Search

It is well known that silicon photomultipliers (SiPMs) are prone to radiation damage. With the increasing popularity of SiPMs among new spaceborne missions, especially on CubeSats, it is of paramount importance to characterize their performance in space environment. In this work, we report the in-orbit ageing of SiPM arrays, so-called multi-pixel photon counters (MPPCs), using measurements acquired by the GRBAlpha and VZLUSAT-2 CubeSats at low Earth orbit (LEO) spanning over three years, which in duration is unique. GRBAlpha is a 1U CubeSat launched on March 22, 2021, to a 550 km altitude sun-synchronous polar orbit (SSO) carrying on board a gamma-ray detector based on CsI(Tl) scintillator readout by eight MPPCs and regularly detecting gamma-ray transients such as gamma-ray bursts and solar flares in the energy range of ~30-900 keV. VZLUSAT-2 is a 3U CubeSat launched on January 13, 2022 also to a 550 km altitude SSO carrying on board, among other payloads, two gamma-ray detectors similar to the one on GRBAlpha. We have flight-proven the Hamamatsu MPPCs S13360-3050 PE and demonstrated that MPPCs, shielded by 2.5 mm of PbSb alloy, can be used in an LEO environment on a scientific mission lasting beyond three years. This manifests the potential of MPPCs being employed in future satellites., Comment: Submitted to Nuclear Instruments and Methods in Physics Research Section A, 13 pages, 14 figures

The world is abundant with diverse materials, each possessing unique surface appearances that play a crucial role in our daily perception and understanding of their properties. Despite advancements in technology enabling the capture and realistic reproduction of material appearances for visualization and quality control, the interoperability of material property information across various measurement representations and software platforms remains a complex challenge. A key to overcoming this challenge lies in the automatic identification of materials' perceptual features, enabling intuitive differentiation of properties stored in disparate material data representations. We reasoned that for many practical purposes, a compact representation of the perceptual appearance is more useful than an exhaustive physical description.This paper introduces a novel approach to material identification by encoding perceptual features obtained from dynamic visual stimuli. We conducted a psychophysical experiment to select and validate 16 particularly significant perceptual attributes obtained from videos of 347 materials. We then gathered attribute ratings from over twenty participants for each material, creating a 'material fingerprint' that encodes the unique perceptual properties of each material. Finally, we trained a multi-layer perceptron model to predict the relationship between statistical and deep learning image features and their corresponding perceptual properties. We demonstrate the model's performance in material retrieval and filtering according to individual attributes. This model represents a significant step towards simplifying the sharing and understanding of material properties in diverse digital environments regardless of their digital representation, enhancing both the accuracy and efficiency of material identification., Comment: 14 pages, 12 figures, 3 tables

Quantum coherence between energy eigenstates of harmonic oscillators is essential for quantum physics. Even the most elementary binary superpositions of the ground and the higher eigenstate are highly required for quantum sensing, thermodynamics, and computing. We derive upper bounds for quantum coherences achieved by classical and Gaussian states and operations and, subsequently, obtain a hierarchy of the thresholds for the off-diagonal elements necessary to reach genuine quantum non-Gaussian coherences. We experimentally demonstrate unambiguous observation of quantum non-Gaussian coherences in mechanical vibrations of a single calcium ion up to the superposition of zero and six phonons. The analysis of the robustness with respect to pure dephasing in a motional Ramsey experiment demonstrates the feasibility of their storage for up to more than 20 ms for superpositions with a large energy difference of participating number states. The presented observations prove that atomic oscillations go deeply into a diverse area of discrete quantum non-Gaussian coherent phenomena critical for their applications.

Inverse protein folding generates valid amino acid sequences that can fold into a desired protein structure, with recent deep-learning advances showing significant potential and competitive performance. However, challenges remain in predicting highly uncertain regions, such as those with loops and disorders. To tackle such low-confidence residue prediction, we propose a \textbf{Ma}sk \textbf{p}rior-guided denoising \textbf{Diff}usion (\textbf{MapDiff}) framework that accurately captures both structural and residue interactions for inverse protein folding. MapDiff is a discrete diffusion probabilistic model that iteratively generates amino acid sequences with reduced noise, conditioned on a given protein backbone. To incorporate structural and residue interactions, we develop a graph-based denoising network with a mask prior pre-training strategy. Moreover, in the generative process, we combine the denoising diffusion implicit model with Monte-Carlo dropout to improve uncertainty estimation. Evaluation on four challenging sequence design benchmarks shows that MapDiff significantly outperforms state-of-the-art methods. Furthermore, the in-silico sequences generated by MapDiff closely resemble the physico-chemical and structural characteristics of native proteins across different protein families and architectures., Comment: 21 pages, 8 figures

This paper proposes a novel trajectory generation method based on Model Predictive Control (MPC) for agile landing of an Unmanned Aerial Vehicle (UAV) onto an Unmanned Surface Vehicle (USV)'s deck in harsh conditions. The trajectory generation exploits the state predictions of the USV to create periodically updated trajectories for a multirotor UAV to precisely land on the deck of a moving USV even in cases where the deck's inclination is continuously changing. We use an MPC-based scheme to create trajectories that consider both the UAV dynamics and the predicted states of the USV up to the first derivative of position and orientation. Compared to existing approaches, our method dynamically modifies the penalization matrices to precisely follow the corresponding states with respect to the flight phase. Especially during the landing maneuver, the UAV synchronizes attitude with the USV's, allowing for fast landing on a tilted deck. Simulations show the method's reliability in various sea conditions up to Rough sea (wave height 4 m), outperforming state-of-the-art methods in landing speed and accuracy, with twice the precision on average. Finally, real-world experiments validate the simulation results, demonstrating robust landings on a moving USV, while all computations are performed in real-time onboard the UAV., Comment: 18 pages, 17 figures, Ocean Engineering

