arXiv subjects

Patrick Huber

Publications and source records attributed to Patrick Huber.

At least 55 records · Page 3Linked to original sources

arXiv2023

FLEE-GNN: A Federated Learning System for Edge-Enhanced Graph Neural Network in Analyzing Geospatial Resilience of Multicommodity Food Flows

Understanding and measuring the resilience of food supply networks is a global imperative to tackle increasing food insecurity. However, the complexity of these networks, with their multidimensional interactions and decisions, presents significant challenges. This paper proposes FLEE-GNN, a novel Federated Learning System for Edge-Enhanced Graph Neural Network, designed to overcome these challenges and enhance the analysis of geospatial resilience of multicommodity food flow network, which is one type of spatial networks. FLEE-GNN addresses the limitations of current methodologies, such as entropy-based methods, in terms of generalizability, scalability, and data privacy. It combines the robustness and adaptability of graph neural networks with the privacy-conscious and decentralized aspects of federated learning on food supply network resilience analysis across geographical regions. This paper also discusses FLEE-GNN's innovative data generation techniques, experimental designs, and future directions for improvement. The results show the advancements of this approach to quantifying the resilience of multicommodity food flow networks, contributing to efforts towards ensuring global food security using AI methods. The developed FLEE-GNN has the potential to be applied in other spatial networks with spatially heterogeneous sub-network distributions.

cs.LG↗

arXiv2023

Discourse Structure Extraction from Pre-Trained and Fine-Tuned Language Models in Dialogues

Discourse processing suffers from data sparsity, especially for dialogues. As a result, we explore approaches to build discourse structures for dialogues, based on attention matrices from Pre-trained Language Models (PLMs). We investigate multiple tasks for fine-tuning and show that the dialogue-tailored Sentence Ordering task performs best. To locate and exploit discourse information in PLMs, we propose an unsupervised and a semi-supervised method. Our proposals achieve encouraging results on the STAC corpus, with F1 scores of 57.2 and 59.3 for unsupervised and semi-supervised methods, respectively. When restricted to projective trees, our scores improved to 63.3 and 68.1.

cs.CL↗

arXiv2023

Mineral Detection of Neutrinos and Dark Matter. A Whitepaper

Minerals are solid state nuclear track detectors - nuclear recoils in a mineral leave latent damage to the crystal structure. Depending on the mineral and its temperature, the damage features are retained in the material from minutes (in low-melting point materials such as salts at a few hundred degrees C) to timescales much larger than the 4.5 Gyr-age of the Solar System (in refractory materials at room temperature). The damage features from the $O(50)$ MeV fission fragments left by spontaneous fission of $^{238}$U and other heavy unstable isotopes have long been used for fission track dating of geological samples. Laboratory studies have demonstrated the readout of defects caused by nuclear recoils with energies as small as $O(1)$ keV. This whitepaper discusses a wide range of possible applications of minerals as detectors for $E_R \gtrsim O(1)$ keV nuclear recoils: Using natural minerals, one could use the damage features accumulated over $O(10)$ Myr$-O(1)$ Gyr to measure astrophysical neutrino fluxes (from the Sun, supernovae, or cosmic rays interacting with the atmosphere) as well as search for Dark Matter. Using signals accumulated over months to few-years timescales in laboratory-manufactured minerals, one could measure reactor neutrinos or use them as Dark Matter detectors, potentially with directional sensitivity. Research groups in Europe, Asia, and America have started developing microscopy techniques to read out the $O(1) - O(100)$ nm damage features in crystals left by $O(0.1) - O(100)$ keV nuclear recoils. We report on the status and plans of these programs. The research program towards the realization of such detectors is highly interdisciplinary, combining geoscience, material science, applied and fundamental physics with techniques from quantum information and Artificial Intelligence.

astro-ph.IM↗

arXiv2023

Testing the Gallium Anomaly

We study the online detection by gallium capture of mono-energetic neutrinos produced by a $^{51}$Cr radioactive source in a scintillation experiment. We find that cerium-doped gadolinium aluminum gallium garnet (GAGG) is a suitable scintillator which contains about 21% of gallium per weight and has a high mass density and light yield. Combined with a highly efficient light detection system this allows tagging of the subsequent germanium decay and thus a clean distinction of gallium capture and elastic neutrino electron scattering events. With 1.5 tons of scintillator and 10 source runs of 3.4MCi, each, we obtain about 760 gallium capture events with a purity of 85% and 680,000 neutrino electron scattering events, where the latter provide a precise normalization independent of any nuclear physics. This configuration would allow to test the gallium anomaly at more than $5σ$ in an independent way.

