Search

Artificial intelligence (AI) has transformed digital pathology by enabling biomarker prediction from high-resolution whole slide images (WSIs). However, current methods are computationally inefficient, processing thousands of redundant tiles per WSI and requiring complex aggregator models. We introduce EAGLE (Efficient Approach for Guided Local Examination), a deep learning framework that emulates pathologists by selectively analyzing informative regions. EAGLE incorporates two foundation models: CHIEF for efficient tile selection and Virchow2 for extracting high-quality features. Benchmarking was conducted against leading slide- and tile-level foundation models across 31 tasks from four cancer types, spanning morphology, biomarker prediction and prognosis. EAGLE outperformed state-of-the-art foundation models by up to 23% and achieved the highest AUROC overall. It processed a slide in 2.27 seconds, reducing computational time by more than 99% compared to existing models. This efficiency enables real-time workflows, allows pathologists to validate all tiles which are used by the model during analysis, and eliminates dependence on high-performance computing, making AI-powered pathology more accessible. By reliably identifying meaningful regions and minimizing artifacts, EAGLE provides robust and interpretable outputs, supporting rapid slide searches, integration into multi-omics pipelines and emerging clinical foundation models.

We study open orbits of symmetric subgroups of a simple connected Lie group G on a causal flag manifold. First we show that a flag manifold M of G carries an invariant causal structure if and only if G is hermitian of tube type and M is the conformal completion of the corresponding simple euclidean Jordan algebra, resp., the Shilov boundary of the associated symmetric tube domain. We then study open orbits in M under symmetric subgroups, also called causal Makarevic spaces, from the perspective of applications in Algebraic Quantum Field Theory (AQFT). A key motivation is the geometry of corresponding modular flows. The open orbits are reductive causal symmetric spaces, which arise in two flavors: compactly causal and non-compactly causal ones. In the non-compactly causal case we determine the corresponding Euler elements and their positivity regions. For compactly causal spaces, modular flows do not always exist and we determine when this is the case. Then the positivity regions of the modular flows are not globally hyperbolic, but these spaces contain other interesting globally hyperbolic subsets that can be described in terms of the conformally flat Jordan coordinates via Cayley charts. We discuss the Lorentzian case, involving de Sitter and anti-de Sitter space in some detail., Comment: 59pp

Fine-tuning large language models (LLMs) is increasingly costly as models scale to hundreds of billions of parameters, and even parameter-efficient fine-tuning (PEFT) methods like LoRA remain resource-intensive. We introduce LowRA, the first framework to enable LoRA fine-tuning below 2 bits per parameter with minimal performance loss. LowRA optimizes fine-grained quantization - mapping, threshold selection, and precision assignment - while leveraging efficient CUDA kernels for scalable deployment. Extensive evaluations across 4 LLMs and 4 datasets show that LowRA achieves a superior performance-precision trade-off above 2 bits and remains accurate down to 1.15 bits, reducing memory usage by up to 50%. Our results highlight the potential of ultra-low-bit LoRA fine-tuning for resource-constrained environments.

We investigate the shape of a tin sheet formed from a droplet struck by a nanosecond laser pulse. Specifically, we examine the dynamics of the process as a function of laser beam properties, focusing on the outstanding riddle of curvature inversion: tin sheets produced in experiments and state-of-the-art extreme ultraviolet (EUV) nanolithography light sources curve in a direction opposite to previous theoretical predictions. We resolve this discrepancy by combining direct numerical simulations with experimental data, demonstrating that curvature inversion can be explained by an instantaneous pressure impulse with low kurtosis. Specifically, we parametrize a dimensionless pressure width, $W$, using a raised cosine function and successfully reproduce the experimentally observed curvature over a wide range of laser-to-droplet diameter ratios, $0.3 < d/D_0 < 0.8$. The simulation process described in this work has applications in the EUV nanolithography industry, where a laser pulse deforms a droplet into a sheet, which is subsequently ionized by a second pulse to produce EUV-emitting plasma., Comment: 16 pages, 10 figures

Most superconducting mechanisms pair electrons within the same band, forming spin singlets. However, the discovery of multi-band superconductivity has opened new scenarios for pairing, particularly in systems with strong spin-orbit coupling. Here, we reveal inter-band pairing in the superconductor \bipd\ by mapping the amplitude of sub-gap Yu-Shiba-Rusinov (YSR) states around Vanadium adatoms deposited on its surface. The surface of \bipd\ is characterized by spin-helical-like bands near the Fermi level. Scanning tunneling spectroscopy reveals anisotropic YSR amplitude oscillations around the impurity, driven by spin-conserving Bogoliubov quasiparticle interference (BQPI). Analysis of the BQPI patterns at the YSR energy exposes inter-band pairing in this material. Interestingly, only a small subset of all possible inter-band scattering processes observed in the normal state contribute to the BQPI patterns. Combining experimental data and theory, we demonstrate that the observed band selectivity results from the hybridization of the band coupled with the impurity with other bands. Our findings reveal unconventional pairing mechanisms in \bipd\ and highlight the crucial role of spin-orbit interactions in their formation., Comment: 6 pages, 5 figures

The use of advanced X-ray sources plays a key role in the study of dynamic processes in magnetically ordered materials. The progress in X-ray free electron lasers enables the direct and simultaneous observation of the femtosecond evolution of electron and spin systems through transient X-ray absorption spectroscopy (XAS) and X-ray magnetic circular dichroism (XMCD), respectively. Such experiments allow us to resolve the response seen in the population of the spin-split valence states upon optical excitation. Here, we utilize circularly polarized ultrashort soft X-ray pulses from the new helical afterburner undulator at the free-electron laser FLASH in Hamburg to study the femtosecond dynamics of a laser-excited CoPt alloy at the Co $L_{3}$ absorption edge. Despite employing a weaker electronic excitation level we find a comparable demagnetization for the Co $3d$-states in CoPt compared to previous measurements on CoPd. This is attributed to distinctly different orbital hybridization and spin-orbit coupling between $3d$ and $4d$ vs. $3d$ and $5d$ elements in the corresponding alloys and multilayers., Comment: 17 pages, 3 figures