We propose a verified executable Scala backend for ASN1SCC, a compiler for ASN.1/ACN. ASN.1 is a language for describing data structures widely employed in ground and space telecommunications. ACN can be used along ASN.1 to describe complex binary formats and legacy protocols. To avoid error-prone and time-consuming manual writing of serializers, we show how to port an ASN.1/ACN code generator to generate Scala code. We then enhance the generator to emit not only the executable code but also strong enough pre-conditions, post-conditions, and lemmas for inductive proofs. This allowed us to verify the resulting generated annotated code using Stainless, a program verifier for Scala. The properties we prove include the absence of runtime errors, such as out-of-bound accesses or divisions by zero. For the base library, we also prove the invertibility of the decoding and encoding functions, showing that decoding yields the encoded value back. Furthermore, our system automatically inserts invertibility proofs for arbitrary records in the generated code, proving over 300'000 verification conditions. We establish key steps towards such proofs for sums and arrays as well.

Model checking of strategic abilities for agents with memory is a notoriously hard problem, and very few attempts have been made to tackle it. In this paper, we present two important steps towards this goal. First, we take the partial-order reduction scheme that was recently proved to preserve individual and coalitional abilities of memoryless agents, and show that it also works for agents with memory. Secondly, we take the Knowledge-Based Subset Construction, that was recently studied for synchronous concurrent games, and adapt it to preserve abilities of memoryful agents in asynchronous MAS. On the way, we also propose a new execution semantics for strategies in asynchronous MAS, that combines elements of Concurrent Game Structures and Interleaved Interpreted Systems in a natural and intuitive way.

We give a status report on new developments within the Whizard event generator. Important new features comprise NLO electroweak automation (incl. extension to BSM processes like SMEFT), loop-induced processes and new developments in the UFO interface. We highlight work in progress and further plans, such as the implementation of electroweak PDFs, photon radiation, the exclusive top threshold and features for exotic new physics searches., Comment: Presented at the International Workshop on Future Linear Colliders (LCWS 2024), 8-11 July 2024, contribution accepted by EPJ Web of Conferences

The transcorrelated (TC) method performs a similarity transformation on the electronic Schr\"odinger equation via Jastrow factorization of the wave function. This has demonstrated significant advancements in computational electronic structure theory by improving basis set convergence and compactifying the description of the wave function. In this work, we introduce a new approach that incorporates pseudopotentials (PPs) into the TC framework, significantly accelerating Jastrow factor optimization and reducing computational costs. Our results for ionization potentials, atomization energies, and dissociation curves of first-row atoms and molecules show that PPs provide chemically accurate descriptions across a range of systems and give guidelines for future theory and applications. The new pseudopotential-based TC method opens possibilities for applying TC to more complex and larger systems, such as transition metals and solid-state systems.

Graphs are ubiquitous in real-world applications, ranging from social networks to biological systems, and have inspired the development of Graph Neural Networks (GNNs) for learning expressive representations. While most research has centered on static graphs, many real-world scenarios involve dynamic, temporally evolving graphs, motivating the need for Continuous-Time Dynamic Graph (CTDG) models. This paper provides a comprehensive review of Graph Representation Learning (GRL) on CTDGs with a focus on Self-Supervised Representation Learning (SSRL). We introduce a novel theoretical framework that analyzes the expressivity of CTDG models through an Information-Flow (IF) lens, quantifying their ability to propagate and encode temporal and structural information. Leveraging this framework, we categorize existing CTDG methods based on their suitability for different graph types and application scenarios. Within the same scope, we examine the design of SSRL methods tailored to CTDGs, such as predictive and contrastive approaches, highlighting their potential to mitigate the reliance on labeled data. Empirical evaluations on synthetic and real-world datasets validate our theoretical insights, demonstrating the strengths and limitations of various methods across long-range, bi-partite and community-based graphs. This work offers both a theoretical foundation and practical guidance for selecting and developing CTDG models, advancing the understanding of GRL in dynamic settings., Comment: 12-page main paper + 8-page appendix

Energy extraction from a measured quantum system is a cornerstone of information thermodynamics as illustrated by Maxwell's demon. The nonequilibrium physics of many-particle systems is additionally strongly influenced by quantum statistics. We here report the first experimental realization of a quantum demon in a many-particle photonic setup made of two identical thermal light beams. We show that single-photon measurements combined with feedforward operation may deterministically increase the mean energy of one beam faster than energy fluctuations, thus improving the thermodynamic stability of the device. We moreover demonstrate that bosonic statistics can enhance the energy output above the classical limit, and further analyze the counterintuitive thermodynamics of the demon using an information-theoretic approach. Our results underscore the pivotal role of many-particle statistics for enhanced energy extraction in quantum thermodynamics., Comment: 10 pages, 7 figures

This paper focuses on enhancing the energy efficiency (EE) of a cooperative network featuring a `miniature' unmanned aerial vehicle (UAV) that operates at terahertz (THz) frequencies, utilizing holographic surfaces to improve the network's performance. Unlike traditional reconfigurable intelligent surfaces (RIS) that are typically used as passive relays to adjust signal reflections, this work introduces a novel concept: Energy harvesting (EH) using reconfigurable holographic surfaces (RHS) mounted on the miniature UAV. In this system, a source node facilitates the simultaneous reception of information and energy signals by the UAV, with the harvested energy from the RHS being used by the UAV to transmit data to a specific destination. The EE optimization involves adjusting non-orthogonal multiple access (NOMA) power coefficients and the UAV's flight path, considering the peculiarities of the THz channel. The optimization problem is solved in two steps. Initially, the trajectory is refined using a successive convex approximation (SCA) method, followed by the adjustment of NOMA power coefficients through a quadratic transform technique. The effectiveness of the proposed algorithm is demonstrated through simulations, showing superior results when compared to baseline methods.

