with clear and theoretically-grounded guidance for identifying reliable evaluation methods, thus facilitating reproducibility in the field., Comment: 35 pages, 15 figures, 5 tables

The production of $\Sigma^0$ hyperons in proton proton collisions at a beam kinetic energy of 3.5 GeV impinging on a liquid hydrogen target was investigated using data collected with the HADES setup. The total production cross section is found to be $\mathrm{\sigma (pK^{+}\Sigma^{0}) [\mu b] = 17.7 \pm 1.7 (stat) \pm 1.6 (syst)}$. Differential cross section distributions of the exclusive channel $\mathrm{pp \rightarrow pK^{+}\Sigma^{0}}$ were analyzed in the center-of-mass, Gottfried-Jackson and helicity reference frames for the first time at the excess energy of 556 MeV. The data support the interplay between pion and kaon exchange mechanisms and clearly demonstrate the contribution of interfering nucleon resonances decaying to $\mathrm{K^{+}\Sigma^{0}}$. The Bonn-Gatchina partial wave analysis was employed to analyse the data. Due to the limited statistics, it was not possible to obtain an unambiguous determination of the relative contribution of intermediate nucleon resonances to the final state. However nucleon resonances with masses around 1.710 $\mathrm{GeV/c^{2}}$ ($\mathrm{N^{*}(1710)}$) and 1.900 $\mathrm{GeV/c^{2}}$ ($\mathrm{N^{*}(1900)}$ or $\mathrm{\Delta^{*}(1900)}$) are preferred by the fit.

The double differential production cross sections, $d^2\sigma/d\Omega dE$, for hydrogen isotopes and charged pions in the reaction of p + Nb at 3.5 GeV proton beam energy have been measured by the High Acceptance DiElectron Spectrometer (HADES). Thanks to the high acceptance of HADES at forward emission angles and usage of its magnetic field, the measured energy range of hydrogen isotopes could be significantly extended in comparison to the relatively scarce experimental data available in the literature. The data provide information about the development of the intranuclear cascade in the proton-nucleus collisions. They can as well be utilized to study the rate of energy/momentum dissipation in the nuclear systems and the mechanism of elementary and composite particle production in excited nuclear matter at normal density. Data of this type are important also for technological and medical applications. Our results are compared to models developed to describe the processes relevant to nuclear spallation (INCL++) or oriented to probe either the elementary hadronic processes in nuclear matter or the behavior of compressed nuclear matter (GiBUU).

Hadron production ($\pi^\pm$, proton, $\Lambda$, $K_S^0$, $K^\pm$) in $\pi^- + \mathrm{C}$ and $\pi^- + \mathrm{W}$ collisions is investigated at an incident pion beam momentum of $1.7~\mathrm{GeV}/c$. This comprehensive set of data measured with HADES at SIS18/GSI significantly extends the existing world data on hadron production in pion induced reactions and provides a new reference for models that are commonly used for the interpretation of heavy-ion collisions. The measured inclusive differential production cross-sections are compared with state-of-the-art transport model (GiBUU, SMASH) calculations. The (semi-) exclusive channel $\pi^- + A \rightarrow \Lambda + K_S^0 +X$, in which the kinematics of the strange hadrons are correlated, is also investigated and compared to a model calculation. Agreement and remaining tensions between data and the current version of the considered transport models are discussed.