hep-ph↗

arXiv2022

Wafer-Scale Fabrication of Hierarchically Porous Silicon and Silica Glass by Active Nanoparticle-Assisted Chemical Etching and Pseudomorphic Thermal Oxidation

Many biological materials exhibit a multiscale porosity with small, mostly nanoscale pores as well as large, macroscopic capillaries to simultaneously achieve optimized mass transport capabilities and lightweight structures with large inner surfaces. Realizing such a hierarchical porosity in artificial materials necessitates often sophisticated and expensive top-down processing that limits scalability. Here we present an approach that combines self-organized porosity based on metal-assisted chemical etching (MACE) with photolithographically induced macroporosity for the synthesis of single-crystalline silicon with a bimodal pore-size distribution, i.e., hexagonally arranged cylindrical macropores with 1 micrometer diameter separated by walls that are traversed by mesopores 60 nm across. The MACE process is mainly guided by a metal-catalyzed reduction-oxidation reaction, where silver nanoparticles (AgNPs) serve as the catalyst. In this process, the AgNPs act as self-propelled particles that are constantly removing silicon along their trajectories. High-resolution X-ray imaging and electron tomography reveal a resulting large open porosity and inner surface for potential applications in high-performance energy storage, harvesting and conversion or for on-chip sensorics and actuorics. Finally, the hierarchically porous silicon membranes can be transformed structure-conserving by thermal oxidation into hierarchically porous amorphous silica, a material that could be of particular interest for opto-fluidic and (bio-)photonic applications due to its multiscale artificial vascularization.

cond-mat.mtrl-sci↗

arXiv2022

Snowmass Neutrino Frontier Report

This report summarizes the current status of neutrino physics and the broad and exciting future prospects identified for the Neutrino Frontier as part of the 2021 Snowmass Process.

hep-ex↗

arXiv2022

How nanoporous silicon-polypyrrole hybrids flex their muscles in aqueous electrolytes: In operando high-resolution x-ray diffraction and electron tomography-based micromechanical computer simulations

Macroscopic strain experiments revealed that Si crystals traversed by parallel, channel-like nanopores functionalized with the muscle polymer polypyrrole exhibit large and reversible electrochemo-mechanical actuation in aqueous electrolytes. On the microscopical level this system still bears open questions, as to how the electrochemical expansion and contraction of PPy acts on to np-Si pore walls and how the collective motorics of the pore array emerges from the single-nanopore behavior. An analysis of in operando X-ray diffraction experiments with micromechanical finite element simulations, based on a 3D reconstruction of the nanoporous medium by TEM tomography, shows that the in-plane mechanical response is dominantly isotropic despite the anisotropic elasticity of the single crystalline host matrix. However, the structural anisotropy originating from the parallel alignment of the nanopores lead to significant differences between the in- and out-of-plane electromechanical response. This response is not describable by a simple 2D arrangement of parallel cylindrical channels. Rather, the simulations highlight that the dendritic shape of the Si pore walls, including pore connections between the main channels, cause complex, inhomogeneous stress-strain fields in the crystalline host. Time-dependent X-ray scattering on the dynamics of the actuator properties hint towards the importance of diffusion limitations, plastic deformation and creep in the nanoconfined polymer upon (counter-)ion adsorption and desorption, the very pore-scale processes causing the macroscopic electroactuation. From a more general perspective, our study demonstrates that the combination of TEM tomography-based micromechanical modeling with high-resolution X-ray scattering experiments provides a powerful approach for in operando analysis of nanoporous composites from the single-nanopore up to the porous-medium scale.

cond-mat.mtrl-sci↗

arXiv2022

Unsupervised Inference of Data-Driven Discourse Structures using a Tree Auto-Encoder

With a growing need for robust and general discourse structures in many downstream tasks and real-world applications, the current lack of high-quality, high-quantity discourse trees poses a severe shortcoming. In order the alleviate this limitation, we propose a new strategy to generate tree structures in a task-agnostic, unsupervised fashion by extending a latent tree induction framework with an auto-encoding objective. The proposed approach can be applied to any tree-structured objective, such as syntactic parsing, discourse parsing and others. However, due to the especially difficult annotation process to generate discourse trees, we initially develop such method to complement task-specific models in generating much larger and more diverse discourse treebanks.