Serious games and gamification (SGG) in eHealth have positive health impacts, but a personalized approach is needed due to diverse user contexts. This introduces challenges in achieving personalization in SGG. A literature search on Web of Science and PubMed identified 31 articles: 22 on serious games and 9 on gamification. These strategies are most applied in behavior change and rehabilitation, with machine learning and AI showing promise for personalization. Reusability of personalisation algorithms and domain knowledge is underemphasized, reported in only 10 articles. Future research should standardize personalized SGG development, focusing on component reuse to streamline design and enhance evaluation., Comment: 24 pages, 9 figures

Segmentation of Earth observation (EO) satellite data is critical for natural hazard analysis and disaster response. However, processing EO data at ground stations introduces delays due to data transmission bottlenecks and communication windows. Using segmentation models capable of near-real-time data analysis onboard satellites can therefore improve response times. This study presents a proof-of-concept using MobileSAM, a lightweight, pre-trained segmentation model, onboard Unibap iX10-100 satellite hardware. We demonstrate the segmentation of water bodies from Sentinel-2 satellite imagery and integrate MobileSAM with PASEOS, an open-source Python module that simulates satellite operations. This integration allows us to evaluate MobileSAM's performance under simulated conditions of a satellite constellation. Our research investigates the potential of fine-tuning MobileSAM in a decentralised way onboard multiple satellites in rapid response to a disaster. Our findings show that MobileSAM can be rapidly fine-tuned and benefits from decentralised learning, considering the constraints imposed by the simulated orbital environment. We observe improvements in segmentation performance with minimal training data and fast fine-tuning when satellites frequently communicate model updates. This study contributes to the field of onboard AI by emphasising the benefits of decentralised learning and fine-tuning pre-trained models for rapid response scenarios. Our work builds on recent related research at a critical time; as extreme weather events increase in frequency and magnitude, rapid response with onboard data analysis is essential., Comment: Accepted at the Sixth IEEE International Conference on Image Processing Applications and Systems (IPAS) 2025

Detecting user frustration in modern-day task-oriented dialog (TOD) systems is imperative for maintaining overall user satisfaction, engagement, and retention. However, most recent research is focused on sentiment and emotion detection in academic settings, thus failing to fully encapsulate implications of real-world user data. To mitigate this gap, in this work, we focus on user frustration in a deployed TOD system, assessing the feasibility of out-of-the-box solutions for user frustration detection. Specifically, we compare the performance of our deployed keyword-based approach, open-source approaches to sentiment analysis, dialog breakdown detection methods, and emerging in-context learning LLM-based detection. Our analysis highlights the limitations of open-source methods for real-world frustration detection, while demonstrating the superior performance of the LLM-based approach, achieving a 16\% relative improvement in F1 score on an internal benchmark. Finally, we analyze advantages and limitations of our methods and provide an insight into user frustration detection task for industry practitioners.

Advancing towards high automation and autonomous operations is crucial for the future of inland waterway transport (IWT) systems. These systems necessitate robust and precise onboard sensory technologies that can perceive the environment under all weather conditions, including static features for local positioning techniques such as Simultaneous Localization and Mapping (SLAM). Traditional marine RADAR, mandatory on vessels and operating in the 9300-9500 MHz frequency band, can cover ranges from 15 to 1200 meters but are inadequate for detecting closer objects, making them unsuitable for automated docking maneuvers, lock entry, or bridge undercrossings. This necessitates the development of reliable close-range sensor technology that functions effectively in all weather conditions. In present research works on vessel automation, LiDAR sensors, operating in the nearinfrared range, are used predominantly to detect the immediate surroundings of vessels but suffer significant degradation in poor visibility. Conversely, automotive RADAR sensors, utilizing the 76-81 GHz frequency band, can detect objects from a few centimeters to up to 200 meters, even in adverse conditions. These sensors are commonly used in advanced autonomous road traffic systems and are evaluated in this study for their suitability in inland navigation and maneuvering. This paper discusses a distributed sensor network of four compact automotive frequencymodulated continuous-wave (FMCW) radars mounted on a cabin boat as a test platform. Initial field experiments demonstrate the RADAR network's ability to perceive closerange static environmental features around the boat in inland waters. The paper also provides a comparative analysis of the environmental detection capabilities of automotive RADAR and LiDAR sensors., Comment: 9 pages, 11 figues, conference, included in conference proceedings of DGON POSNAV 2024

Recent advances in AI -- including generative approaches -- have resulted in technology that can support humans in scientific discovery and decision support but may also disrupt democracies and target individuals. The responsible use of AI increasingly shows the need for human-AI teaming, necessitating effective interaction between humans and machines. A crucial yet often overlooked aspect of these interactions is the different ways in which humans and machines generalise. In cognitive science, human generalisation commonly involves abstraction and concept learning. In contrast, AI generalisation encompasses out-of-domain generalisation in machine learning, rule-based reasoning in symbolic AI, and abstraction in neuro-symbolic AI. In this perspective paper, we combine insights from AI and cognitive science to identify key commonalities and differences across three dimensions: notions of generalisation, methods for generalisation, and evaluation of generalisation. We map the different conceptualisations of generalisation in AI and cognitive science along these three dimensions and consider their role in human-AI teaming. This results in interdisciplinary challenges across AI and cognitive science that must be tackled to provide a foundation for effective and cognitively supported alignment in human-AI teaming scenarios.

This article investigates the influence of luck and strategic considerations on performance of teams participating in the M6 investment challenge. We find that there is insufficient evidence to suggest that the extreme Sharpe ratios observed are beyond what one would expect by chance, given the number of teams, and thus not necessarily indicative of the possibility of consistently attaining abnormal returns. Furthermore, we introduce a stylized model of the competition to derive and analyze a portfolio strategy optimized for attaining the top rank. The results demonstrate that the task of achieving the top rank is not necessarily identical to that of attaining the best investment returns in expectation. It is possible to improve one's chances of winning, even without the ability to attain abnormal returns, by choosing portfolio weights adversarially based on the current competition ranking. Empirical analysis of submitted portfolio weights aligns with this finding.