The recent measurement of the CP-asymmetry in the decay $D \to K^+ K^-$ by LHCb, combined with $\Delta A_{\text{CP}}$, evidences a sizable CP-asymmetry in $D \to \pi^+ \pi^-$ decays, which requires a dynamical enhancement of standard model higher-order contributions over tree-level ones by a factor of two. The data furthermore imply huge U-spin breaking, about 4-5 times larger than the nominal standard model one of $\lesssim 30 \%$ in charm. Enhanced breakdown of the two approximate symmetries points to models that violate U-spin and CP and disfavors flavor singlet contributions such as chromomagnetic dipole operators as explanations of the data. We analyze the reach of flavorful $Z^\prime$ models for charm CP-asymmetries. Models feature explicit U-spin and isospin breaking, allowing for correlations with $D \to \pi^0 \pi^0$ and $D^+ \to \pi^+ \pi^0$ decays with corresponding CP-asymmetries at a similar level and sign as $D \to \pi^+ \pi^-$, about $ {\cal{O}}(1-2) \cdot 10^{-3}$. Experimental and theoretical constraints narrow down the shape of viable models: anomaly-free models are leptophobic -- or at least electro- and muo-phobic -- with light $Z^\prime$ below ${\cal{O}}(20)$ GeV, and can be searched for in low mass dijets at the LHC, $\Upsilon$ and charmonium decays, and dark photon signatures. A $Z^\prime$ around $\sim 3$ GeV or $\sim (5-7)$ GeV can relieve the tensions in the $J/\psi \to \pi^+ \pi^-$ and $\psi^\prime \to \pi^+ \pi^-$ branching ratios with pion form factors from fits to Babar and JLab data, and simultaneously explain the charm CP asymmetries. Models also feature sizable branching ratios into light right-handed neutrinos or vector-like dark fermions, which can be searched for in $e^+ e^- \to$~hadrons + invisibles at Belle II and BESIII. Due to the low new physics scale dark fermions may induce an early Landau pole which requires UV-completion near the TeV-scale., Comment: 13 pages, 5 figures. v2: Fig.1 and discussion on U-Spin improved, minor corrections and references added, v3: match published version

The need for interpretable models has fostered the development of self-explainable classifiers. Prior approaches are either based on multi-stage optimization schemes, impacting the predictive performance of the model, or produce explanations that are not transparent, trustworthy or do not capture the diversity of the data. To address these shortcomings, we propose ProtoVAE, a variational autoencoder-based framework that learns class-specific prototypes in an end-to-end manner and enforces trustworthiness and diversity by regularizing the representation space and introducing an orthonormality constraint. Finally, the model is designed to be transparent by directly incorporating the prototypes into the decision process. Extensive comparisons with previous self-explainable approaches demonstrate the superiority of ProtoVAE, highlighting its ability to generate trustworthy and diverse explanations, while not degrading predictive performance.

High precision measurements of flow coefficients $v_{n}$ ($n = 1 - 4$) for protons, deuterons and tritons relative to the first-order spectator plane have been performed in Au+Au collisions at $\sqrt{s_{NN}} = 2.4$ GeV with the High-Acceptance Di-Electron Spectrometer (HADES) at the SIS18/GSI. Flow coefficients are studied as a function of transverse momentum $p_{t}$ and rapidity $y_{cm}$ over a large region of phase space and for several classes of collision centrality. A clear mass hierarchy is found for the slope of $v_{1}$, $d v_{1}/d y^{\prime}|_{y^{\prime} = 0}$ where $y^{\prime}$ is the scaled rapidity, and for $v_{2}$ at mid-rapidity. Scaling with the number of nucleons is observed for the $p_{t}$ dependence of $v_{2}$ and $v_{4}$ at mid-rapidity, which is indicative for nuclear coalescence as the main process responsible for light nuclei formation. $v_{2}$ is found to scale with the initial eccentricity $\langle \epsilon_{2} \rangle$, while $v_{4}$ scales with $\langle \epsilon_{2} \rangle^{2}$ and $\langle \epsilon_{4} \rangle$. The multi-differential high-precision data on $v_{1}$, $v_{2}$, $v_{3}$, and $v_{4}$ provides important constraints on the equation-of-state of compressed baryonic matter., Comment: 21 pages, 18 figures, 4 tables; Figs. 11 updated; The manuscript has associated data in the HEPData repository: https://doi.org/10.17182/hepdata.152804

It is common practice to reuse models initially trained on different data to increase downstream task performance. Especially in the computer vision domain, ImageNet-pretrained weights have been successfully used for various tasks. In this work, we investigate the impact of transfer learning for segmentation problems, being pixel-wise classification problems that can be tackled with encoder-decoder architectures. We find that transfer learning the decoder does not help downstream segmentation tasks, while transfer learning the encoder is truly beneficial. We demonstrate that pretrained weights for a decoder may yield faster convergence, but they do not improve the overall model performance as one can obtain equivalent results with randomly initialized decoders. However, we show that it is more effective to reuse encoder weights trained on a segmentation or reconstruction task than reusing encoder weights trained on classification tasks. This finding implicates that using ImageNet-pretrained encoders for downstream segmentation problems is suboptimal. We also propose a contrastive self-supervised approach with multiple self-reconstruction tasks, which provides encoders that are suitable for transfer learning in segmentation problems in the absence of segmentation labels.