The Tadah! code provides a versatile platform for developing and optimizing Machine Learning Interatomic Potentials (MLIPs). By integrating composite descriptors, it allows for a nuanced representation of system interactions, customized with unique cutoff functions and interaction distances. Tadah! supports Bayesian Linear Regression (BLR) and Kernel Ridge Regression (KRR) to enhance model accuracy and uncertainty management. A key feature is its hyperparameter optimization cycle, iteratively refining model architecture to improve transferability. This approach incorporates performance constraints, aligning predictions with experimental and theoretical data. Tadah! provides an interface for LAMMPS, enabling the deployment of MLIPs in molecular dynamics simulations. It is designed for broad accessibility, supporting parallel computations on desktop and HPC systems. Tadah! leverages a modular C++ codebase, utilizing both compile-time and runtime polymorphism for flexibility and efficiency. Neural network support and predefined bonding schemes are potential future developments, and Tadah! remains open to community-driven feature expansion. Comprehensive documentation and command-line tools further streamline the development and application of MLIPs.

For the $\beta$-Hermite, Laguerre, and Jacobi ensembles of dimension $N$ there exist central limit theorems for the freezing case $\beta\to\infty$ such that the associated means and covariances can be expressed in terms of the associated Hermite, Laguerre, and Jacobi polynomials of order $N$ respectively as well as via the associated dual polynomials in the sense of de Boor and Saff. In this paper we derive limits for $N\to\infty$ for the covariances of the $r\in\mathbb N$ largest (and smallest) eigenvalues for these frozen Jacobi ensembles in terms of Bessel functions. These results correspond to the hard edge analysis in the frozen Laguerre cases by Andraus and Lerner-Brecher and to known results for finite $\beta$.

In statistical classification and machine learning, classification error is an important performance measure, which is minimized by the Bayes decision rule. In practice, the unknown true distribution is usually replaced with a model distribution estimated from the training data in the Bayes decision rule. This substitution introduces a mismatch between the Bayes error and the model-based classification error. In this work, we apply classification error bounds to study the relationship between the error mismatch and the Kullback-Leibler divergence in machine learning. Motivated by recent observations of low model-based classification errors in many machine learning tasks, bounding the Bayes error to be lower, we propose a linear approximation of the classification error bound for low Bayes error conditions. Then, the bound for class priors are discussed. Moreover, we extend the classification error bound for sequences. Using automatic speech recognition as a representative example of machine learning applications, this work analytically discusses the correlations among different performance measures with extended bounds, including cross-entropy loss, language model perplexity, and word error rate., Comment: accepted at ICASSP 2025

The almost identical charge-to-mass ratios of fully ionized helium-4 and carbon-12 ions enable simultaneous acceleration in hadron therapy synchrotrons. At the same energy per mass, helium ions exhibit a stopping range approximately three times greater than carbon ions. They can therefore be exploited for online range verification downstream of the patient during carbon ion beam irradiation. One possibility for creating this mixed beam is accelerating the two ion species sequentially through the LINAC and subsequently "mixing" them at injection energy in the synchrotron with a double injection scheme. This work reports the first successful generation, acceleration, and extraction of a mixed helium and carbon ion beam using this double multi-turn injection scheme, which was achieved at the MedAustron therapy accelerator in Austria. A description of the double multi-turn injection scheme, particle tracking simulations, and details on the implementation at the MedAustron accelerator facility are presented and discussed. Finally, measurements of the mixed beam at delivery in the irradiation room using a radiochromic film and a low-gain avalanche diode (LGAD) detector are presented., Comment: 14 pages, 12 figures

We present three possible extensions of the bit-flip channel to qutrit systems based on the diverse interpretations of the channel. Also, we extended them to higher-dimensional qudit systems, formulating different versions of dit flip channels. Finally, we studied their impact on the Negativity, as an entanglement measure, of qubit-qutrit and 2-qutrit Werner states. In doing so, we showed the inequivalence of these versions as they affect in very distinct ways the states entanglement., Comment: 8 pages, 10 figures

Software security is of utmost importance for most software systems. Developers must systematically select, plan, design, implement, and especially, maintain and evolve security features -- functionalities to mitigate attacks or protect personal data such as cryptography or access control -- to ensure the security of their software. Although security features are usually available in libraries, integrating security features requires writing and maintaining additional security-critical code. While there have been studies on the use of such libraries, surprisingly little is known about how developers engineer security features, how they select what security features to implement and which ones may require custom implementation, and the implications for maintenance. As a result, we currently rely on assumptions that are largely based on common sense or individual examples. However, to provide them with effective solutions, researchers need hard empirical data to understand what practitioners need and how they view security -- data that we currently lack. To fill this gap, we contribute an exploratory study with 26 knowledgeable industrial participants. We study how security features of software systems are selected and engineered in practice, what their code-level characteristics are, and what challenges practitioners face. Based on the empirical data gathered, we provide insights into engineering practices and validate four common assumptions., Comment: Accepted at the 47th IEEE/ACM International Conference on Software Engineering (ICSE 2025)