Though large language models (LLMs) have demonstrated exceptional performance across numerous problems, their application to predictive tasks in relational databases remains largely unexplored. In this work, we address the notion that LLMs cannot yield satisfactory results on relational databases due to their interconnected tables, complex relationships, and heterogeneous data types. Using the recently introduced RelBench benchmark, we demonstrate that even a straightforward application of LLMs achieves competitive performance on these tasks. These findings establish LLMs as a promising new baseline for ML on relational databases and encourage further research in this direction.

For the almost complete intersection ideals $(x_1^2, \dots, x_n^2, (x_1 + \cdots + x_n)^k)$, we compute their reduced Gr\"obner basis for any term ordering, revealing a combinatorial structure linked to lattice paths, elementary symmetric polynomials, and Catalan numbers. Using this structure, we classify the weak Lefschetz property for these ideals. Additionally, we provide a new proof of the well-known result that the squarefree algebra satisfies the strong Lefschetz property. Finally, we compute the Betti numbers of the initial ideals and construct a minimal free resolution using a Mayer-Vietoris tree approach., Comment: 33 pages, 3 figures

Conformally invariant scalar equation permits the Robin boundary condition at infinity of asymptotically-AdS spacetimes. We show how the dynamics of conformal cubic scalar field on the Reissner-Nordstr\"om-anti-de Sitter background depend on the black hole size, charge and the choice of the boundary condition. We study the whole range of admissible charges, including the extremal case. In particular, we observe the transition in stability of the field for large black holes at the specific critical value of the charge. Similarities between Reissner-Nordstr\"om and Kerr black hole let us suspect that similar effect may also occur in rotating black holes., Comment: 13 pages, 8 figures

We investigate the temperature, photon and shadow radii, quasinormal modes (QNMs), greybody factors, and emission rates of deformed AdS black holes, focusing on the effects of the deformation parameter $ \alpha $ and control parameter $ \beta $. Increasing $ \alpha $ enhances the oscillation frequency and damping rate of gravitational waves, while $ \beta $ shows non-linear behaviour. Electromagnetic perturbations exhibit similar trends, though with lower frequencies and damping rates. Greybody factors are mainly influenced by multipole moment $ l $ and $ \alpha $, with $ \beta $ having a more subtle effect. These findings provide insights into black hole dynamics, mergers, and gravitational wave emissions., Comment: 17 pages, 16 figures

Safety fine-tuning algorithms are commonly used to fine-tune language models to reduce harmful outputs, but the exact internal mechanisms of how those models achieve this remain unclear. In studying direct preference optimisation (DPO) for toxicity reduction, current explanations claim that DPO works by dampening the most toxic MLP neurons to learn an offset to avert toxic regions in the residual stream. However, by ablating the most toxic neurons and applying activation patching, we find this explanation incomplete. By projecting neuron activation changes onto a toxicity probe, we find that only 31.8\% of toxicity reduction comes from dampened toxic neurons. Instead, DPO reduces toxicity by accumulating effects across multiple neuron groups, both reducing writing in the toxic direction and promoting anti-toxicity in the residual stream. Moreover, DPO gives noisy adjustments to neuron activations, with many neurons actually increasing toxicity. This indicates that DPO is a balancing process between opposing neuron effects to achieve toxicity reduction.

Organic based semiconductor materials offer emerging and sustainable solutions for solar energy conversion technologies and electronics. However, knowledge of their intrinsic (photo)physical properties and light-matter interactions is often limited, especially with respect to the frequency dependent dielectric properties on the relevant timescales of exciton separation and charge generation (fs-ps). By using terahertz-time-domain spectroscopy (THz-TDS), we show that the complex permittivity and THz conductivity of different polymer materials and graphitic carbon nitrides - melon and poly(heptazine imides) (K-PHI) - can be measured directly and accurately in different environmental humidities. Its effects are most strongly observed in the ionic and 2D carbon nitride, K-PHI, where both real permittivity and THz conductivity double from dry to humid conditions (~4-8 and 75-150 S/m, respectively) surpassing the intrinsic dielectric response from water or K-PHI through synergistic effects. Our findings are backed by fs-ps transient absorption spectroscopy (TAS), confirming the impact of humidity on light conversion behaviour on the ps-timescale in K-PHI. We highlight the importance of dielectric property characterization at THz frequencies as critical for understanding photophysical behaviour at exciton or charge separation time scales, especially in the presence of water and hydrated ions, which may also be beneficial for computation, exploring next generation photocatalysts, electronics or ionotronics.

This study explores how the Large Language Models (LLMs) adjust linguistic features to create personalized persuasive outputs. While research showed that LLMs personalize outputs, a gap remains in understanding the linguistic features of their persuasive capabilities. We identified 13 linguistic features crucial for influencing personalities across different levels of the Big Five model of personality. We analyzed how prompts with personality trait information influenced the output of 19 LLMs across five model families. The findings show that models use more anxiety-related words for neuroticism, increase achievement-related words for conscientiousness, and employ fewer cognitive processes words for openness to experience. Some model families excel at adapting language for openness to experience, others for conscientiousness, while only one model adapts language for neuroticism. Our findings show how LLMs tailor responses based on personality cues in prompts, indicating their potential to create persuasive content affecting the mind and well-being of the recipients., Comment: 31 pages

The increasing volume and complexity of IoT systems demand a transition from the cloud-centric model to a decentralized IoT architecture in the so-called Computing Continuum, with no or minimal reliance on central servers. This paradigm shift, however, raises novel research concerns for decentralized coordination, calling for accurate policies. However, building such strategies is not trivial. Our work aims to relieve the DevOps engineers from this concern and propose a solution for autonomous, decentralized task allocation at runtime for IoT systems. To this end, we present a semantic communication approach and an ad-hoc lightweight coordination strategy based on Ant Colony Optimization (ACO). We compare the ACO strategy with Random Search and Gossip protocol-based algorithms. We conduct accurate experiments with up to a hundred nodes in both a static and a dynamic environment, i.e., with device outages. We show that ACO finds a matching node with the smallest hops and messages sent. While the Gossip strategy can allocate the most tasks successfully, ACO scales better, thus being a promising candidate for decentralized task coordination in IoT clusters.

