Search

Combining a precise sputter etching method with subsequent AlO$_x$ growth within an atomic layer deposition chamber enables fabrication of atomically flat lateral patterns of SiO$_2$ and AlO$_x$. The transfer of MoSe$_2$ monolayers onto these dielectrically modulated substrates results in formation of lateral heterojunctions, with the flat substrate topography leading to minimal strain across the junction. Kelvin probe force microscopy (KPFM) measurements show significant variations in the contact potential difference (CPD) across the interface, with AlO$_x$ regions inducing a 230~mV increase in CPD. Spatially resolved photoluminescence spectroscopy reveals shifts in spectral weight of neutral and charged exciton species across the different dielectric regions. On the AlO$_x$ side, the Fermi energy moves closer to the conduction band, leading to a higher trion-to-exciton ratio, indicating a bandgap shift consistent with CPD changes. In addition, transient reflection spectroscopy highlights the influence of the dielectric environment on carrier dynamics, with the SiO$_2$ side exhibiting rapid carrier decay typical of neutral exciton recombination. In contrast, the AlO$_x$ side shows slower, mixed decay behavior consistent with conversion of trions back into excitons. These results demonstrate how dielectric substrate engineering can tune the electronic and optical characteristics of proximal two-dimensional materials, allowing scalable fabrication of advanced junctions for novel (opto)electronics applications., Comment: 24 pages, 5 figures. SI: 2 pages, 3 figures. We welcome your feedback!

A good process model is expected not only to reflect the behavior of the process, but also to be as easy to read and understand as possible. Because preferences vary across different applications, numerous measures provide ways to reflect the complexity of a model with a numeric score. However, this abundance of different complexity measures makes it difficult to select one for analysis. Furthermore, most complexity measures are defined for BPMN or EPC, but not for workflow nets. This paper is an extended analysis of complexity measures and their formal properties. It adapts existing complexity measures to the world of workflow nets. It then compares these measures with a set of properties originally defined for software complexity, as well as new extensions to it. We discuss the importance of the properties in theory by evaluating whether matured complexity measures should fulfill them or whether they are optional. We find that not all inspected properties are mandatory, but also demonstrate that the behavior of evolutionary process discovery algorithms is influenced by some of these properties. Our findings help analysts to choose the right complexity measure for their use-case., Comment: 73 pages, 68 figures

Multiplicative linear logic is a very well studied formal system, and most such studies are concerned with the one-sided sequent calculus. In this paper we look in detail at existing translations between a deep inference system and the standard sequent calculus one, provide a simplified translation, and provide a formal proof that a standard approach to modelling is indeed invariant to all these translations. En route we establish a necessary condition for provable sequents related to the number of pars and tensors in a formula that seems to be missing from the literature.

For the realization of the future urban air mobility, reliable information exchange based on robust and efficient communication between all airspace participants will be one of the key factors to ensure safe operations. Especially in dense urban scenarios, the direct and fast information exchange between drones based on Drone-to-Drone communications is a promising technology for enabling reliable collision avoidance systems. However, to mitigate collisions and to increase overall reliability, unmanned aircraft still lack a redundant, higher-level safety net to coordinate and monitor traffic, as is common in today's civil aviation. In addition, direct and fast information exchange based on ad hoc communication is needed to cope with the very short reaction times required to avoid collisions and to cope with the the high traffic densities. Therefore, we are developing a \ac{d2d} communication and surveillance system, called DroneCAST, which is specifically tailored to the requirements of a future urban airspace and will be part of a multi-link approach. In this work we discuss challenges and expected safety-critical applications that will have to rely on communications for \ac{uam} and present our communication concept and necessary steps towards DroneCAST. As a first step towards an implementation, we equipped two drones with hardware prototypes of the experimental communication system and performed several flights around the model city to evaluate the performance of the hardware and to demonstrate different applications that will rely on robust and efficient communications.