Automatic speech recognition (ASR) systems are well known to perform poorly on dysarthric speech. Previous works have addressed this by speaking rate modification to reduce the mismatch with typical speech. Unfortunately, these approaches rely on transcribed speech data to estimate speaking rates and phoneme durations, which might not be available for unseen speakers. Therefore, we combine unsupervised rhythm and voice conversion methods based on self-supervised speech representations to map dysarthric to typical speech. We evaluate the outputs with a large ASR model pre-trained on healthy speech without further fine-tuning and find that the proposed rhythm conversion especially improves performance for speakers of the Torgo corpus with more severe cases of dysarthria. Code and audio samples are available at https://idiap.github.io/RnV ., Comment: Accepted at ICASSP 2025 Satellite Workshop: Workshop on Speech Pathology Analysis and DEtection (SPADE)

The decoding of continuously spoken speech from neuronal activity has the potential to become an important clinical solution for paralyzed patients. Deep Learning Brain Computer Interfaces (BCIs) have recently successfully mapped neuronal activity to text contents in subjects who attempted to formulate speech. However, only small BCI datasets are available. In contrast, labeled data and pre-trained models for the closely related task of speech recognition from audio are widely available. One such model is Wav2Vec2 which has been trained in a self-supervised fashion to create meaningful representations of speech audio data. In this study, we show that patterns learned by Wav2Vec2 are transferable to brain data. Specifically, we replace its audio feature extractor with an untrained Brain Feature Extractor (BFE) model. We then execute full fine-tuning with pre-trained weights for Wav2Vec2, training ''from scratch'' without pre-trained weights as well as freezing a pre-trained Wav2Vec2 and training only the BFE each for 45 different BFE architectures. Across these experiments, the best run is from full fine-tuning with pre-trained weights, achieving a Character Error Rate (CER) of 18.54\%, outperforming the best training from scratch run by 20.46\% and that of frozen Wav2Vec2 training by 15.92\% percentage points. These results indicate that knowledge transfer from audio speech recognition to brain decoding is possible and significantly improves brain decoding performance for the same architectures. Related source code is available at https://github.com/tfiedlerdev/Wav2Vec2ForBrain., Comment: Paper was submitted to ICASSP 2025 but marginally rejected

We investigate the advantages of high-dimensional encoding for a quantum key distribution protocol. In particular, we address a BBM92-like protocol where the dimension of the systems can be larger than two and more than two mutually unbiased bases (MUBs) can be employed. Indeed, it is known that, for a system whose dimension $d$ is a prime or the power of a prime, up to $d+1$ MUBs can be found. We derive an analytic expression for the asymptotic key rate when $d+1$ MUBs are exploited and show the effects of using different numbers of MUBs on the performance of the protocol. Then, we move to the non-asymptotic case and optimize the finite key rate against collective and coherent attacks for generic dimension of the systems and all possible numbers of MUBs. In the finite-key scenario, we find that, if the number of rounds is small enough, the highest key rate is obtained by exploiting three MUBs, instead of $d+1$ as one may expect., Comment: 11 pages, 6 figures

Exploration of unknown environments is crucial for autonomous robots; it allows them to actively reason and decide on what new data to acquire for tasks such as mapping, object discovery, and environmental assessment. Existing methods, such as frontier-based methods, rely heavily on 3D map operations, which are limited by map quality and often overlook valuable context from visual cues. This work aims at leveraging 2D visual cues for efficient autonomous exploration, addressing the limitations of extracting goal poses from a 3D map. We propose a image-only frontier-based exploration system, with FrontierNet as a core component developed in this work. FrontierNet is a learning-based model that (i) detects frontiers, and (ii) predicts their information gain, from posed RGB images enhanced by monocular depth priors. Our approach provides an alternative to existing 3D-dependent exploration systems, achieving a 16% improvement in early-stage exploration efficiency, as validated through extensive simulations and real-world experiments.

Current time-synchronous sequence-to-sequence automatic speech recognition (ASR) models are trained by using sequence level cross-entropy that sums over all alignments. Due to the discriminative formulation, incorporating the right label context into the training criterion's gradient causes normalization problems and is not mathematically well-defined. The classic hybrid neural network hidden Markov model (NN-HMM) with its inherent generative formulation enables conditioning on the right label context. However, due to the HMM state-tying the identity of the right label context is never modeled explicitly. In this work, we propose a factored loss with auxiliary left and right label contexts that sums over all alignments. We show that the inclusion of the right label context is particularly beneficial when training data resources are limited. Moreover, we also show that it is possible to build a factored hybrid HMM system by relying exclusively on the full-sum criterion. Experiments were conducted on Switchboard 300h and LibriSpeech 960h., Comment: Accepted for presentation at ICASSP 2025

Security must be considered in almost every software system. Unfortunately, selecting and implementing security features remains challenging due to the variety of security threats and possible countermeasures. While security standards are intended to help developers, they are usually too abstract and vague to help implement security features, or they merely help configure such. A resource that describes security features at an abstraction level between high-level (i.e., rather too general) and low-level (i.e., rather too specific) security standards could facilitate secure systems development. To realize security features, developers typically use external security frameworks, to minimize implementation mistakes. Even then, developers still make mistakes, often resulting in security vulnerabilities. When security incidents occur or the system needs to be audited or maintained, it is essential to know the implemented security features and, more importantly, where they are located. This task, commonly referred to as feature location, is often tedious and error-prone. Therefore, we have to support long-term tracking of implemented security features. We present a study of security features in the literature and their coverage in popular security frameworks. We contribute (1) a taxonomy of 68 functional implementation-level security features including a mapping to widely used security standards, (2) an examination of 21 popular security frameworks concerning which of these security features they provide, and (3) a discussion on the representation of security features in source code. Our taxonomy aims to aid developers in selecting appropriate security features and frameworks and relating them to security standards when they need to choose and implement security features for a software system.