Among various topological insulators, Sn-doped Bi$_{1.1}$Sb$_{0.9}$Te$_{2}$S stands out for its exceptional properties. It has a wide energy gap and typically exhibits a well-isolated Dirac point and a Fermi level positioned within the gap. The samples we present display metallic-like low-temperature resistivity attributed to surface states, pronounced quantum oscillations observable even at 40 K, and a Fermi level located approximately 100 meV above the Dirac point. In this work, we report an unusual effect: a strong temperature dependence of the quantum oscillation frequency, which decreases by around 10\% between 2 and 40 K. This reduction significantly exceeds the expected effects of the Sommerfeld and topological corrections for Dirac quasi-particles, which could account for only one-eighth of the observed change. We attribute this change to the temperature-induced renormalization of the bulk band gap size due to electron-phonon interactions, which in turn affect the position of the surface Dirac point within the gap. Furthermore, we propose that in this compound, surface quantum oscillations can serve as a precise tool for investigating the low-temperature evolution of the bulk band gap size., Comment: 7 pages, 3 figures

Recent literature proposes combining short-term experimental and long-term observational data to provide credible alternatives to conventional observational studies for identification of long-term average treatment effects (LTEs). I show that experimental data have an auxiliary role in this context. They bring no identifying power without additional modeling assumptions. When modeling assumptions are imposed, experimental data serve to amplify their identifying power. If the assumptions fail, adding experimental data may only yield results that are farther from the truth. Motivated by this, I introduce two assumptions on treatment response that may be defensible based on economic theory or intuition. To utilize them, I develop a novel two-step identification approach that centers on bounding temporal link functions -- the relationship between short-term and mean long-term potential outcomes. The approach provides sharp bounds on LTEs for a general class of assumptions, and allows for imperfect experimental compliance -- extending existing results.

Continual learning (CL) has emerged as a critical area in machine learning, enabling neural networks to learn from evolving data distributions while mitigating catastrophic forgetting. However, recent research has identified the stability gap -- a phenomenon where models initially lose performance on previously learned tasks before partially recovering during training. Such learning dynamics are contradictory to the intuitive understanding of stability in continual learning where one would expect the performance to degrade gradually instead of rapidly decreasing and then partially recovering later. To better understand and alleviate the stability gap, we investigate it at different levels of the neural network architecture, particularly focusing on the role of the classification head. We introduce the nearest-mean classifier (NMC) as a tool to attribute the influence of the backbone and the classification head on the stability gap. Our experiments demonstrate that NMC not only improves final performance, but also significantly enhances training stability across various continual learning benchmarks, including CIFAR100, ImageNet100, CUB-200, and FGVC Aircrafts. Moreover, we find that NMC also reduces task-recency bias. Our analysis provides new insights into the stability gap and suggests that the primary contributor to this phenomenon is the linear head, rather than the insufficient representation learning., Comment: Accepted at WACV 2025

CNNs exhibit inherent equivariance to image translation, leading to efficient parameter and data usage, faster learning, and improved robustness. The concept of translation equivariant networks has been successfully extended to rotation transformation using group convolution for discrete rotation groups and harmonic functions for the continuous rotation group encompassing $360^\circ$. We explore the compatibility of the SA mechanism with full rotation equivariance, in contrast to previous studies that focused on discrete rotation. We introduce the Harmformer, a harmonic transformer with a convolutional stem that achieves equivariance for both translation and continuous rotation. Accompanied by an end-to-end equivariance proof, the Harmformer not only outperforms previous equivariant transformers, but also demonstrates inherent stability under any continuous rotation, even without seeing rotated samples during training., Comment: Appears in NeurIPS 2024 Workshop on Symmetry and Geometry in Neural Representations

Energy is a critical driver of modern economic systems. Accurate energy price forecasting plays an important role in supporting decision-making at various levels, from operational purchasing decisions at individual business organizations to policy-making. A significant body of literature has looked into energy price forecasting, investigating a wide range of methods to improve accuracy and inform these critical decisions. Given the evolving landscape of forecasting techniques, the literature lacks a thorough empirical comparison that systematically contrasts these methods. This paper provides an in-depth review of the evolution of forecasting modeling frameworks, from well-established econometric models to machine learning methods, early sequence learners such LSTMs, and more recent advancements in deep learning with transformer networks, which represent the cutting edge in forecasting. We offer a detailed review of the related literature and categorize forecasting methodologies into four model families. We also explore emerging concepts like pre-training and transfer learning, which have transformed the analysis of unstructured data and hold significant promise for time series forecasting. We address a gap in the literature by performing a comprehensive empirical analysis on these four family models, using data from the EU energy markets, we conduct a large-scale empirical study, which contrasts the forecasting accuracy of different approaches, focusing especially on alternative propositions for time series transformers.