We demonstrate three-dimensional magnetic resonance tomography with a resolution down to 5.99 +- 0.07 nm. Our measurements use lithographically fabricated microwires as a source of three-dimensional magnetic field gradients, which we use to image NV centers in a densely doped diamond by Fourier-accelerated magnetic resonance tomography. We also present a compressed sensing scheme for imaging of a spatially localized ensemble from undersampled data, which allows for a direct visual interpretation without numerical optimization. The resolution achieved in our work approaches the positioning accuracy of site-directed spin labeling, paving the way to three-dimensional structure analysis by magnetic-gradient based tomography.

Recent developments in surrogate construction predominantly focused on two strategies to improve surrogate accuracy. Firstly, component-wise domain scaling informed by cross-validation. Secondly, regression to construct response surfaces using additional information in the form of additional function-values sampled from multi-fidelity models and gradients. Component-wise domain scaling reliably improves the surrogate quality at low dimensions but has been shown to suffer from high computational costs for higher dimensional problems. The second strategy, adding gradients to train surrogates, typically results in regression surrogates. Counter-intuitively, these gradient-enhanced regression-based surrogates do not exhibit improved accuracy compared to surrogates only interpolating function values. This study empirically establishes three main findings. Firstly, constructing the surrogate in poorly scaled domains is the predominant cause of deteriorating response surfaces when regressing with additional gradient information. Secondly, surrogate accuracy improves if the surrogates are constructed in a fully transformed domain, by scaling and rotating the original domain, not just simply scaling the domain. The domain transformation scheme should be based on the local curvature of the approximation surface and not its global curvature. Thirdly, the main benefit of gradient information is to efficiently determine the (near) optimal domain in which to construct the surrogate. This study proposes a foundational transformation algorithm that performs near-optimal transformations for lower dimensional problems. The algorithm consistently outperforms cross-validation-based component-wise domain scaling for higher dimensional problems. A carefully selected test problem set that varies between 2 and 16-dimensional problems is used to clearly demonstrate the three main findings of this study., Comment: 20 pages, 28 figures

The rapid progress of AI is fueled by increasingly large and computationally intensive machine learning models and datasets. As a consequence, the amount of compute used in training state-of-the-art models is exponentially increasing (doubling every 10 months between 2015 and 2022), resulting in a large carbon footprint. Federated Learning (FL) - a collaborative machine learning technique for training a centralized model using data of decentralized entities - can also be resource-intensive and have a significant carbon footprint, particularly when deployed at scale. Unlike centralized AI that can reliably tap into renewables at strategically placed data centers, cross-device FL may leverage as many as hundreds of millions of globally distributed end-user devices with diverse energy sources. Green AI is a novel and important research area where carbon footprint is regarded as an evaluation criterion for AI, alongside accuracy, convergence speed, and other metrics. In this paper, we propose the concept of Green FL, which involves optimizing FL parameters and making design choices to minimize carbon emissions consistent with competitive performance and training time. The contributions of this work are two-fold. First, we adopt a data-driven approach to quantify the carbon emissions of FL by directly measuring real-world at-scale FL tasks running on millions of phones. Second, we present challenges, guidelines, and lessons learned from studying the trade-off between energy efficiency, performance, and time-to-train in a production FL system. Our findings offer valuable insights into how FL can reduce its carbon footprint, and they provide a foundation for future research in the area of Green AI.

This book is the result of a seminar in which we reviewed multimodal approaches and attempted to create a solid overview of the field, starting with the current state-of-the-art approaches in the two subfields of Deep Learning individually. Further, modeling frameworks are discussed where one modality is transformed into the other, as well as models in which one modality is utilized to enhance representation learning for the other. To conclude the second part, architectures with a focus on handling both modalities simultaneously are introduced. Finally, we also cover other modalities as well as general-purpose multi-modal models, which are able to handle different tasks on different modalities within one unified architecture. One interesting application (Generative Art) eventually caps off this booklet.