A fundamental alcove $\mathcal{A}$ is a tile in a paving of a vector space $V$ by an affine reflection group $W_{\mathrm{aff}}$. Its geometry encodes essential features of $W_{\mathrm{aff}}$, such as its affine Dynkin diagram $\widetilde{D}$ and fundamental group $\Omega$. In this article we investigate its full isometry group $\mathrm{Aut}(\mathcal{A})$. It is well known that the isometry group of a regular polyhedron is generated by hyperplane reflections on its faces. Being a simplex, an alcove $\mathcal{A}$ is the simplest of polyhedra, nevertheless it is seldom a regular one. In our first main result we show that $\mathrm{Aut}(\mathcal{A})$ is isomorphic to $\mathrm{Aut}(\widetilde{D})$. Building on this connection, we establish that $\mathrm{Aut}(\mathcal{A})$ is an abstract Coxeter group, with generators given by affine isometric involutions of the ambient space. Although these involutions are seldom reflections, our second main result leverages them to construct, by slicing the Komrakov--Premet fundamental polytope $\mathcal{K}$ for the action of $\Omega$, a family of fundamental polytopes for the action of $\mathrm{Aut}(\mathcal{A})$ on $\mathcal{A}$, whose vertices are contained in the vertices of $\mathcal{K}$ and whose faces are parametrized by the so-called balanced minuscule roots, which we introduce here. In an appendix, we discuss some related negative results on stratified centralizers and equivariant triangulations., Comment: 6 Figures, 3 Tables

We demonstrate the first microchip semiconductor membrane external-cavity surface-emitting laser. This compact type of laser consists solely of a semiconductor gain region present as a micron-thin membrane, sandwiched between two transparent heat spreaders. The heat spreaders have a highly reflective coating on their outer facets, which assembles the laser's plane-parallel solid-state cavity with a total length of just ~1 mm. One of the coatings with slightly reduced reflectivity acts as outcoupling mirror. The microchip membrane external-cavity surface-emitting laser (microchip MECSEL) is optically pumped with a standard fiber-coupled diode laser module emitting at 808 nm and stabilizes itself due to an occurring thermal lens. More than one watt of continuous wave output power around 1123 nm and a record value in fitted slope efficiency of ~51.4 % with MECSELs, while maintaining excellent beam quality (TEM_00, M^2 < 1.05), is demonstrated. Important properties of semiconductor lasers such as the efficiency, beam quality, and polarization were investigated. Further, this setup was used to characterize the thermal lens and it's dependence on the absorbed pump power in the microchip MECSEL. Such systems represent an attractive solution, when high-power output at customizable emission wavelength with excellent beam quality is needed in combination with very compact built size.

We investigate the Hatsugai-Kohmoto (HK) model on a square lattice, which describes both a Mott insulator at half-filling and a non-Fermi liquid phase on doping. Through the solution of this exactly solvable model with the inclusion of pairing interactions, we demonstrate the emergence of strong pair-density-wave (PDW) fluctuations associated with center-of-mass momentum $\mathbf{Q}=(\pi,\pi)$ at finite temperatures for low dopings up to a critical doping and intermediate $U$ interaction. Furthermore, we also confirm that a superconducting instability appears in the model within a wide regime of interaction $U$ and doping parameter $x$. In view of the fact that it has been recently put forward that the metal-insulator transition of the HK model belongs to the same universality class as the Mott transition of the paradigmatic Hubbard model [Huang \emph{et al.}, Nat. Phys. \textbf{18}, 511 (2022)], our work may thus shed light on an interesting scenario regarding the emergence of a fluctuating PDW phase on doping a Mott insulator, which has been argued to be relevant for understanding the physics of the cuprate superconductors in the underdoped regime., Comment: 10 pages, 4 figures

Systems may depend on parameters which one may control, or which serve to optimise the system, or are imposed externally, or they could be uncertain. This last case is taken as the ``Leitmotiv'' for the following. A reduced order model is produced from the full order model by some kind of projection onto a relatively low-dimensional manifold or subspace. The parameter dependent reduction process produces a function of the parameters into the manifold. One now wants to examine the relation between the full and the reduced state for all possible parameter values of interest. Similarly, in the field of machine learning, also a function of the parameter set into the image space of the machine learning model is learned on a training set of samples, typically minimising the mean-square error. This set may be seen as a sample from some probability distribution, and thus the training is an approximate computation of the expectation, giving an approximation to the conditional expectation, a special case of an Bayesian updating where the Bayesian loss function is the mean-square error. This offers the possibility of having a combined look at these methods, and also of introducing more general loss functions., Comment: 28 pages, 2 appendices

High-dimensional encoding for quantum key distribution (QKD) enables increased achievable key rates and robustness as compared to the standard qubit-based systems. However, experimental implementations of such systems are more complicated, expensive, and require complex security analysis. In this work we present a proof of principle high-dimensional time-phase BB84 QKD experiment using only one single-photon detector per measurement basis. We employ the temporal Talbot effect to detect QKD symbols in the control basis, and show experimentally-obtained values for the two-dimensional and four-dimensional case. We present a comparison of the secret key rates obtained from two different security proofs to display security issues stemming from asymmetric detection efficiencies in the two bases. Our results contribute to the discussion of the benefits of high-dimensional encoding and highlight the impact of security analysis on the achievable QKD performance., Comment: 8 pages, 8 figures

Antiferromagnets are promising candidates for ultrafast spintronic applications, leveraging current-induced spin-orbit torques. However, experimentally distinguishing between different switching mechanisms of the staggered magnetization (N\'eel vector) driven by current pulses remains a challenge. In an exemplary study of the collinear antiferromagnetic compound Mn$_2$Au, we demonstrate that slower thermomagnetoelastic effects predominantly govern switching over a wide parameter range. In the regime of short current pulses in the nanosecond range, however, we observe fully N\'eel spin-orbit torque driven switching. We show that this ultrafast mechanism enables the complete directional alignment of the N\'eel vector by current pulses in device structures., Comment: 8 pages, 6 figures