We explore the applicability of the transcorrelated method to the elements in the second row of the periodic table. We use transcorrelated Hamiltonians in conjunction with full configuration interaction quantum Monte Carlo and coupled cluster techniques to obtain total energies and ionisation potentials, investigating their dependence on the nature and size of the basis sets used. Transcorrelation accelerates convergence to the complete basis set limit relative to conventional approaches, and chemically accurate results can generally be obtained with the cc-pVTZ basis, even with a frozen Ne core in the post-Hartree--Fock treatment., Comment: 11 pages, 5 figures

We demonstrate that epitaxial thin film antiferromagnet Mn2As exhibits the quench-switching effect, which was previously reported only in crystallographically similar antiferromagnetic CuMnAs thin films. Quench switching in Mn2As shows stronger increase in resistivity, reaching hundreds of percent at 5K, and significantly longer retention time of the metastable high-resistive state before relaxation towards the low-resistive uniform magnetic state. Qualitatively, Mn2As and CuMnAs show analogous parametric dependence of the magnitude and relaxation of the quench-switching signal. Quantitatively, relaxation dynamics in both materials show direct proportionality to the N\'eel temperature. This confirms that the quench switching has magnetic origin in both materials. The presented results suggest that the antiferromagnets crystalizing in the Cu2Sb structure are well suited for exploring and exploiting the intriguing physics of highly non-uniform magnetic states associated with the quench switching., Comment: 7pages, 4 figures and supplemental material

Quantum resistance is vital for emerging cryptographic systems as quantum technologies continue to advance towards large-scale, fault-tolerant quantum computers. Resistance may be offered by quantum key distribution (QKD), which provides information-theoretic security using quantum states of photons, but may be limited by transmission loss at long distances. An alternative approach uses classical means and is conjectured to be resistant to quantum attacks, so-called post-quantum cryptography (PQC), but it is yet to be rigorously proven, and its current implementations are computationally expensive. To overcome the security and performance challenges present in each, here we develop hybrid protocols by which QKD and PQC inter-operate within a joint quantum-classical network. In particular, we consider different hybrid designs that may offer enhanced speed and/or security over the individual performance of either approach. Furthermore, we present a method for analyzing the security of hybrid protocols in key distribution networks. Our hybrid approach paves the way for joint quantum-classical communication networks, which leverage the advantages of both QKD and PQC and can be tailored to the requirements of various practical networks., Comment: 6 pages, 3 figures, including extra supplementary materials

Space has emerged as an exciting new application area for machine learning, with several missions equipping deep learning capabilities on-board spacecraft. Pre-processing satellite data through on-board training is necessary to address the satellite downlink deficit, as not enough transmission opportunities are available to match the high rates of data generation. To scale this effort across entire constellations, collaborated training in orbit has been enabled through federated learning (FL). While current explorations of FL in this context have successfully adapted FL algorithms for scenario-specific constraints, these theoretical FL implementations face several limitations that prevent progress towards real-world deployment. To address this gap, we provide a holistic exploration of the FL in space domain on several fronts. 1) We develop a method for space-ification of existing FL algorithms, evaluated on 2) FLySTacK, our novel satellite constellation design and hardware aware testing platform where we perform rigorous algorithm evaluations. Finally we introduce 3) AutoFLSat, a generalized, hierarchical, autonomous FL algorithm for space that provides a 12.5% to 37.5% reduction in model training time than leading alternatives.

The rapid expansion of Internet of Things (IoT), edge, and embedded devices in the past decade has introduced numerous challenges in terms of security and configuration management. Simultaneously, advances in cloud-native development practices have greatly enhanced the development experience and facilitated quicker updates, thereby enhancing application security. However, applying these advances to IoT, edge, and embedded devices remains a complex task, primarily due to the heterogeneous environments and the need to support devices with extended lifespans. WebAssembly and the WebAssembly System Interface (WASI) has emerged as a promising technology to bridge this gap. As WebAssembly becomes more popular on IoT, edge, and embedded devices, there is a growing demand for hardware interface support in WebAssembly programs. This work presents WASI proposals and proof-of-concept implementations to enable hardware interaction with I2C and USB, which are two commonly used protocols in IoT, directly from WebAssembly applications. This is achieved by running the device drivers within WebAssembly as well. A thorough evaluation of the proof of concepts shows that WASI-USB introduces a minimal overhead of at most 8% compared to native operating system USB APIs. However, the results show that runtime initialization overhead can be significant in low-latency applications., Comment: Preprint of article submitted to IEEE/IFIP Network Operations and Management Symposium 2025

Modern computer systems store vast amounts of personal data, enabling advances in AI and ML but risking user privacy and trust. For privacy reasons, it is desired sometimes for an ML model to forget part of the data it was trained on. This paper presents a new approach to machine unlearning using forgetting neural networks (FNN). FNNs are neural networks with specific forgetting layers, that take inspiration from the processes involved when a human brain forgets. While FNNs had been proposed as a theoretical construct, they have not been previously used as a machine unlearning method. We describe four different types of forgetting layers and study their properties. In our experimental evaluation, we report our results on the MNIST handwritten digit recognition and fashion datasets. The effectiveness of the unlearned models was tested using Membership Inference Attacks (MIA). Successful experimental results demonstrate the great potential of our proposed method for dealing with the machine unlearning problem.

Major software failures are reported to be due to misconfiguration. As manual configuration is too error-prone to be deemed a reliable strategy for dynamic and complex systems, automated configuration management has become a standard. Countermeasures against misconfiguration can be focused on prevention or, if failure already occurred, detection. Configuration is often used as a broad term for any set of parameters or system states that dictate how an application will behave, but in this paper, we only focus on parameters consumed on process startup, usually from configuration files. Our objective is to enhance configuration management processes in environments based on the distributed cloud model, a novel cloud model that allows dynamic allocation of strategically located resources. The two mechanisms we propose are configuration validation using schemas and configuration version control with support for detecting differences between configuration versions. Our solution reduces the risk of incorrect configuration as schemas prevent any non-compliant configuration from reaching applications. However, if failure still occurs because the schema was incomplete or a valid configuration revealed existing software bugs, the version control system can precisely locate configuration changes that triggered the failure.