We address the notorious metastability of the standard model (SM) and promote it to a model building task: What are the new ingredients required to stabilize the SM up to the Planck scale without encountering subplanckian Landau poles? Using the SM extended by vector-like fermions (VLFs), we chart out the corresponding landscape of Higgs stability. We find that the gauge portal mechanism, triggered by new SM charge carriers, opens up sizeable room for stability in a minimally invasive manner. We also find models with Higgs criticality, and Yukawa portals opening up at stronger coupling. Several models allow for VLFs in the TeV-range, which can be searched for at the LHC. For nontrivial flavor structure severe flavor-changing neutral current constraints arise which complement those from stability, and push lower fermion masses up to $\mathcal{O}(10^3\,\text{TeV})$., Comment: 18 pages, 16 figures. v2: match published version

The global polarization of {\Lambda} hyperons along the total orbital angular momentum of a relativistic heavy-ion collision is presented based on the high statistics data samples collected in Au+Au collisions at \sqrt{s_{NN}} = 2.4 GeV and Ag+Ag at 2.55 GeV with the High-Acceptance Di-Electron Spectrometer (HADES) at GSI, Darmstadt. This is the first measurement below the strangeness production threshold in nucleon-nucleon collisions. Results are reported as a function of the collision centrality as well as a function of the hyperon transverse momentum (p_T) and rapidity (y_{CM}) for the range of centrality 0--40%. We observe a strong centrality dependence of the polarization with an increasing signal towards peripheral collisions. For mid-central (20--40%) collisions the polarization magnitudes are (%) = 6.8 \pm 1.3 (stat.) \pm 2.1 (syst.) for Au+Au and (%) = 6.2 \pm 0.4 (stat.) \pm 0.6 (syst.) for Ag+Ag, which are the largest values observed so far. This observation thus provides a continuation of the increasing trend previously observed by STAR and contrasts expectations from recent theoretical calculations predicting a maximum in the region of collision energies about 3 GeV. The observed polarization is of a similar magnitude as predicted by 3D fluid dynamics and the UrQMD plus thermal vorticity model and significantly above results from the AMPT model., Comment: 8 pages, 4 figures

Deep Neural Networks (DNNs) excel at learning complex abstractions within their internal representations. However, the concepts they learn remain opaque, a problem that becomes particularly acute when models unintentionally learn spurious correlations. In this work, we present DORA (Data-agnOstic Representation Analysis), the first data-agnostic framework for analyzing the representational space of DNNs. Central to our framework is the proposed Extreme-Activation (EA) distance measure, which assesses similarities between representations by analyzing their activation patterns on data points that cause the highest level of activation. As spurious correlations often manifest in features of data that are anomalous to the desired task, such as watermarks or artifacts, we demonstrate that internal representations capable of detecting such artifactual concepts can be found by analyzing relationships within neural representations. We validate the EA metric quantitatively, demonstrating its effectiveness both in controlled scenarios and real-world applications. Finally, we provide practical examples from popular Computer Vision models to illustrate that representations identified as outliers using the EA metric often correspond to undesired and spurious concepts., Comment: 24 pages, 18 figures

First information on the time-like electromagnetic structure of baryons in the second resonance region has been obtained from measurements of dielectron (e+ e-) invariant-mass and angular distributions in the quasi-free reaction $\pi-$ p $\rightarrow$ n e+ e- at $\sqrt{s_{\pi p}}$ = 1.49 GeV with the High Acceptance Di-Electron Spectrometer (HADES) at GSI using the pion beam impinging on a CH$_2$ target. We find a total cross section $\sigma$ = 2.97 $\pm$ 0.07data $\pm$ 0.21acc $\pm$ 0.31Zeff $\mu$b. In complement to the analysis of the inclusive e+ e- channel, this data set provides a crucial test of the description of baryon time-like transitions. Approaches based on a Vector Meson Dominance amplitude containing direct photon and vector meson ($\rho$) couplings to the baryon provide a satisfactory agreement with the data. A good description is also obtained by electromagnetic time-like baryon transition form factors in a covariant spectator-quark model, pointing to the dominance of meson-cloud effects. The dielectron angular distributions exhibit the contributions of virtual photons ($\gamma^*$) with longitudinal polarization, in contrast to real photons. The virtual photon angular dependence supports the dominance of J=3/2, I=1/2 contributions observed in both the $\gamma^*$n and the $\pi \pi$n channels., Comment: revised version, 7 pages, 3 figures