We show rigidity results for the operator equations T(f.g) = Tf.Tg, T(f*g) = Tf.Tg and T(f.g) = Tf*Tg for bijective operators T acting on sufficently large spaces of smooth functions. Typically a condition like |T(f.g) - Tf.Tg| < a for all f, g with a fixed function a will imply T(f.g) = Tf.Tg. Theorems of Alesker, Artstein-Avidan, Faifman and Milman then yield characterizations (up to diffeomorphisms) of the Fourier transform by mapping products into convolutions and vice-versa on the Schwartz space.

A longstanding conjecture in $\phi^4_4$ theory is that primitive graphs dominate the beta function asymptotically at large loop order in the minimal-subtraction scheme. Here we investigate this issue by exploiting additional combinatorial structure coming from an extension to vectors with $O(N)$ symmetry. For the 0-dimensional case, we calculate the $N$-dependent generating function of primitive graphs and its asymptotics, including arbitrarily many subleading corrections. We find that the leading asymptotic growth rate becomes visible only above $\approx 25$ loops, while data at lower order is suggestive of a wrong asymptotics. Our results also yield the exact asymptotics of Martin invariants. In 4D, each graph comes with a nontrivial Feynman integral, its period. We give bounds on the degree in $N$ for primitive and non-primitive graphs, and construct the primitive graphs of highest degree explicitly. We calculate the 4D primitive beta function numerically up to 17 loops. The qualitative behaviour is similar to the 0D series, with a small but systematic tendency for the 4D data to grow faster with $N$, indicating a correlation between periods and $O(N)$-symmetry factors. The zeros of the 4D primitive beta function approach their asymptotic locations quickly, but, like in 0D, the growth rate of the 4D primitive beta function does not match its asymptotics even at 17 loops. Our results improve on the knowledge of asymptotics in QFT by providing concrete analytic and numerical values, and putting individual observables into a broader context of $\phi^4_4$ theory in 0D and 4D. We demonstrate that even if certain quantities are in agreement with the asymptotics already below 10 loops, this must not be mistaken as evidence that overall an asymptotic regime has been reached., Comment: 64 pages, 26 figures, 7 tables. Further raw data available from the first author

Keyword-based searches are today's standard in digital libraries. Yet, complex retrieval scenarios like in scientific knowledge bases, need more sophisticated access paths. Although each document somewhat contributes to a domain's body of knowledge, the exact structure between keywords, i.e., their possible relationships, and the contexts spanned within each single document will be crucial for effective retrieval. Following this logic, individual documents can be seen as small-scale knowledge graphs on which graph queries can provide focused document retrieval. We implemented a full-fledged graph-based discovery system for the biomedical domain and demonstrated its benefits in the past. Unfortunately, graph-based retrieval methods generally follow an 'exact match' paradigm, which severely hampers search efficiency, since exact match results are hard to rank by relevance. This paper extends our existing discovery system and contributes effective graph-based unsupervised ranking methods, a new query relaxation paradigm, and ontological rewriting. These extensions improve the system further so that users can retrieve results with higher precision and higher recall due to partial matching and ontological rewriting., Comment: Technical Report of our accepted paper at AI4LAC@JCDL2024. 11 pages, 5 figures

Digital libraries provide different access paths, allowing users to explore their collections. For instance, paper recommendation suggests literature similar to some selected paper. Their implementation is often cost-intensive, especially if neural methods are applied. Additionally, it is hard for users to understand or guess why a recommendation should be relevant for them. That is why we tackled the problem from a different perspective. We propose XGPRec, a graph-based and thus explainable method which we integrate into our existing graph-based biomedical discovery system. Moreover, we show that XGPRec (1) can, in terms of computational costs, manage a real digital library collection with 37M documents from the biomedical domain, (2) performs well on established test collections and concept-centric information needs, and (3) generates explanations that proved to be beneficial in a preliminary user study. We share our code so that user libraries can build upon XGPRec., Comment: Technical Report of our accepted paper at AI4LAC@JCDL2024. 12 pages, 3 figures

We present an in-depth analysis of ChiSCAT, a recently introduced interferometric microscopy scheme to detect recurring micromotion events in cells. Experimentally, we demonstrate that illumination with low-coherence sources can greatly improve the robustness of the scheme to vibrations. Theoretically, we analyze the performance of ChiSCAT under various noise models, in particular photon shot noise and noise dominated by cellular motions other than the signal. We finally propose ways to improve performance, especially in a setting dominated by cell motions, and conclude with an outlook on potential future directions.

Quantum Key Distribution (QKD) is a promising technology for secure communication. Nevertheless, QKD is still treated with caution in certain contexts due to potential gaps between theoretical models and actual QKD implementations. A common assumption in security proofs is that the detection probability at the receiver, for a given input state, is independent of the measurement basis, which might not always be verified and could lead to security loopholes. This paper presents a security proof for QKD protocols that does not rely on the above assumption and is thus applicable in scenarios with detection probability mismatches, even when induced by the adversary. We demonstrate, through simulations, that our proof can extract positive key rates for setups vulnerable to large detection probability mismatches. This is achieved by monitoring whether an adversary is actively exploiting such vulnerabilities, instead of considering the worst-case scenario as in previous proofs. Our work highlights the importance of accounting for basis-dependent detection probabilities and provides a concrete solution for improving the security of practical QKD systems., Comment: 14 pages and 5 figures in main text; 5 appendices