We prove that Floer theory induces a filtration by ideals on equivariant quantum cohomology of symplectic manifolds equipped with a $\mathbb{C}^*$-action. In particular, this gives rise to Hilbert-Poincar\'e polynomials on ordinary cohomology that depend on Floer theory. En route, the paper develops structural properties of filtrations on three versions of equivariant Floer cohomology. We obtain an explicit presentation for equivariant symplectic cohomology in the Calabi-Yau and Fano settings., Comment: 72 pages, 12 figures, fixed Theorem 1.10 and 14.3, new Remark 1.12 about the persistence module, very minor other changes

Solving complex tasks in a modern information-driven society requires novel materials and concepts for energy-efficient hardware. Antiferromagnets offer a promising platform for seeking such approaches due to their exceptional features: low power consumption and possible high integration density are desirable for information storage and processing or applications in unconventional computing. Among antiferromagnets, CuMnAs stands out for atomic-level scalable magnetic textures, analogue multilevel storage capability, and the magnetic state's control by a single electrical or femtosecond laser pulse. Using a pair of excitation laser pulses, this work examines synaptic and neuronal functionalities of CuMnAs for information processing, readily incorporating two principles of distinct characteristic timescales. Laser-induced transient heat dynamics at sub-nanosecond times represents the short-term memory and causes resistance switching due to quenching into a magnetically fragmented state. This quench switching, detectable electrically from ultrashort times to hours after writing, reminisces the long-term memory. The versatility of the principles' combination is demonstrated by operations commonly used in neural networks. Temporal latency coding, fundamental to spiking neural networks, is utilized to encode data from a grayscale image into sub-nanosecond pulse delays. Applying input laser pulses with distinct amplitudes then allows for pulse-pattern recognition. The results open pathways for designing novel computing architectures., Comment: 20 pages, 10 figures

We propose a novel approach to denoising diffusion magnetic resonance images (dMRI) using convolutional neural networks, that exploits the benefits of data acquired at multiple b-values to offset the need for many redundant observations. Denoising is especially relevant in dMRI since noise can have a deleterious impact on both quantification accuracy and image preprocessing. The most successful methods proposed to date, like Marchenko-Pastur Principal Component Analysis (MPPCA) denoising, are tailored to diffusion-weighting repeated for many encoding directions. They exploit high redundancy of the dataset that oversamples the diffusion-encoding direction space, since many directions have collinear components. However, there are many dMRI techniques that do not entail a large number of encoding directions or repetitions, and are therefore less suited to this approach. For example, clinical dMRI exams may include as few as three encoding directions, with low or negligible data redundancy across directions. Moreover, promising new dMRI approaches, like spherical b-tensor encoding (STE), benefit from high b-values while sensitizing the signal to diffusion along all directions in just a single shot. We introduce a convolutional neural network approach that we call multi-b-value-based denoising (MBD). MBD exploits the similarity in diffusion-weighted images (DWI) across different b-values but along the same diffusion encoding direction. It allows denoising of diffusion images with high noise variance while avoiding blurring, and using just a small number input images., Comment: this is a biomedical engineering work using machine learning to enhance medical images

An accurate approximation of solutions to elliptic problems in infinite domains is challenging from a computational point of view. This is due to the need to replace the infinite domain with a sufficiently large and bounded computational domain, and posing artificial boundary conditions on the boundary of the truncated computational geometry, which will then pollute the solution in an interior region of interest. For elliptic problems with periodically varying coefficients (with a possibly unknown period), a modelling strategy based on exponentially regularized elliptic problem was previously developed and analysed. The main idea was to replace the infinite domain periodic problem with a regularized elliptic problem posed over a finite domain, while retaining an accuracy decaying exponentially with respect to the size of the truncated domain. In this article, we extend the analysis to problems, where no structural assumptions on the coefficient are assumed. Moreover, the analysis here uncovers an interesting property of the right hand side in the Fourier domain for the method to converge fast for problems beyond periodicity., Comment: 20 pages, 6 figures

The NOvA collaboration reports cross-section measurements for $\nu_{\mu}$ charged-current interactions with low hadronic energy (maximum kinetic energy of 250 MeV for protons and 175 MeV for pions) in the NOvA Near Detector. The results are presented as a double-differential cross section as a function of the direct observables of the final-state muon kinematics. Results are also presented as a single-differential cross section as a function of the derived square of the four-momentum transfer, $Q^{2}$, and as a function of the derived neutrino energy. The data correspond to an accumulated 8.09$\times10^{20}$ protons-on-target (POT) in the neutrino mode of the NuMI beam, with a narrow band of neutrino energies peaked at 1.8 GeV. The analysis provides a sample of neutrino-nucleus interactions with an enhanced fraction of quasi-elastic and two-particle-two-hole (2p2h) interactions. This enhancement allows quantitative comparisons with various nuclear models. We find strong disagreement between data and theory-based models in various regions of the muon kinematic phase space, especially in the forward muon direction., Comment: 20 pages, 12 figures. The second version includes an additional citation and adds four previously missing authors

We present results of cosmological zoom-in simulations of a massive protocluster down to redshift $z\approx4$ (when the halo mass is $\approx10^{13}$ M$_\odot$) using the SWIFT code and the EAGLE galaxy formation model, focusing on supermassive black hole (BH) physics. The BH was seeded with a mass of $10^4$ M$_\odot$ at redshift $z\approx17$. We compare the base model that uses an Eddington limit on the BH accretion rate and thermal isotropic feedback by the AGN, with one where super-Eddington accretion is allowed, as well as two other models with BH spin and jets. In the base model, the BH grows at the Eddington limit from $z=9$ to $z=5.5$, when it becomes massive enough to halt its own and its host galaxy's growth through feedback. We find that allowing super-Eddington accretion leads to drastic differences, with the BH going through an intense but short super-Eddington growth burst around $z\approx7.5$, during which it increases its mass by orders of magnitude, before feedback stops further growth (of both the BH and the galaxy). By $z\approx4$ the galaxy is only half as massive in the super-Eddington cases, and an order of magnitude more extended, with the half-mass radius reaching values of a few physical kpc instead of a few hundred pc. The BH masses in our simulations are consistent with the intrinsic BH mass$-$stellar mass relation inferred from high-redshift observations by JWST. This shows that galaxy formation models using the $\Lambda$CDM cosmology are capable of reproducing the observed massive BHs at high redshift. Allowing jets, either at super- or sub-Eddington rates, has little impact on the host galaxy properties, but leads to lower BH masses as a result of stronger self-regulation, which is itself a consequence of higher feedback efficiencies., Comment: Submitted to MNRAS

