Showing 1–50 of 77 results for author: Weber, N

MyoGestic: EMG Interfacing Framework for Decoding Multiple Spared Degrees of Freedom of the Hand in Individuals with Neural Lesions

Authors: Raul C. Sîmpetru, Dominik I. Braun, Arndt U. Simon, Michael März, Vlad Cnejevici, Daniela Souza de Oliveira, Nico Weber, Jonas Walter, Jörg Franke, Daniel Höglinger, Cosima Prahm, Matthias Ponfick, Alessandro Del Vecchio

Abstract: Restoring limb motor function in individuals with spinal cord injury (SCI), stroke, or amputation remains a critical challenge, one which affects millions worldwide. Recent studies show through surface electromyography (EMG) that spared motor neurons can still be voluntarily controlled, even without visible limb movement . These signals can be decoded and used for motor intent estimation; however,… ▽ More Restoring limb motor function in individuals with spinal cord injury (SCI), stroke, or amputation remains a critical challenge, one which affects millions worldwide. Recent studies show through surface electromyography (EMG) that spared motor neurons can still be voluntarily controlled, even without visible limb movement . These signals can be decoded and used for motor intent estimation; however, current wearable solutions lack the necessary hardware and software for intuitive interfacing of the spared degrees of freedom after neural injuries. To address these limitations, we developed a wireless, high-density EMG bracelet, coupled with a novel software framework, MyoGestic. Our system allows rapid and tailored adaptability of machine learning models to the needs of the users, facilitating real-time decoding of multiple spared distinctive degrees of freedom. In our study, we successfully decoded the motor intent from two participants with SCI, two with spinal stroke , and three amputees in real-time, achieving several controllable degrees of freedom within minutes after wearing the EMG bracelet. We provide a proof-of-concept that these decoded signals can be used to control a digitally rendered hand, a wearable orthosis, a prosthesis, or a 2D cursor. Our framework promotes a participant-centered approach, allowing immediate feedback integration, thus enhancing the iterative development of myocontrol algorithms. The proposed open-source software framework, MyoGestic, allows researchers and patients to focus on the augmentation and training of the spared degrees of freedom after neural lesions, thus potentially bridging the gap between research and clinical application and advancing the development of intuitive EMG interfaces for diverse neural lesions. △ Less

Submitted 14 August, 2024; originally announced August 2024.

Comments: 23 pages, 8 figures

ACM Class: H.5.2; J.3; I.5.4; D.2.13

Neutron imaging of high-temperature Na-Zn Cells: implications for cell design and fabrication

Authors: William Nash, Martins Sarma, Tobias Lappan, Pavel Trtik, Catherine K. W. Solem, Zhaohui Wang, Alberto Beltrán, Norbert Weber, Tom Weier

Abstract: Electrochemical cells employing Sodium (Na) and Zinc (Zn) electrodes and a chloride salt electrolyte have been imaged by neutron radiography during cycling. The use of such abundant raw materials confers a very low energy-normalised cost to the Na-Zn system, but its operation requires them to be entirely molten, and therefore to be operated at 600 °C. To suppress the self-discharge that results fr… ▽ More Electrochemical cells employing Sodium (Na) and Zinc (Zn) electrodes and a chloride salt electrolyte have been imaged by neutron radiography during cycling. The use of such abundant raw materials confers a very low energy-normalised cost to the Na-Zn system, but its operation requires them to be entirely molten, and therefore to be operated at 600 °C. To suppress the self-discharge that results from this all-molten configuration, porous ceramic diaphragms are used to partition the electrolyte and thereby impede the movement of the Zn2+ ions responsible towards the Na electrode. Neutron images reveal large gas bubbles trapped beneath these diaphragms, formed during the cell fabrication process due to the large volume change that accompanies melting/solidifying of the electrolyte. Cycling data confirm that these bubbles interfere with cell operation by substantially increasing ohmic resistance. They indicate the need for either a new diaphragm design, or a cell fabrication process that prevents their formation in the first instance. △ Less

Submitted 24 July, 2024; originally announced July 2024.

Comments: 8 pages, 3 figures

MSC Class: 76D99

Laboratory-Scale AI: Open-Weight Models are Competitive with ChatGPT Even in Low-Resource Settings

Authors: Robert Wolfe, Isaac Slaughter, Bin Han, Bingbing Wen, Yiwei Yang, Lucas Rosenblatt, Bernease Herman, Eva Brown, Zening Qu, Nic Weber, Bill Howe

Abstract: The rapid proliferation of generative AI has raised questions about the competitiveness of lower-parameter, locally tunable, open-weight models relative to high-parameter, API-guarded, closed-weight models in terms of performance, domain adaptation, cost, and generalization. Centering under-resourced yet risk-intolerant settings in government, research, and healthcare, we see for-profit closed-wei… ▽ More The rapid proliferation of generative AI has raised questions about the competitiveness of lower-parameter, locally tunable, open-weight models relative to high-parameter, API-guarded, closed-weight models in terms of performance, domain adaptation, cost, and generalization. Centering under-resourced yet risk-intolerant settings in government, research, and healthcare, we see for-profit closed-weight models as incompatible with requirements for transparency, privacy, adaptability, and standards of evidence. Yet the performance penalty in using open-weight models, especially in low-data and low-resource settings, is unclear. We assess the feasibility of using smaller, open-weight models to replace GPT-4-Turbo in zero-shot, few-shot, and fine-tuned regimes, assuming access to only a single, low-cost GPU. We assess value-sensitive issues around bias, privacy, and abstention on three additional tasks relevant to those topics. We find that with relatively low effort, very low absolute monetary cost, and relatively little data for fine-tuning, small open-weight models can achieve competitive performance in domain-adapted tasks without sacrificing generality. We then run experiments considering practical issues in bias, privacy, and hallucination risk, finding that open models offer several benefits over closed models. We intend this work as a case study in understanding the opportunity cost of reproducibility and transparency over for-profit state-of-the-art zero shot performance, finding this cost to be marginal under realistic settings. △ Less

Submitted 27 May, 2024; originally announced May 2024.