We study the three-nucleon system at next-to-next-to-next-to-leading order ($\mathrm{N^3LO}$) in the framework of chiral effective field theory (EFT) on the lattice. Our calculations do not rely on a perturbative treatment of subleading contributions to the nuclear forces. For the two-nucleon potential, we apply the previously developed $\mathrm{N^3LO}$ lattice interaction. For the leading contribution to the three-nucleon force, we determine the two low-energy constants (LECs) in the contact interactions by adjusting the ground state energy and half-life of triton, where the latter employs the nuclear axial current at $\mathrm{N^2LO}$ in chiral EFT. Additionally, the ground state energy of helion and the charge radii of the two considered nuclei are computed. No effect of the smearing regularization in the three-nucleon contact interaction is observed here. We compare our results with recent lattice-EFT calculations that are based on a potential tuned to light and medium-mass nuclei using the wave-function-matching technique to circumvent the Monte-Carlo sign problem., Comment: 8 pages, 1 figure, 1 table, to be published in proceedings of the 11th International Workshop on Chiral Dynamics - CD2024

We explore ideas for scaling verification methods for quantum circuits using SMT (Satisfiability Modulo Theories) solvers. We propose two primary strategies: (1) decomposing proof obligations via compositional verification and (2) leveraging linear over-approximation techniques for gate effects. We present two examples and demonstrate the application of these ideas to proof Hamming weight preservation.

Recent approaches have successfully focused on the segmentation of static reconstructions, thereby equipping downstream applications with semantic 3D understanding. However, the world in which we live is dynamic, characterized by numerous interactions between the environment and humans or robotic agents. Static semantic maps are unable to capture this information, and the naive solution of rescanning the environment after every change is both costly and ineffective in tracking e.g. objects being stored away in drawers. With Lost & Found we present an approach that addresses this limitation. Based solely on egocentric recordings with corresponding hand position and camera pose estimates, we are able to track the 6DoF poses of the moving object within the detected interaction interval. These changes are applied online to a transformable scene graph that captures object-level relations. Compared to state-of-the-art object pose trackers, our approach is more reliable in handling the challenging egocentric viewpoint and the lack of depth information. It outperforms the second-best approach by 34% and 56% for translational and orientational error, respectively, and produces visibly smoother 6DoF object trajectories. In addition, we illustrate how the acquired interaction information in the dynamic scene graph can be employed in the context of robotic applications that would otherwise be unfeasible: We show how our method allows to command a mobile manipulator through teach & repeat, and how information about prior interaction allows a mobile manipulator to retrieve an object hidden in a drawer. Code, videos and corresponding data are accessible at https://behretj.github.io/LostAndFound., Comment: Webpage: https://behretj.github.io/LostAndFound

We propose a new approach to studying hyperbolic Kac-Moody algebras, focussing on the rank-3 algebra $\mathfrak{F}$ first investigated by Feingold and Frenkel. Our approach is based on the concrete realization of this Lie algebra in terms of a Hilbert space of transverse and longitudinal physical string states, which are expressed in a basis using DDF operators. When decomposed under its affine subalgebra $A_1^{(1)}$, the algebra $\mathfrak{F}$ decomposes into an infinite sum of affine representation spaces of $A_1^{(1)}$ for all levels $\ell\in\mathbb{Z}$. For $|\ell| >1$ there appear in addition coset Virasoro representations for all minimal models of central charge $c<1$, but the different level-$\ell$ sectors of $\mathfrak{F}$ do not form proper representations of these because they are incompletely realized in $\mathfrak{F}$. To get around this problem we propose to nevertheless exploit the coset Virasoro algebra for each level by identifying for each level a (for $|\ell|\geq 3$ infinite) set of `Virasoro ground states' that are not necessarily elements of $\mathfrak{F}$ (in which case we refer to them as `virtual'), but from which the level-$\ell$ sectors of $\mathfrak{F}$ can be fully generated by the joint action of affine and coset Virasoro raising operators. We conjecture (and present partial evidence) that the Virasoro ground states for $|\ell|\geq 3$ in turn can be generated from a finite set of `maximal ground states' by the additional action of the `spectator' coset Virasoro raising operators present for all levels $|\ell| > 2$. Our results hint at an intriguing but so far elusive secret behind Einstein's theory of gravity, with possibly important implications for quantum cosmology., Comment: 48 pages

Modern reconstruction techniques can effectively model complex 3D scenes from sparse 2D views. However, automatically assessing the quality of novel views and identifying artifacts is challenging due to the lack of ground truth images and the limitations of no-reference image metrics in predicting detailed artifact maps. The absence of such quality metrics hinders accurate predictions of the quality of generated views and limits the adoption of post-processing techniques, such as inpainting, to enhance reconstruction quality. In this work, we propose a new no-reference metric, Puzzle Similarity, which is designed to localize artifacts in novel views. Our approach utilizes image patch statistics from the input views to establish a scene-specific distribution that is later used to identify poorly reconstructed regions in the novel views. We test and evaluate our method in the context of 3D reconstruction; to this end, we collected a novel dataset of human quality assessment in unseen reconstructed views. Through this dataset, we demonstrate that our method can not only successfully localize artifacts in novel views, correlating with human assessment, but do so without direct references. Surprisingly, our metric outperforms both no-reference metrics and popular full-reference image metrics. We can leverage our new metric to enhance applications like automatic image restoration, guided acquisition, or 3D reconstruction from sparse inputs.