Spin defects in hexagonal Boron Nitride (hBN) attract increasing interest for quantum technology since they represent optically-addressable qubits in a van der Waals material. In particular, negatively-charged boron vacancy centers (${V_B}^-$) in hBN have shown promise as sensors of temperature, pressure, and static magnetic fields. However, the short spin coherence time of this defect currently limits its scope for quantum technology. Here, we apply dynamical decoupling techniques to suppress magnetic noise and extend the spin coherence time by nearly two orders of magnitude, approaching the fundamental $T_1$ relaxation limit. Based on this improvement, we demonstrate advanced spin control and a set of quantum sensing protocols to detect electromagnetic signals in the MHz range with sub-Hz resolution. This work lays the foundation for nanoscale sensing using spin defects in an exfoliable material and opens a promising path to quantum sensors and quantum networks integrated into ultra-thin structures.

Various privacy-preserving frameworks that respect the individual's privacy in the analysis of data have been developed in recent years. However, available model classes such as simple statistics or generalized linear models lack the flexibility required for a good approximation of the underlying data-generating process in practice. In this paper, we propose an algorithm for a distributed, privacy-preserving, and lossless estimation of generalized additive mixed models (GAMM) using component-wise gradient boosting (CWB). Making use of CWB allows us to reframe the GAMM estimation as a distributed fitting of base learners using the $L_2$-loss. In order to account for the heterogeneity of different data location sites, we propose a distributed version of a row-wise tensor product that allows the computation of site-specific (smooth) effects. Our adaption of CWB preserves all the important properties of the original algorithm, such as an unbiased feature selection and the feasibility to fit models in high-dimensional feature spaces, and yields equivalent model estimates as CWB on pooled data. Next to a derivation of the equivalence of both algorithms, we also showcase the efficacy of our algorithm on a distributed heart disease data set and compare it with state-of-the-art methods.

Integrating quantum emitters into nanocavities which simultaneously couple to the photonic and mechanical modes is critical for interfacing electron spins, photons and phonons in the cavity QED system. Here, we investigate the interaction between the charged boron vacancy $V_B^-$, ultra-high-Q ($\sim10^5$) cavity photonic modes and local phonon modes. A pronounced asymmetry is observed in the emission spectrum for cavities with Q-factor above a threshold of 10$^4$. Similar asymmetries are not observed for cavities without $V_B^-$ centers. To explain our findings, we model the system with phonon-induced light-matter coupling based on $V_B^-$ centers, and compare to the Jaynes-Cummings model for usual emitters. Our results reveal that the multipartite interplay arises during the light-matter coupling of $V_B^-$ centers, illustrating that it is phonon-induced, rather than being caused by thermal population of phonon modes. Such emitter-optomechanical interaction between different photon ($V_B^-$ emission, cavity photonic) and phonon ($V_B^-$ phonon, cavity mechanical) modes provides a novel system to interface spin defects, photons and phonons in condensed matters.

Magnetic resonance imaging of spin systems combines scientific applications in medicine, chemistry and physics. Here, we investigate the pixel-wise coherent quantum dynamics of spins consisting of a 40 by 40 micron sized region of interest implanted with nitrogen vacancy centers (NV) coupled to a nano-magnetic flake of $\mathrm{CrTe_2}$. $\mathrm{CrTe_2}$ is an in-plane van der Waals ferromagnet, which we can probe quantitatively by the NV electron's spin signal even at room temperature. First, we combine the nano-scale sample shapes measured by atomic force microscope with the magnetic resonance imaging data. We then map out the coherent dynamics of the colour centers coupled to the van der Waals ferromagnet using pixel-wise coherent Rabi and Ramsey imaging of the NV sensor layer. Next, we fit the pixel-wise solution of the Hamiltonian to the quantum sensor data. Combining data and model, we can explore the detuning range of the spin oscillation with a quantum dynamic range of over $\left|\Delta_{max}\right|= 60 { }\mathrm{MHz} $ in the Ramsey interferometry mode. Finally, we show the effect of the $\mathrm{CrTe_2}$ van der Waals magnet on the coherence of the NV sensor layer and measure a 70 times increase in the maximum frequency of the quantum oscillation going from the Rabi to the Ramsey imaging mode.