Comments: Accepted at the ACM Conference on Fairness, Accountability, and Transparency (FAccT) 2024

Higher-Rank Irreducible Cartesian Tensors for Equivariant Message Passing

Authors: Viktor Zaverkin, Francesco Alesiani, Takashi Maruyama, Federico Errica, Henrik Christiansen, Makoto Takamoto, Nicolas Weber, Mathias Niepert

Abstract: The ability to perform fast and accurate atomistic simulations is crucial for advancing the chemical sciences. By learning from high-quality data, machine-learned interatomic potentials achieve accuracy on par with ab initio and first-principles methods at a fraction of their computational cost. The success of machine-learned interatomic potentials arises from integrating inductive biases such as… ▽ More The ability to perform fast and accurate atomistic simulations is crucial for advancing the chemical sciences. By learning from high-quality data, machine-learned interatomic potentials achieve accuracy on par with ab initio and first-principles methods at a fraction of their computational cost. The success of machine-learned interatomic potentials arises from integrating inductive biases such as equivariance to group actions on an atomic system, e.g., equivariance to rotations and reflections. In particular, the field has notably advanced with the emergence of equivariant message-passing architectures. Most of these models represent an atomic system using spherical tensors, tensor products of which require complicated numerical coefficients and can be computationally demanding. This work introduces higher-rank irreducible Cartesian tensors as an alternative to spherical tensors, addressing the above limitations. We integrate irreducible Cartesian tensor products into message-passing neural networks and prove the equivariance of the resulting layers. Through empirical evaluations on various benchmark data sets, we consistently observe on-par or better performance than that of state-of-the-art spherical models. △ Less

Submitted 23 May, 2024; originally announced May 2024.

Yamdb: easily accessible thermophysical properties of liquid metals and molten salts

Authors: Tom Weier, William Nash, Paolo Personnettaz, Norbert Weber

Abstract: Yamdb (Yet another materials data base) addresses the need to provide thermophysical properties of liquid metals and molten salts in an easily accessible manner. Mathematical relations describing material properties - usually determined by experiment - are taken from the literature. Equations and their coefficients are stored separately. The former can be implemented in any programming language (P… ▽ More Yamdb (Yet another materials data base) addresses the need to provide thermophysical properties of liquid metals and molten salts in an easily accessible manner. Mathematical relations describing material properties - usually determined by experiment - are taken from the literature. Equations and their coefficients are stored separately. The former can be implemented in any programming language (Python and Go in this case) and the latter are kept in YAML files together with additional information (source, temperature range, composition, accuracy if available, etc). △ Less

Submitted 28 January, 2024; originally announced February 2024.

Comments: 22 pages, 5 figures

MSC Class: 76A99 ACM Class: J.2

On the use of associative memory in Hopfield networks designed to solve propositional satisfiability problems

Authors: Natalya Weber, Werner Koch, Ozan Erdem, Tom Froese

Abstract: Hopfield networks are an attractive choice for solving many types of computational problems because they provide a biologically plausible mechanism. The Self-Optimization (SO) model adds to the Hopfield network by using a biologically founded Hebbian learning rule, in combination with repeated network resets to arbitrary initial states, for optimizing its own behavior towards some desirable goal s… ▽ More Hopfield networks are an attractive choice for solving many types of computational problems because they provide a biologically plausible mechanism. The Self-Optimization (SO) model adds to the Hopfield network by using a biologically founded Hebbian learning rule, in combination with repeated network resets to arbitrary initial states, for optimizing its own behavior towards some desirable goal state encoded in the network. In order to better understand that process, we demonstrate first that the SO model can solve concrete combinatorial problems in SAT form, using two examples of the Liars problem and the map coloring problem. In addition, we show how under some conditions critical information might get lost forever with the learned network producing seemingly optimal solutions that are in fact inappropriate for the problem it was tasked to solve. What appears to be an undesirable side-effect of the SO model, can provide insight into its process for solving intractable problems. △ Less

Submitted 4 March, 2024; v1 submitted 31 July, 2023; originally announced July 2023.

Comments: 7 pages, 4 figures

Journal ref: 2023 IEEE Symposium Series on Computational Intelligence (SSCI) 1352-1358

A continuous multiphase model for liquid metal batteries

Authors: Omar E. Godinez-Brizuela, Carolina Duczek, Norbert Weber, Kristian E. Einarsrud

Abstract: Liquid metal batteries (LMBs) are a promising alternative for large-scale stationary energy storage for renewable applications. Using high-abundance electrode materials such as Sodium and Zinc is highly desirable due to their low cost and excellent cell potential. LMBs undergo multiple complex mass transport dynamics and as a result, their operation limits and other critical parameters are not ful… ▽ More Liquid metal batteries (LMBs) are a promising alternative for large-scale stationary energy storage for renewable applications. Using high-abundance electrode materials such as Sodium and Zinc is highly desirable due to their low cost and excellent cell potential. LMBs undergo multiple complex mass transport dynamics and as a result, their operation limits and other critical parameters are not fully understood yet. In this work, a multiphase numerical model was developed to resolve electrode and electrolyte components in 1D and simulate the discharge process of a Na-Zn battery including the interfacial displacement of the molten metal electrodes. The variation in electrolyte composition was predicted throughout the process, including the species distribution and its effect on the cell conductivity and capacity. Volume change and species redistribution were found to be important in predicting the maximum theoretical capacity of the cell when neglecting convective phenomena. △ Less

Submitted 8 June, 2023; originally announced June 2023.