The advancement of novel combinatorial CRISPR screening technologies enables the identification of synergistic gene combinations on a large scale. This is crucial for developing novel and effective combination therapies, but the combinatorial space makes exhaustive experimentation infeasible. We introduce NAIAD, an active learning framework that efficiently discovers optimal gene pairs capable of driving cells toward desired cellular phenotypes. NAIAD leverages single-gene perturbation effects and adaptive gene embeddings that scale with the training data size, mitigating overfitting in small-sample learning while capturing complex gene interactions as more data is collected. Evaluated on four CRISPR combinatorial perturbation datasets totaling over 350,000 genetic interactions, NAIAD, trained on small datasets, outperforms existing models by up to 40\% relative to the second-best. NAIAD's recommendation system prioritizes gene pairs with the maximum predicted effects, resulting in the highest marginal gain in each AI-experiment round and accelerating discovery with fewer CRISPR experimental iterations. Our NAIAD framework (https://github.com/NeptuneBio/NAIAD) improves the identification of novel, effective gene combinations, enabling more efficient CRISPR library design and offering promising applications in genomics research and therapeutic development.

In both the random hopping model and at topological phase transitions in one-dimensional chiral systems, the Lyapunov exponent vanishes at zero energy, but is here shown to have an inverse logarithmic increase with a coefficient that is computed explicitly. This is the counterpart of the Dyson spike in the density of states. The argument also transposes to the free energy density of the random field Ising model, and more generally to many so-called balanced hyperbolic critical points. It is based on the fact that the Furstenberg measure in rescaled logarithmic Dyson-Schmidt variables can be well-approximated by an absolutely continuous measure with trapezoidal density.

Blazars are a subclass of active galactic nuclei with relativistic jets pointing toward the observer. They are notable for their flux variability at all observed wavelengths and timescales. Together with simultaneous measurements at lower energies, the very-high-energy (VHE) emission observed during blazar flares may be used to probe the population of accelerated particles. However, optimally triggering observations of blazar high states can be challenging. Notable examples include identifying a flaring episode in real time and predicting VHE flaring activity based on lower-energy observables. For this purpose, we have developed a novel deep learning analysis framework, based on data-driven anomaly detection techniques. It is capable of detecting various types of anomalies in real-world, multiwavelength light curves, ranging from clear high states to subtle correlations across bands. Based on unsupervised anomaly detection and clustering methods, we differentiate source variability from noisy background activity, without the need for a labeled training data set of flaring states. The framework incorporates measurement uncertainties and is robust given data quality challenges, such as varying cadences and observational gaps. We evaluate our approach using both historical data and simulations of blazar light curves in two energy bands, corresponding to sources observable with the Fermi Large Area Telescope and the upcoming Cherenkov Telescope Array Observatory. In a statistical analysis, we show that our framework can reliably detect known historical flares.

This paper addresses the computation of ground states of multicomponent Bose-Einstein condensates, defined as the global minimiser of an energy functional on an infinite-dimensional generalised oblique manifold. We establish the existence of the ground state, prove its uniqueness up to scaling, and characterise it as the solution to a coupled nonlinear eigenvector problem. By equipping the manifold with several Riemannian metrics, we introduce a suite of Riemannian gradient descent and Riemannian Newton methods. Metrics that incorporate first- or second-order information about the energy are particularly advantageous, effectively preconditioning the resulting methods. For a Riemannian gradient descent method with an energy-adaptive metric, we provide a qualitative global and quantitative local convergence analysis, confirming its reliability and robustness with respect to the choice of the spatial discretisation. Numerical experiments highlight the computational efficiency of both the Riemannian gradient descent and Newton methods.

This work examines a discrete elastic energy system with local interactions described by a discrete second-order functional in the symmetric gradient and additional non-local random long-range interactions. We analyze the asymptotic behavior of this model as the grid size tends to zero. Assuming that the occurrence of long-range interactions is Bernoulli distributed and depends only on the distance between the considered grid points, we derive - in an appropriate scaling regime - a fractional p-Laplace-type term as the long-range interactions' homogenized limit. A specific feature of the presented homogenization process is that the random weights of the p-Laplace-type term are non-stationary, thus making the use of standard ergodic theorems impossible. For the entire discrete energy system, we derive a non-local fractional p-Laplace-type term and a local second-order functional in the symmetric gradient. Our model can be used to describe the elastic energy of standard, homogeneous, materials that are reinforced with long-range stiff fibers.

Digital libraries that maintain extensive textual collections may want to further enrich their content for certain downstream applications, e.g., building knowledge graphs, semantic enrichment of documents, or implementing novel access paths. All of these applications require some text processing, either to identify relevant entities, extract semantic relationships between them, or to classify documents into some categories. However, implementing reliable, supervised workflows can become quite challenging for a digital library because suitable training data must be crafted, and reliable models must be trained. While many works focus on achieving the highest accuracy on some benchmarks, we tackle the problem from a digital library practitioner. In other words, we also consider trade-offs between accuracy and application costs, dive into training data generation through distant supervision and large language models such as ChatGPT, LLama, and Olmo, and discuss how to design final pipelines. Therefore, we focus on relation extraction and text classification, using the showcase of eight biomedical benchmarks., Comment: JCD2024 Full Paper, 12 pages, 6 figures

This paper explores the rapid development of a telephone call summarization system utilizing large language models (LLMs). Our approach involves initial experiments with prompting existing LLMs to generate summaries of telephone conversations, followed by the creation of a tailored synthetic training dataset utilizing stronger frontier models. We place special focus on the diversity of the generated data and on the ability to control the length of the generated summaries to meet various use-case specific requirements. The effectiveness of our method is evaluated using two state-of-the-art LLM-as-a-judge-based evaluation techniques to ensure the quality and relevance of the summaries. Our results show that fine-tuned Llama-2-7B-based summarization model performs on-par with GPT-4 in terms of factual accuracy, completeness and conciseness. Our findings demonstrate the potential for quickly bootstrapping a practical and efficient call summarization system., Comment: Accepted at the The International Conference on Foundation and Large Language Models (FLLM2024)