Distributed statistical analyses provide a promising approach for privacy protection when analysing data distributed over several databases. It brings the analysis to the data and not the data to the analysis. The analyst receives anonymous summary statistics which are combined to a aggregated result. We are interested to calculate the AUC of a prediction score based on a distributed approach without getting to know the data of involved individual subjects distributed over different databases. We use DataSHIELD as the technology to carry out distributed analyses and use a newly developed algorithms to perform the validation of the prediction score. Calibration can easily be implemented in the distributed setting. But, discrimination represented by a respective ROC curve and its AUC is challenging. We base our approach on the ROC-GLM algorithm as well as on ideas of differential privacy. The proposed algorithms are evaluated in a simulation study. A real-word application is described: The audit use case of DIFUTURE (Medical Informatics Initiative) with the goal to validate a treatment prediction rule of patients with newly diagnosed multiple sclerosis.

Negatively charged boron vacancies ($V_B^-$) in hexagonal boron nitride (hBN) exhibit a broad emission spectrum due to strong electron-phonon coupling and Jahn-Teller mixing of electronic states. As such, the direct measurement of zero-phonon line (ZPL) of $V_B^-$ has remained elusive. Here, we measure the room-temperature ZPL wavelength to be $773\pm2$ nm by coupling the hBN layer to the high-Q nanobeam cavity. As the wavelength of cavity mode is tuned, we observe a pronounced intensity resonance, indicating the coupling to $V_B^-$. Our observations are consistent with the spatial redistribution of $V_B^-$ emission. Spatially resolved measurements show a clear Purcell effect maximum at the midpoint of the nanobeam, in accord with the optical field distribution of the cavity mode. Our results are in good agreement with theoretical calculations, opening the way to using $V_B^-$ as cavity spin-photon interfaces.

This document includes seven technical appendices that accompany the main report, "Trade Adjustment Assistance Community College and Career Training Grant Program: Round 4 Early Outcomes Study Report." The appendices include: (1) Baseline Data; (2) Follow-up Survey Data; (3) Unemployment Insurance Wage Data; (4) Service Impacts; (5) Methodology for Estimating Outcomes by Program; (6) Implementation Data Collection; and (7) Expanded Results. [This document was produced in partnership with Capital Research Corporation. For the main report, see ED610190.]

Componentwise boosting (CWB), also known as model-based boosting, is a variant of gradient boosting that builds on additive models as base learners to ensure interpretability. CWB is thus often used in research areas where models are employed as tools to explain relationships in data. One downside of CWB is its computational complexity in terms of memory and runtime. In this paper, we propose two techniques to overcome these issues without losing the properties of CWB: feature discretization of numerical features and incorporating Nesterov momentum into functional gradient descent. As the latter can be prone to early overfitting, we also propose a hybrid approach that prevents a possibly diverging gradient descent routine while ensuring faster convergence. We perform extensive benchmarks on multiple simulated and real-world data sets to demonstrate the improvements in runtime and memory consumption while maintaining state-of-the-art estimation and prediction performance.

In practice, machine learning (ML) workflows require various different steps, from data preprocessing, missing value imputation, model selection, to model tuning as well as model evaluation. Many of these steps rely on human ML experts. AutoML - the field of automating these ML pipelines - tries to help practitioners to apply ML off-the-shelf without any expert knowledge. Most modern AutoML systems like auto-sklearn, H20-AutoML or TPOT aim for high predictive performance, thereby generating ensembles that consist almost exclusively of black-box models. This, in turn, makes the interpretation for the layperson more intricate and adds another layer of opacity for users. We propose an AutoML system that constructs an interpretable additive model that can be fitted using a highly scalable componentwise boosting algorithm. Our system provides tools for easy model interpretation such as visualizing partial effects and pairwise interactions, allows for a straightforward calculation of feature importance, and gives insights into the required model complexity to fit the given task. We introduce the general framework and outline its implementation autocompboost. To demonstrate the frameworks efficacy, we compare autocompboost to other existing systems based on the OpenML AutoML-Benchmark. Despite its restriction to an interpretable model space, our system is competitive in terms of predictive performance on most data sets while being more user-friendly and transparent., Comment: 6 pages, 4 figures, ECML-PKDD Workshop on Automating Data Science 2021