Ferroelastic twin angles at the surface of CaTiO$_\mathrm{3}$ quantified by PhotoEmission Electron Microscopy

Authors: G. Magagnin, C. Lubin, M. Escher, N. Weber, L. Tortech, N. Barrett

Abstract: We use photoemission electron microscopy to measure the ferroelastic twin wall angles at the surface of CaTiO$_\mathrm{3}$(001) and deduce the strain ordering. We analyze the angular dependence of the photoelectron emission from different domain surfaces, each with their own characteristic tilt angle in the factory roof-like topography. By considering the surface topography as a field perturbation… ▽ More We use photoemission electron microscopy to measure the ferroelastic twin wall angles at the surface of CaTiO$_\mathrm{3}$(001) and deduce the strain ordering. We analyze the angular dependence of the photoelectron emission from different domain surfaces, each with their own characteristic tilt angle in the factory roof-like topography. By considering the surface topography as a field perturbation, the offset in the photoemission threshold can be directly related to the tilt angles. With knowledge of the symmetry allowed twin walls we quantify twin topography between 179.1° to 180.8°. △ Less

Submitted 17 April, 2023; originally announced April 2023.

Soft-Search: Two Datasets to Study the Identification and Production of Research Software

Authors: Eva Maxfield Brown, Lindsey Schwartz, Richard Lewei Huang, Nicholas Weber

Abstract: Software is an important tool for scholarly work, but software produced for research is in many cases not easily identifiable or discoverable. A potential first step in linking research and software is software identification. In this paper we present two datasets to study the identification and production of research software. The first dataset contains almost 1000 human labeled annotations of so… ▽ More Software is an important tool for scholarly work, but software produced for research is in many cases not easily identifiable or discoverable. A potential first step in linking research and software is software identification. In this paper we present two datasets to study the identification and production of research software. The first dataset contains almost 1000 human labeled annotations of software production from National Science Foundation (NSF) awarded research projects. We use this dataset to train models that predict software production. Our second dataset is created by applying the trained predictive models across the abstracts and project outcomes reports for all NSF funded projects between the years of 2010 and 2023. The result is an inferred dataset of software production for over 150,000 NSF awards. We release the Soft-Search dataset to aid in identifying and understanding research software production: https://github.com/si2-urssi/eager △ Less

Submitted 27 February, 2023; originally announced February 2023.

Simulation of potential and species distribution in a Li||Bi liquid metal battery using coupled meshes

Authors: Carolina Duczek, Norbert Weber, Omar E. Godinez-Brizuela, Tom Weier

Abstract: In this work a 1D finite volume based model using coupled meshes is introduced to capture potential and species distribution throughout the discharge process in a lithium bismuth liquid metal battery while neglecting hydrodynamic effects, focusing on the electrochemical properties of the cell and the mass transport in electrolyte and cathode. Interface reactions in the electrical double layer are… ▽ More In this work a 1D finite volume based model using coupled meshes is introduced to capture potential and species distribution throughout the discharge process in a lithium bismuth liquid metal battery while neglecting hydrodynamic effects, focusing on the electrochemical properties of the cell and the mass transport in electrolyte and cathode. Interface reactions in the electrical double layer are considered through the introduction of a discrete jump of the potential modelled as periodic boundary condition to resolve interfacial discontinuities in the cell potential. A balanced-force like approach is implemented to ensure consistent calculation at the interface level. It is found that mass transport and concentration gradients have a significant effect on the cell overpotentials and thus on cell performance and cell voltage. By quantifying overvoltages in the Li Bi cell with a mixed cation electrolyte, it is possible to show that diffusion and migration current density could have counteractive effects on the cell voltage. Furthermore, the simulated limiting current density is observed to be much lower than experimentally measured, which can be attributed to convective effects in the electrolyte that need to be addressed in future simulations. The solver is based on the open source library OpenFOAM and thoroughly verified against the equivalent system COMSOL multiphysics and further validated with experimental results. It is openly available at https://doi.org/10.14278/rodare.2313. △ Less

Submitted 26 May, 2023; v1 submitted 22 November, 2022; originally announced November 2022.

Journal ref: Electrochimica Acta 437 (2023) 141413

Scaling up the self-optimization model by means of on-the-fly computation of weights

Authors: Natalya Weber, Werner Koch, Tom Froese

Abstract: The Self-Optimization (SO) model is a useful computational model for investigating self-organization in "soft" Artificial life (ALife) as it has been shown to be general enough to model various complex adaptive systems. So far, existing work has been done on relatively small network sizes, precluding the investigation of novel phenomena that might emerge from the complexity arising from large numb… ▽ More The Self-Optimization (SO) model is a useful computational model for investigating self-organization in "soft" Artificial life (ALife) as it has been shown to be general enough to model various complex adaptive systems. So far, existing work has been done on relatively small network sizes, precluding the investigation of novel phenomena that might emerge from the complexity arising from large numbers of nodes interacting in interconnected networks. This work introduces a novel implementation of the SO model that scales as $\mathcal{O}\left(N^{2}\right)$ with respect to the number of nodes $N$, and demonstrates the applicability of the SO model to networks with system sizes several orders of magnitude higher than previously was investigated. Removing the prohibitive computational cost of the naive $\mathcal{O}\left(N^{3}\right)$ algorithm, our on-the-fly computation paves the way for investigating substantially larger system sizes, allowing for more variety and complexity in future studies. △ Less

Submitted 3 November, 2022; originally announced November 2022.

Comments: 7 pages, 7 figures

Journal ref: 2022 IEEE Symposium Series on Computational Intelligence (SSCI), Singapore, Singapore, 2022, pp. 1276-1282

Nanoscale friction controlled by top layer thickness in [LaMnO$_{3}$]$_{m}$/[SrMnO$_{3}$]$_{n}$ superlattices

Authors: Niklas A. Weber, Miru Lee, Florian Schönewald, Leonard Schüler, Vasily Moshnyaga, Matthias Krüger, Cynthia A. Volkert

Abstract: We conducted lateral force microscopy measurements on seven [LaMnO$_{3}$]$_{m}$/[SrMnO$_{3}$]$_{n}$ superlattices with varied layer thicknesses. We observe that the friction forces and the friction coefficients initially increase with increasing LaMnO3 top layer thickness, followed by saturation when the top layer thickness exceeds a few nanometers. These observations clearly demonstrate that slid… ▽ More We conducted lateral force microscopy measurements on seven [LaMnO$_{3}$]$_{m}$/[SrMnO$_{3}$]$_{n}$ superlattices with varied layer thicknesses. We observe that the friction forces and the friction coefficients initially increase with increasing LaMnO3 top layer thickness, followed by saturation when the top layer thickness exceeds a few nanometers. These observations clearly demonstrate that sliding friction is affected by sub-surface material properties to a depth of several nanometers and is not just determined by dynamics in the contact interface. We argue that the sub-surface dissipated energy is governed by damping in the elastically strained volume below the AFM tip, an effect which we estimate via thermoelasticity. The absence of a correlation between friction and the thermal resistivity of our superlattices shows furthermore that high-frequency phonons and heat conduction do not play a role in determining friction. Our observations thus demonstrate that friction can be tailored by sub-surface material properties. △ Less

Submitted 18 October, 2022; originally announced October 2022.