The performance of neural networks scales with both their size and the amount of data they have been trained on. This is shown in both language and image generation. However, this requires scaling-friendly network architectures as well as large-scale datasets. Even though scaling-friendly architectures like transformers have emerged for 3D vision tasks, the GPT-moment of 3D vision remains distant due to the lack of training data. In this paper, we introduce ARKit LabelMaker, the first large-scale, real-world 3D dataset with dense semantic annotations. Specifically, we complement ARKitScenes dataset with dense semantic annotations that are automatically generated at scale. To this end, we extend LabelMaker, a recent automatic annotation pipeline, to serve the needs of large-scale pre-training. This involves extending the pipeline with cutting-edge segmentation models as well as making it robust to the challenges of large-scale processing. Further, we push forward the state-of-the-art performance on ScanNet and ScanNet200 dataset with prevalent 3D semantic segmentation models, demonstrating the efficacy of our generated dataset.

Gaussian splatting and single/multi-view depth estimation are typically studied in isolation. In this paper, we present DepthSplat to connect Gaussian splatting and depth estimation and study their interactions. More specifically, we first contribute a robust multi-view depth model by leveraging pre-trained monocular depth features, leading to high-quality feed-forward 3D Gaussian splatting reconstructions. We also show that Gaussian splatting can serve as an unsupervised pre-training objective for learning powerful depth models from large-scale unlabeled datasets. We validate the synergy between Gaussian splatting and depth estimation through extensive ablation and cross-task transfer experiments. Our DepthSplat achieves state-of-the-art performance on ScanNet, RealEstate10K and DL3DV datasets in terms of both depth estimation and novel view synthesis, demonstrating the mutual benefits of connecting both tasks., Comment: Project page: https://haofeixu.github.io/depthsplat/ Code: https://github.com/cvg/depthsplat

The shift towards decentralized and renewable energy sources has introduced significant challenges to traditional power systems, necessitating innovative market designs. Local energy markets present a viable solution for integrating distributed energy resources such as photovoltaic systems, electric vehicles, and heat pumps within various grid topologies. This study investigates the techno-economic benefits of local energy markets compared to conventional market designs, focusing on their impact on average energy prices and operational peak power, using a self-developed agent-based energy system simulation tool. Through comprehensive simulations across the countryside, rural, suburban, and urban grid topologies with varying penetration levels of the distributed energy resources, totaling 400 simulation setups, we demonstrate that local energy markets can enhance economic efficiency and grid stability with 99 % of the scenarios boasting lower average energy prices and 80 % lower operational peak power levels. Our findings suggest that local energy markets can play a role in the future energy system, especially in areas with high shares of PV and HP, provided that additional infrastructure, management costs, and bureaucratic complexity are kept to a minimum., Comment: 39 pages, 9 figures, 4 tables

Archives of libraries contain many materials, which have not yet been made available to the public. The prioritization of which content to provide and especially how to design effective access paths depend on potential users' needs. As a case study we interviewed researchers working on topics related to one German philosopher to map out their information interaction workflow. Additionally, we deeply analyze study participants' requirements for a digital library system. Moreover, we discuss how existing methods may meet their requirements and which implications these methods may have in a practical digital library setting, e.g., computational costs and hallucinations. In brief, this paper contributes the findings of our digital humanities case study resulting in system requirements., Comment: Technical Report of our accepted JCDL 2024 Poster, 6 pages

Human-robot interaction through mixed reality (MR) technologies enables novel, intuitive interfaces to control robots in remote operations. Such interfaces facilitate operations in hazardous environments, where human presence is risky, yet human oversight remains crucial. Potential environments include disaster response scenarios and areas with high radiation or toxic chemicals. In this paper we present an interface system projecting a 3D representation of a scanned room as a scaled-down 'dollhouse' hologram, allowing users to select and manipulate objects using a straightforward drag-and-drop interface. We then translate these drag-and-drop user commands into real-time robot actions based on the recent Spot-Compose framework. The Unity-based application provides an interactive tutorial and a user-friendly experience, ensuring ease of use. Through comprehensive end-to-end testing, we validate the system's capability in executing pick-and-place tasks and a complementary user study affirms the interface's intuitive controls. Our findings highlight the advantages of this interface in improving user experience and operational efficiency. This work lays the groundwork for a robust framework that advances the potential for seamless human-robot collaboration in diverse applications. Paper website: https://holospot.github.io/, Comment: 6 pages, 8 figures, submitted to ICRA 2025

The incompressible Euler equations are an important model system in computational fluid dynamics. Fast high-order methods for the solution of this time-dependent system of partial differential equations are of particular interest: due to their exponential convergence in the polynomial degree they can make efficient use of computational resources. To address this challenge we describe a novel timestepping method which combines a hybridised Discontinuous Galerkin method for the spatial discretisation with IMEX timestepping schemes, thus achieving high-order accuracy in both space and time. The computational bottleneck is the solution of a (block-) sparse linear system to compute updates to pressure and velocity at each stage of the IMEX integrator. Following Chorin's projection approach, this update of the velocity and pressure fields is split into two stages. As a result, the hybridised equation for the implicit pressure-velocity problem is reduced to the well-known system which arises in hybridised mixed formulations of the Poisson- or diffusion problem and for which efficient multigrid preconditioners have been developed. Splitting errors can be reduced systematically by embedding this update into a preconditioned Richardson iteration. The accuracy and efficiency of the new method is demonstrated numerically for two time-dependent testcases that have been previously studied in the literature., Comment: 28 pages, 11 figures, 8 tables