cs.CL↗

arXiv2022

Towards Domain-Independent Supervised Discourse Parsing Through Gradient Boosting

Discourse analysis and discourse parsing have shown great impact on many important problems in the field of Natural Language Processing (NLP). Given the direct impact of discourse annotations on model performance and interpretability, robustly extracting discourse structures from arbitrary documents is a key task to further improve computational models in NLP. To this end, we present a new, supervised paradigm directly tackling the domain adaptation issue in discourse parsing. Specifically, we introduce the first fully supervised discourse parser designed to alleviate the domain dependency through a staged model of weak classifiers by introducing the gradient boosting framework.

cs.CL↗

arXiv2022

Large Discourse Treebanks from Scalable Distant Supervision

Discourse parsing is an essential upstream task in Natural Language Processing with strong implications for many real-world applications. Despite its widely recognized role, most recent discourse parsers (and consequently downstream tasks) still rely on small-scale human-annotated discourse treebanks, trying to infer general-purpose discourse structures from very limited data in a few narrow domains. To overcome this dire situation and allow discourse parsers to be trained on larger, more diverse and domain-independent datasets, we propose a framework to generate "silver-standard" discourse trees from distant supervision on the auxiliary task of sentiment analysis.

cs.CL↗

arXiv2022

Improving Topic Segmentation by Injecting Discourse Dependencies

Recent neural supervised topic segmentation models achieve distinguished superior effectiveness over unsupervised methods, with the availability of large-scale training corpora sampled from Wikipedia. These models may, however, suffer from limited robustness and transferability caused by exploiting simple linguistic cues for prediction, but overlooking more important inter-sentential topical consistency. To address this issue, we present a discourse-aware neural topic segmentation model with the injection of above-sentence discourse dependency structures to encourage the model make topic boundary prediction based more on the topical consistency between sentences. Our empirical study on English evaluation datasets shows that injecting above-sentence discourse structures to a neural topic segmenter with our proposed strategy can substantially improve its performances on intra-domain and out-of-domain data, with little increase of model's complexity.

cs.CL↗

arXiv2022

Acoustically Induced Giant Synthetic Hall Voltages in Graphene

Any departure from graphene's flatness leads to the emergence of artificial gauge fields that act on the motion of the Dirac fermions through an associated pseudomagnetic field. Here, we demonstrate the tunability of strong gauge fields in non-local experiments using a large planar graphene sheet that conforms to the deformation of a piezoelectric layer by a surface acoustic wave. The acoustic wave induces a longitudinal and a giant synthetic Hall voltage in the absence of external magnetic fields. The superposition of a synthetic Hall potential and a conventional Hall voltage can annihilate the sample's transversal potential at large external magnetic fields. Surface acoustic waves thus provide a promising and facile avenue for the exploit of gauge fields in large planar graphene systems.

cond-mat.mes-hall↗

arXiv2022

Multiple glassy dynamics of a homologous series of triphenylene-based columnar liquid crystals -- A study by broadband dielectric spectroscopy and advanced calorimetry

Hexakis(n-alkyloxy)triphenylene) (HATn) consisting of an aromatic triphenylene core and alkyl side chains are model discotic liquid crystal (DLC) systems forming a columnar mesophase. In the mesophase, the molecules of HATn self-assemble in columns, which has one-dimensional high charge carrier mobility along the columns. Here, a homologous series of HATn with different length of the alkyl chain (n=5,6,8,10,12) is investigated using differential scanning calorimetry (DSC), broadband dielectric spectroscopy (BDS) and advanced calorimetric techniques including fast scanning calorimetry (FSC) and specific heat spectroscopy (SHS). The investigation of the phase behavior was done utilizing DSC experiments and the influence of the alkyl chain length on the phase behavior was revealed. By the dielectric investigations probing the molecular mobility, a $γ$-relaxation due to localized fluctuations as well as two glassy dynamics the $α$ core and $α$ alkyl relaxation were observed in the temperature range of the plastic crystalline phase. Moreover, the observed glassy dynamics were further studied employing advanced calorimetry. All observed relaxation processes are attributed to the possible specific molecular fluctuations and discussed in detail. From the results a transition at around n=8 from a rigid constrained (n=5,6) to a softer system (n=10,12) was revealed with increasing alkyl chain length. A counterbalance of two competing effects of a polyethylene like behavior of the alkyl chains in the intercolumnar domains and self-organized confinement is discussed in the context of a hindered glass transition.