Comments: The document contains both the manuscript and the supporting information (starting on page 22)

Extending Open Bandit Pipeline to Simulate Industry Challenges

Authors: Bram van den Akker, Niklas Weber, Felipe Moraes, Dmitri Goldenberg

Abstract: Bandit algorithms are often used in the e-commerce industry to train Machine Learning (ML) systems when pre-labeled data is unavailable. However, the industry setting poses various challenges that make implementing bandit algorithms in practice non-trivial. In this paper, we elaborate on the challenges of off-policy optimisation, delayed reward, concept drift, reward design, and business rules con… ▽ More Bandit algorithms are often used in the e-commerce industry to train Machine Learning (ML) systems when pre-labeled data is unavailable. However, the industry setting poses various challenges that make implementing bandit algorithms in practice non-trivial. In this paper, we elaborate on the challenges of off-policy optimisation, delayed reward, concept drift, reward design, and business rules constraints that practitioners at Booking.com encounter when applying bandit algorithms. Our main contributions is an extension to the Open Bandit Pipeline (OBP) framework. We provide simulation components for some of the above-mentioned challenges to provide future practitioners, researchers, and educators with a resource to address challenges encountered in the e-commerce industry. △ Less

Submitted 9 September, 2022; originally announced September 2022.

Comments: Published at the CONSEQUENCES+REVEAL '22 workshop @ Recsys 2022, Seattle, WA, USA, Sept. 22nd-23rd 2022

ACM Class: I.2.6; I.6.8

Research Software Publication Policy Case Study

Authors: Nic Weber

Abstract: Research software is increasingly recognized as a vital component of the scholarly record. Journals offer authors the opportunity to publish research software papers, but often have different requirements for how these publications should be structured and how code should be verified. In this short case study we gather data from 20 Physical Science journals to trace the frequency, quality control,… ▽ More Research software is increasingly recognized as a vital component of the scholarly record. Journals offer authors the opportunity to publish research software papers, but often have different requirements for how these publications should be structured and how code should be verified. In this short case study we gather data from 20 Physical Science journals to trace the frequency, quality control, and publishing criteria for software papers. Our goal with the case study is to provide a proof-of-concept for doing descriptive empirical work with software publication policies across numerous domains of science and engineering. In the narrative we therefore provide descriptive statistics showing how these journals differ in criteria required for archiving, linking, verifying, and documenting software as part of a formal publication. The contribution of this preliminary work is twofold: 1. We provide case study of Physical Science research software publications over time; 2. We demonstrate the use of a new survey method for analyzing research software publication policies. In our conclusion, we describe how comparative research into software publication policies can provide better criteria and requirements for an emerging software publication landscape. △ Less

Submitted 10 June, 2022; originally announced June 2022.

Fed-DART and FACT: A solution for Federated Learning in a production environment

Authors: Nico Weber, Patrick Holzer, Tania Jacob, Enislay Ramentol

Abstract: Federated Learning as a decentralized artificial intelligence (AI) solution solves a variety of problems in industrial applications. It enables a continuously self-improving AI, which can be deployed everywhere at the edge. However, bringing AI to production for generating a real business impact is a challenging task. Especially in the case of Federated Learning, expertise and resources from multi… ▽ More Federated Learning as a decentralized artificial intelligence (AI) solution solves a variety of problems in industrial applications. It enables a continuously self-improving AI, which can be deployed everywhere at the edge. However, bringing AI to production for generating a real business impact is a challenging task. Especially in the case of Federated Learning, expertise and resources from multiple domains are required to realize its full potential. Having this in mind we have developed an innovative Federated Learning framework FACT based on Fed-DART, enabling an easy and scalable deployment, helping the user to fully leverage the potential of their private and decentralized data. △ Less

Submitted 23 May, 2022; originally announced May 2022.

Ethics of Open Data

Authors: Nic Weber, Brandon Locke

Abstract: This chapter addresses emergent ethical issues in producing, using, curating, and providing services for open data. Our goal is to provide an introduction to how ethical topics in open data manifest in practical dilemmas for scholarly communications and some approaches to understanding and working through them. We begin with a brief overview of what can be thought of as three basic theories of eth… ▽ More This chapter addresses emergent ethical issues in producing, using, curating, and providing services for open data. Our goal is to provide an introduction to how ethical topics in open data manifest in practical dilemmas for scholarly communications and some approaches to understanding and working through them. We begin with a brief overview of what can be thought of as three basic theories of ethics that intersect with dilemmas in openness, accountability, transparency, and fairness in data: Virtue, Consequential, and Non-consequential ethics. We then map these kinds of ethics to the practical questions that arise in provisioning infrastructures, providing services, and supporting sustainable research in science and scholarship that depends upon open access to data. Throughout, we attempt to offer concrete examples of potential ethical dilemmas facing scholarly communication with respect to open data, and try to make clear what kinds of ethical positions are helpful to practitioners. In doing so, we hope to both clarify the ethical questions facing librarians doing practical work to support open data access, as well as situate current debates in the field with respect to these three kinds of ethics. △ Less

Submitted 20 May, 2022; originally announced May 2022.