In response to climate change and urban heat island effects, enhancing human thermal comfort in cities is crucial for sustainable urban development. Traditional methods for investigating the urban thermal environment and corresponding human thermal comfort level are often resource intensive, inefficient, and limited in scope. To address these challenges, we (1) introduce the concept of thermal affordance, which represents the inherent capacity of a streetscape to influence human thermal comfort based on its visual and physical features; and (2) a method to evaluate it (visual assessment of thermal affordance -- VATA), which combines street view imagery (SVI), online and in-filed surveys, and statistical learning algorithms. VATA extracts five categories of image features from SVI data and establishes 19 visual-perceptual indicators for streetscape visual assessment. Using a multi-task neural network and elastic net regression, we model their chained relationship to predict and comprehend thermal affordance for Singapore. VATA predictions are validated with field-investigated OTC data, providing a cost-effective and scalable method to assess the thermal comfort potential of urban streetscape. This framework can inform streetscape design to support sustainable, livable, and resilient urban environments.

Double- and single-differential cross sections for inclusive charged-current neutrino-nucleus scattering are reported for the kinematic domain 0 to 2 GeV/c in three-momentum transfer and 0 to 2 GeV in available energy, at a mean muon-neutrino energy of 1.86 GeV. The measurements are based on an estimated 995,760 muon-neutrino CC interactions in the scintillator medium of the NOvA Near Detector. The subdomain populated by 2-particle-2-hole reactions is identified by the cross-section excess relative to predictions for neutrino-nucleus scattering that are constrained by a data control sample. Models for 2-particle-2- hole processes are rated by chi-square comparisons of the predicted-versus-measured muon-neutrino CC inclusive cross section over the full phase space and in the restricted subdomain. Shortfalls are observed in neutrino generator predictions obtained using the theory-based Val`encia and SuSAv2 2p2h models., Comment: 20 pages, 14 figures

A novel approach for robust state estimation of marine vessels in rough water is proposed in this paper to enable tight collaboration between Unmanned Aerial Vehicles (UAVs) and a marine vessel, such as cooperative landing or object manipulation, regardless of weather conditions. Our study of marine vessel (in our case Unmanned Surface Vehicle (USV)) dynamics influenced by strong wave motion has resulted in a novel nonlinear mathematical USV model with 6 degrees of freedom (DOFs), which is required for precise USV state estimation and motion prediction. The proposed state estimation approach fuses data from multiple sensors onboard the UAV and the USV to enable redundancy and robustness under varying weather conditions of real-world applications. The proposed approach provides estimated states of the USV with 6 DOFs and predicts its future states to enable tight control of both vehicles on a receding control horizon. The proposed approach was extensively tested in the realistic Gazebo simulator and successfully experimentally validated in many real-world experiments representing different application scenarios, including agile landing on an oscillating and moving USV. A comparative study indicates that the proposed approach significantly surpassed the current state-of-the-art.

Starting from a prototypical model of elasto-plasticity in the small-strain and quasi-static setting, where the evolution of the plastic distortion is driven exclusively by the motion of discrete dislocations, this work performs a rigorous homogenization procedure to a model involving continuously-distributed dislocation fields. Our main result shows the existence of rate-independent evolutions driven by the motion of dislocation fields, obtained as limits of discrete dislocation evolutions. For all notions of solutions we employ the recent concepts of space-time integral and normal currents, which is richer than the classical approach using the Kr\"oner dislocation density tensor. The key technical challenge is to find discrete dislocation evolutions approximating a given dislocation field evolution, which requires a careful recovery construction of space-time slip trajectories and associated displacements. These methods enable one to transfer the properties, most importantly the quasi-static stability, from the discrete to the field regime., Comment: 46 pages

Spin polarization measurements were performed in three 2D Electron Gases in GaAs with densities n = 9.1, 7.2 and 6.5 x10^10 cm-2, in the quantum Hall regime. Full spin polarization at v = 1 surrounded by rapid depolarization due to Skyrmion formation was observed in all devices, consistent with past measurements. Depolarization of the v = 4/3, 8/5 states and repolarization of the v = 5/3 state was also measured, in remarkable agreement with a non-interacting, disorder-free Composite Fermion model. Optical power and temperature dependent measurements of the v = 1 state suggest a regime of non-linear optics.

Robot soccer, in its full complexity, poses an unsolved research challenge. Current solutions heavily rely on engineered heuristic strategies, which lack robustness and adaptability. Deep reinforcement learning has gained significant traction in various complex robotics tasks such as locomotion, manipulation, and competitive games (e.g., AlphaZero, OpenAI Five), making it a promising solution to the robot soccer problem. This paper introduces MARLadona. A decentralized multi-agent reinforcement learning (MARL) training pipeline capable of producing agents with sophisticated team play behavior, bridging the shortcomings of heuristic methods. Further, we created an open-source multi-agent soccer environment based on Isaac Gym. Utilizing our MARL framework and a modified a global entity encoder as our core architecture, our approach achieves a 66.8% win rate against HELIOS agent, which employs a state-of-the-art heuristic strategy. Furthermore, we provided an in-depth analysis of the policy behavior and interpreted the agent's intention using the critic network., Comment: Fixed redundant "-" in the title