cond-mat.soft↗

arXiv2022

On the Issue of Textured Crystallization of Ba(NO$_3$)$_2$ in Mesoporous SiO$_2$: Raman Spectroscopy and Lattice Dynamics Analysis

The lattice dynamics of preferentially aligned nanocrystals formed upon drying of aqueous Ba(NO$_3$)$_2$ solutions in a mesoporous silica glass traversed by tubular pores of approximately 12 nm are explored by Raman scattering. To interpret the experiments on the confined nanocrystals polarized Raman spectra of bulk single crystals and X-ray diffraction experiments are also performed. Since a cubic symmetry is inherent to Ba(NO$_3$)$_2$, a special Raman scattering geometry was utilized to separate the phonon modes of A$_g$ and E$_g$ species. Combining group-theory analysis and \textit{ab initio} lattice dynamics calculations a full interpretation of all Raman lines of the bulk single crystal is achieved. Apart from a small confinement-induced line broadening, the peak positions and normalized peak intensities of the Raman spectra of the nanoconfined and macroscopic crystals are identical. Interestingly, the Raman scattering experiment indicates the existence of comparatively large, $\sim$10-20 $μ$m, single-crystalline regions of Ba(NO$_3$)$_2$ embedded in the porous host, near three orders of magnitude larger than the average size of single nanopores. This is contrast to the initial assumption of non-interconnected pores. It rather indicates an inter-pore propagation of the crystallization front, presumably via microporosity in the pore walls.

cond-mat.mtrl-sci↗

arXiv2022

CCQA: A New Web-Scale Question Answering Dataset for Model Pre-Training

With the rise of large-scale pre-trained language models, open-domain question-answering (ODQA) has become an important research topic in NLP. Based on the popular pre-training fine-tuning approach, we posit that an additional in-domain pre-training stage using a large-scale, natural, and diverse question-answering (QA) dataset can be beneficial for ODQA. Consequently, we propose a novel QA dataset based on the Common Crawl project in this paper. Using the readily available schema.org annotation, we extract around 130 million multilingual question-answer pairs, including about 60 million English data-points. With this previously unseen number of natural QA pairs, we pre-train popular language models to show the potential of large-scale in-domain pre-training for the task of question-answering. In our experiments, we find that pre-training question-answering models on our Common Crawl Question Answering dataset (CCQA) achieves promising results in zero-shot, low resource and fine-tuned settings across multiple tasks, models and benchmarks.

cs.CL↗

arXiv2022

Towards Understanding Large-Scale Discourse Structures in Pre-Trained and Fine-Tuned Language Models

With a growing number of BERTology work analyzing different components of pre-trained language models, we extend this line of research through an in-depth analysis of discourse information in pre-trained and fine-tuned language models. We move beyond prior work along three dimensions: First, we describe a novel approach to infer discourse structures from arbitrarily long documents. Second, we propose a new type of analysis to explore where and how accurately intrinsic discourse is captured in the BERT and BART models. Finally, we assess how similar the generated structures are to a variety of baselines as well as their distribution within and between models.

cs.CL↗

arXiv2022

The Physics Case for a Neutrino Factory

Neutrino factories, neutrino beams produced in the decay of a muon or antimuon beam inside a storage ring, yield cleaner, richer, and more flexible neutrino beams relative to super-beams. We explore the physics case for this type of beam both for standard oscillation as well as new physics searches and present some machine options. We argue that there is a rich program beyond what the current neutrino program can cover and a string synergy with the muon collider program.

hep-ph↗

arXiv2022

Passive low energy nuclear recoil detection with color centers -- PALEOCCENE

The PALEOCCENE concept offers the potential for room-temperature, passive and robust detectors in the gram to kilogram range for the detection of low-energy nuclear recoil events. Nuclear recoil events can be caused by neutron scattering, coherent elastic neutrino nucleus scattering (CEvNS) or dark matter scattering and therefore, PALEOCCENE could find applications in all three areas. In this white paper we present current and planned R&D efforts to study the feasibility of this technique.

hep-ex↗