Comments: Chapter accepted for publication in ACRL's 'Scholarly Communication and Open Culture'

SOL: Reducing the Maintenance Overhead for Integrating Hardware Support into AI Frameworks

Authors: Nicolas Weber

Abstract: The increased interest in Artificial Intelligence (AI) raised the need for highly optimized and sophisticated AI frameworks. Starting with the Lua-based Torch many frameworks have emerged over time, such as Theano, Caffe, Chainer, CNTK, MxNet, PyTorch, DL4J, or TensorFlow. All of these provide a high level scripting API that allows users to easily design neural networks and run these on various ki… ▽ More The increased interest in Artificial Intelligence (AI) raised the need for highly optimized and sophisticated AI frameworks. Starting with the Lua-based Torch many frameworks have emerged over time, such as Theano, Caffe, Chainer, CNTK, MxNet, PyTorch, DL4J, or TensorFlow. All of these provide a high level scripting API that allows users to easily design neural networks and run these on various kinds of hardware. What the user usually does not see is the high effort put into these frameworks to provide peak execution performance. While mainstream CPUs and GPUs have the "luxury" to have a wide spread user base in the open source community, less mainstream CPU, GPU or accelerator vendors need to put in a high effort to get their hardware supported by these frameworks. This includes not only the development of highly efficient compute libraries such as CUDNN, OneDNN or VEDNN but also supporting an ever growing number of simpler compute operations such as summation and multiplications. Each of these frameworks, nowadays, supports several hundred of unique operations, with tensors of various sizes, shapes and data types, which end up in thousands of compute kernels required for each device type. And the number of operations keeps increasing. That is why NEC Laboratories Europe started developing the SOL AI Optimization project already years ago, to deliver optimal performance to users while keeping the maintenance burden minimal. △ Less

Submitted 19 May, 2022; originally announced May 2022.

Friction on layered media: How deep do phonons reach?

Authors: Miru Lee, Niklas Weber, Cynthia A. Volkert, Matthias Krüger

Abstract: We theoretically study the frictional damping of a small probe object on a coated planar surface, analyzing the resulting phonon modes via a theory of viscoelasticity. Three different types of excitations are found to contribute to friction in distinct ways: traveling (3D) spherical waves, traveling (2D) surface waves, and evanescent waves. While traveling waves transport energy away from the prob… ▽ More We theoretically study the frictional damping of a small probe object on a coated planar surface, analyzing the resulting phonon modes via a theory of viscoelasticity. Three different types of excitations are found to contribute to friction in distinct ways: traveling (3D) spherical waves, traveling (2D) surface waves, and evanescent waves. While traveling waves transport energy away from the probe, determined by long range elastic properties (wavelength), evanescent waves transform energy into heat in a near-field range, characterized by the size of the probe. Thus, fundamentally different behaviors are predicted, depending on coating thickness and material properties. △ Less

Submitted 23 May, 2023; v1 submitted 2 May, 2022; originally announced May 2022.

Comments: 6 pages, 5 figures, supplementary material

Journal ref: Europhysics Letters 142 46001 (2023)

Councils in Action: Automating the Curation of Municipal Governance Data for Research

Authors: Eva Maxfield Brown, Nicholas Weber

Abstract: Large scale comparative research into municipal governance is often prohibitively difficult due to a lack of high-quality data. But, recent advances in speech-to-text algorithms and natural language processing has made it possible to more easily collect and analyze data about municipal governments. In this paper, we introduce an open-source platform, the Council Data Project (CDP), to curate novel… ▽ More Large scale comparative research into municipal governance is often prohibitively difficult due to a lack of high-quality data. But, recent advances in speech-to-text algorithms and natural language processing has made it possible to more easily collect and analyze data about municipal governments. In this paper, we introduce an open-source platform, the Council Data Project (CDP), to curate novel datasets for research into municipal governance. The contribution of this work is two-fold: 1. We demonstrate that CDP, as an infrastructure, can be used to assemble reliable comparative data on municipal governance; 2. We provide exploratory analysis of three municipalities to show how CDP data can be used to gain insight into how municipal governments perform over time. We conclude by describing future directions for research on and with CDP such as the development of machine learning models for speaker annotation, outline generation, and named entity recognition for improved linked data. △ Less

Submitted 31 August, 2022; v1 submitted 19 April, 2022; originally announced April 2022.

Comments: Keywords: public interest technology; municipal governance; data curation; computational data access; natural language processing To Be Published with 2022 ASIS&T Annual Meeting (https://www.asist.org/am22/)

Toward Unsupervised Test Scenario Extraction for Automated Driving Systems from Urban Naturalistic Road Traffic Data

Authors: Nico Weber, Christoph Thiem, Ulrich Konigorski

Abstract: Scenario-based testing is a promising approach to solve the challenge of proving the safe behavior of vehicles equipped with automated driving systems. Since an infinite number of concrete scenarios can theoretically occur in real-world road traffic, the extraction of scenarios relevant in terms of the safety-related behavior of these systems is a key aspect for their successful verification and v… ▽ More Scenario-based testing is a promising approach to solve the challenge of proving the safe behavior of vehicles equipped with automated driving systems. Since an infinite number of concrete scenarios can theoretically occur in real-world road traffic, the extraction of scenarios relevant in terms of the safety-related behavior of these systems is a key aspect for their successful verification and validation. Therefore, a method for extracting multimodal urban traffic scenarios from naturalistic road traffic data in an unsupervised manner, minimizing the amount of (potentially biased) prior expert knowledge, is proposed. Rather than an (elaborate) rule-based assignment by extracting concrete scenarios into predefined functional scenarios, the presented method deploys an unsupervised machine learning pipeline. The approach allows exploring the unknown nature of the data and their interpretation as test scenarios that experts could not have anticipated. The method is evaluated for naturalistic road traffic data at urban intersections from the inD and the Silicon Valley Intersections datasets. For this purpose, it is analyzed with which clustering approach (K-Means, hierarchical clustering, and DBSCAN) the scenario extraction method performs best (referring to an elaborate rule-based implementation). Subsequently, using hierarchical clustering the results show both a jump in overall accuracy of around 20% when moving from 4 to 5 clusters and a saturation effect starting at 41 clusters with an overall accuracy of 84%. These observations can be a valuable contribution in the context of the trade-off between the number of functional scenarios (i.e., clustering accuracy) and testing effort. Possible reasons for the observed accuracy variations of different clusters, each with a fixed total number of given clusters, are discussed. △ Less

Submitted 21 April, 2023; v1 submitted 14 February, 2022; originally announced February 2022.

Comments: 16 pages, 9 figures

Report number: SAE 12-06-03-0017 ACM Class: D.2; I.2

Journal ref: SAE Intl. J CAV 6(3):2023

Layer coupling between solutal and thermal convection in liquid metal batteries

Authors: Paolo Personnettaz, Tanja Sophia Klopper, Norbert Weber, Tom Weier

Abstract: For longer than one decade, liquid metal batteries (LMBs) are developed with the primary aim to provide economic stationary energy storage. Featuring two liquid metal electrodes separated by a molten salt electrolyte, LMBs operate at elevated temperature as simple concentration cells. Therefore, efficient mass transfer is a basic prerequisite for their economic operation. Understanding these mecha… ▽ More For longer than one decade, liquid metal batteries (LMBs) are developed with the primary aim to provide economic stationary energy storage. Featuring two liquid metal electrodes separated by a molten salt electrolyte, LMBs operate at elevated temperature as simple concentration cells. Therefore, efficient mass transfer is a basic prerequisite for their economic operation. Understanding these mechanisms cannot be limited at the single layer level. With this motivation, the effects of solutal- and thermally-driven flow are studied, as well as the flow coupling between the three liquid layers of the cell. It is shown that solutal convection appears first and thermal convection much later. While the presence of solutal flow depends on the mode of operation (charge or discharge), the occurrence of thermal convection is dictated by the geometry (thickness of layers). The coupling of the flow phenomena between the layers is intriguing: while thermal convection is confined to its area of origin, i.e. the electrolyte, solutal convection is able to drive flow in the positive electrode and in the electrolyte. △ Less

Submitted 7 February, 2022; originally announced February 2022.

Comments: preprint

Journal ref: International Journal of Heat and Mass Transfer 188 (2022) 122555

Cell voltage model for Li-Bi liquid metal batteries

Authors: Norbert Weber, Carolina Duczek, Gerrit M. Horstmann, Steffen Landgraf, Michael Nimtz, Paolo Personnettaz, Tom Weier, Donald R. Sadoway

Abstract: Lithium-bismuth bimetallic cells are amongst the best explored liquid metal batteries. A simple and fast quasi-one-dimensional cell voltage model for such devices is presented. The equilibrium cell potential is obtained from a complex two-dimensional fit of data drawn from multiple studies of equilibrium cell potential and rendered congruent with the phase diagram. Likewise, several analytical and… ▽ More Lithium-bismuth bimetallic cells are amongst the best explored liquid metal batteries. A simple and fast quasi-one-dimensional cell voltage model for such devices is presented. The equilibrium cell potential is obtained from a complex two-dimensional fit of data drawn from multiple studies of equilibrium cell potential and rendered congruent with the phase diagram. Likewise, several analytical and fit functions for the ohmic potential drop across the electrolyte are provided for different battery geometries. Mass transport overpotentials originating from the alloying of Li into Bi are modelled by solving a diffusion equation, either analytically or numerically, and accounting for the volume change of the positive electrode. The applicability and limitations of the model are finally illustrated in three distinct experimental settings. △ Less

Submitted 13 January, 2022; originally announced January 2022.

Journal ref: Applied Energy 309 (2022) 118331

Transient behaviour of electrovortex flow in a cylindrical container

Authors: Ke Liu, Frank Stefani, Norbert Weber, Tom Weier, Ben-Wen Li

Abstract: This study is a continuation of the combined experimental and numerical investigation [1] of the flow of the eutectic GaInSn alloy inside a cylindrical vessel exposed to a constant electrical current. The emerging electrovortex flow driven by the interaction of the current, which is applied through a tapered electrode, with its own magnetic field might have both detrimental and advantageous effect… ▽ More This study is a continuation of the combined experimental and numerical investigation [1] of the flow of the eutectic GaInSn alloy inside a cylindrical vessel exposed to a constant electrical current. The emerging electrovortex flow driven by the interaction of the current, which is applied through a tapered electrode, with its own magnetic field might have both detrimental and advantageous effects in liquid metal batteries. While the former work [1] was mainly concerned with time-averaged results, this paper focuses on the transient behaviour of the electrovortex flow which becomes most relevant under the influence of an external axial field. The additional Lorentz force generated by the interaction of the imposed current with the vertical component of the geomagnetic field bz drives the ordinary electrovortex jet flow into a swirling motion. Velocity distributions and motion characteristics, such as spiral streamlines and shortened and irregularly swinging jet regions, are investigated. The mechanism is analysed in detail for bz = 25.5 μT. The maximum angular velocity of the rotating jet is basically linearly dependent on bz, at least for the values studied here. A good agreement between the transient simulation and the experimental result is shown. △ Less

Submitted 21 January, 2022; originally announced January 2022.

Journal ref: Magnetohydrodynamics 57 (2021) 437-448

Decoupled coordinates for machine learning-based molecular fragment linking

Authors: Markus Fleck, Noah Weber, Christopher Trummer

Abstract: Recent developments in machine-learning based molecular fragment linking have demonstrated the importance of informing the generation process with structural information specifying the relative orientation of the fragments to be linked. However, such structural information has not yet been provided in the form of a complete relative coordinate system. Mathematical details for a decoupled set of bo… ▽ More Recent developments in machine-learning based molecular fragment linking have demonstrated the importance of informing the generation process with structural information specifying the relative orientation of the fragments to be linked. However, such structural information has not yet been provided in the form of a complete relative coordinate system. Mathematical details for a decoupled set of bond lengths, bond angles and torsion angles are elaborated and the coordinate system is demonstrated to be complete. Significant impact on the quality of the generated linkers is demonstrated numerically. The amount of reliable information within the different types of degrees of freedom is investigated. Ablation studies and an information-theoretical analysis are performed. The presented benefits suggest the application of a complete and decoupled relative coordinate system as a standard good practice in linker design. △ Less

Submitted 1 November, 2021; originally announced November 2021.

Comments: 16 pages, 5 Figures

ACM Class: I.2

A Needle in a Haystack -- How to Derive Relevant Scenarios for Testing Automated Driving Systems in Urban Areas

Authors: Nico Weber, Christoph Thiem, Ulrich Konigorski

Abstract: While there was great progress regarding the technology and its implementation for vehicles equipped with automated driving systems (ADS), the problem of how to proof their safety as a necessary precondition prior to market launch remains unsolved. One promising solution are scenario-based test approaches; however, there is no commonly accepted way of how to systematically generate and extract the… ▽ More While there was great progress regarding the technology and its implementation for vehicles equipped with automated driving systems (ADS), the problem of how to proof their safety as a necessary precondition prior to market launch remains unsolved. One promising solution are scenario-based test approaches; however, there is no commonly accepted way of how to systematically generate and extract the set of relevant scenarios to be tested to sufficiently capture the real-world traffic dynamics, especially for urban operational design domains. Within the scope of this paper, the overall concept of a novel simulation-based toolchain for the development and testing of ADS-equipped vehicles in urban environments is presented. Based on previous work regarding highway environments, the developed novel enhancements aim at empowering the toolchain to be able to deal with the increased complexity due to the more complex road networks with multi-modal interactions of various traffic participants. Based on derived requirements, a thorough explanation of different modules constituting the toolchain is given, showing first results and identified research gaps, respectively. A closer look is taken on two use cases: First, it is investigated whether the toolchain is capable to serve as synthetic data source within the development phase of ADS-equipped vehicles to enrich a scenario database in terms of extent, complexity and impacts of different what-if-scenarios for future mixed traffic. Second, it is analyzed how to combine the individual advantages of real recorded data and an agent-based simulation within a so-called adaptive replay-to-sim approach to support the testing phase of an ADS-equipped vehicle. The developed toolchain contributes to the overarching goal of a commonly accepted methodology for the validation and safety proof of ADS-equipped vehicles, especially in urban environments. △ Less

Submitted 8 September, 2021; originally announced September 2021.

Comments: 37 pages, 15 figures, Preprint

Toward Generating Sufficiently Valid Test Case Results: A Method for Systematically Assigning Test Cases to Test Bench Configurations in a Scenario-Based Test Approach for Automated Vehicles

Authors: Markus Steimle, Nico Weber, Markus Maurer

Abstract: To successfully launch automated vehicles into the consumer market, there must be credible proof that the vehicles will operate safely. However, finding a method to validate the vehicles' safe operation is a challenging problem. While scenario-based test approaches seem to be possible solutions, they require execution of a large number of test cases. Several test benches, ranging from actual test… ▽ More To successfully launch automated vehicles into the consumer market, there must be credible proof that the vehicles will operate safely. However, finding a method to validate the vehicles' safe operation is a challenging problem. While scenario-based test approaches seem to be possible solutions, they require execution of a large number of test cases. Several test benches, ranging from actual test vehicles to partly or fully simulated environments, are available to execute these test cases. Each test bench provides different elements, which in turn, have different parameters and parameter ranges. The composition of elements with their specific parameter values at a specific test bench that is used to execute a test case is referred to as a test bench configuration. However, selecting the most suitable test bench configuration is difficult. The selected test bench configuration determines whether the execution of a specific test case provides sufficiently valid test case results with respect to the intended purpose, for example, validating a vehicle's safe operation. The effective and efficient execution of a large number of test cases requires a method for systematically assigning test cases to the most suitable test bench configuration. Based on a proposed method for classifying test bench configurations, we propose and illustrate a method for systematically assigning test cases to test bench configurations in a scenario-based test approach for automated vehicles. This assignment method allows for the effective and efficient execution of a large number of test cases while generating sufficiently valid test case results. △ Less

Submitted 20 January, 2022; v1 submitted 7 September, 2021; originally announced September 2021.

Comments: Published in IEEE Access, vol. 10, pp. 6260-6285, 2022, doi: 10.1109/ACCESS.2022.3141198

Self-discharge mitigation in a liquid metal displacement battery

Authors: Kashif Mushtaq, Ji Zhao, Norbert Weber, Adelio Mendes, Donald R. Sadoway

Abstract: Recently, a disruptive idea was reported about the discovery of a new type of battery named Liquid Displacement Battery (LDB) comprising liquid metal electrodes and molten salt electrolyte. This cell featured a novel concept of a porous electronically conductive faradaic membrane instead of the traditional ion-selective ceramic membrane. LDBs are attractive for stationary storage applications but… ▽ More Recently, a disruptive idea was reported about the discovery of a new type of battery named Liquid Displacement Battery (LDB) comprising liquid metal electrodes and molten salt electrolyte. This cell featured a novel concept of a porous electronically conductive faradaic membrane instead of the traditional ion-selective ceramic membrane. LDBs are attractive for stationary storage applications but need mitigation against self-discharge. In the instant battery chemistry, Li|LiCl-PbCl$_2$|Pb, reducing the diffusion coefficient of lead ions can be a way forward and a solution can be the addition of PbO to the electrolyte. The latter acts as a supplementary barrier and complements the function of the faradaic membrane. The remedial actions improved the cell's coulombic efficiency from 92% to 97% without affecting the voltage efficiency. In addition, the limiting current density of a 500 mAh cell increased from 575 to 831 mA cm$^{-2}$ and the limiting power from 2.53 to 3.66 W. Finally, the effect of PbO on the impedance and polarization of the cell was also studied. △ Less

Submitted 24 August, 2021; originally announced August 2021.

Journal ref: Journal of Energy Chemistry 66 (2022) 390-396

Anode-metal drop formation and detachment mechanisms in liquid metal batteries

Authors: Sabrina Bénard, Norbert Weber, Gerrit Maik Horstmann, Steffen Landgraf, Tom Weier

Abstract: We study numerically localised short circuits in Li||Bi liquid metal batteries. In the prototype of a classical, three liquid-layer system, we assume a perceptible local deformation of the Li-salt interface. We find that there always exists a critical current at which a Li-droplet is cut off from this hump, and transferred to the Bi-phase. In a second case, we assume that the molten Li is containe… ▽ More We study numerically localised short circuits in Li||Bi liquid metal batteries. In the prototype of a classical, three liquid-layer system, we assume a perceptible local deformation of the Li-salt interface. We find that there always exists a critical current at which a Li-droplet is cut off from this hump, and transferred to the Bi-phase. In a second case, we assume that the molten Li is contained in a metal foam, and that a small Li-droplet emerges below this foam due to insufficient wetting. This droplet is deformed by Lorentz forces, until eventually being pinched off. Here, the critical current is slightly lower than in the three-layer system, and both, a droplet transfer and complete short circuits are observed. Finally, we discuss the relevance of our simulations for experimentally observed short circuits and non-faradaic Li-transfer. △ Less

Submitted 23 August, 2021; originally announced August 2021.

Journal ref: Journal of Power Sources 510 (2021) 230339

HALF: Holistic Auto Machine Learning for FPGAs

Authors: Jonas Ney, Dominik Loroch, Vladimir Rybalkin, Nico Weber, Jens Krüger, Norbert Wehn

Abstract: Deep Neural Networks (DNNs) are capable of solving complex problems in domains related to embedded systems, such as image and natural language processing. To efficiently implement DNNs on a specific FPGA platform for a given cost criterion, e.g. energy efficiency, an enormous amount of design parameters has to be considered from the topology down to the final hardware implementation. Interdependen… ▽ More Deep Neural Networks (DNNs) are capable of solving complex problems in domains related to embedded systems, such as image and natural language processing. To efficiently implement DNNs on a specific FPGA platform for a given cost criterion, e.g. energy efficiency, an enormous amount of design parameters has to be considered from the topology down to the final hardware implementation. Interdependencies between the different design layers have to be taken into account and explored efficiently, making it hardly possible to find optimized solutions manually. An automatic, holistic design approach can improve the quality of DNN implementations on FPGA significantly. To this end, we present a cross-layer design space exploration methodology. It comprises optimizations starting from a hardware-aware topology search for DNNs down to the final optimized implementation for a given FPGA platform. The methodology is implemented in our Holistic Auto machine Learning for FPGAs (HALF) framework, which combines an evolutionary search algorithm, various optimization steps and a library of parametrizable hardware DNN modules. HALF automates both the exploration process and the implementation of optimized solutions on a target FPGA platform for various applications. We demonstrate the performance of HALF on a medical use case for arrhythmia detection for three different design goals, i.e. low-energy, low-power and high-throughput respectively. Our FPGA implementation outperforms a TensorRT optimized model on an Nvidia Jetson platform in both throughput and energy consumption. △ Less

Submitted 20 October, 2021; v1 submitted 28 June, 2021; originally announced June 2021.

Comments: 2021 31st International Conference on Field-Programmable Logic and Applications (FPL). IEEE, 2021

Human Schema Curation via Causal Association Rule Mining

Authors: Noah Weber, Anton Belyy, Nils Holzenberger, Rachel Rudinger, Benjamin Van Durme

Abstract: Event schemas are structured knowledge sources defining typical real-world scenarios (e.g., going to an airport). We present a framework for efficient human-in-the-loop construction of a schema library, based on a novel script induction system and a well-crafted interface that allows non-experts to "program" complex event structures. Associated with this work we release a schema library: a machine… ▽ More Event schemas are structured knowledge sources defining typical real-world scenarios (e.g., going to an airport). We present a framework for efficient human-in-the-loop construction of a schema library, based on a novel script induction system and a well-crafted interface that allows non-experts to "program" complex event structures. Associated with this work we release a schema library: a machine readable resource of 232 detailed event schemas, each of which describe a distinct typical scenario in terms of its relevant sub-event structure (what happens in the scenario), participants (who plays a role in the scenario), fine-grained typing of each participant, and the implied relational constraints between them. We make our schema library and the SchemaBlocks interface available online. △ Less

Submitted 23 May, 2022; v1 submitted 18 April, 2021; originally announced April 2021.

Comments: 12 pages, 6 figures, 6 tables

Conductivity influence on interfacial waves in liquid metal batteries and related two-layer systems

Authors: Tom Weier, Ilmars Grants, Gerrit Maik Horstmann, Steffen Landgraf, Michael Nimtz, Paolo Personnettaz, Frank Stefani, Norbert Weber

Abstract: Fluid flows in liquid metal batteries can be generated by a number of effects. We start with a short overview of different driving mechanisms and then address questions specific to the metal pad role instabilities in three-layer systems. We focus on the role of the conductivity distribution in the cell, noting at the same time that interfacial tension should be considered as well for smaller cells… ▽ More Fluid flows in liquid metal batteries can be generated by a number of effects. We start with a short overview of different driving mechanisms and then address questions specific to the metal pad role instabilities in three-layer systems. We focus on the role of the conductivity distribution in the cell, noting at the same time that interfacial tension should be considered as well for smaller cells. Following this discussion, numerical results on the excitation of interfacial waves in two-layer liquid metal systems with miscibility gaps bearing an interface normal electric current are presented. Confirming recent results from the literature, we find that magnetic damping plays a decisive role for strong vertical magnetic fields. In addition, boundary conditions for the electric field strongly influence critical currents and growth rates. △ Less

Submitted 9 April, 2021; originally announced April 2021.

Journal ref: Magnetohydrodynamics 56 (2020), No. 2-3, pp. 237-245

Effects of current distribution on mass transport in the positive electrode of a liquid metal battery

Authors: Paolo Personnettaz, Steffen Landgraf, Michael Nimtz, Norbert Weber, Tom Weier

Abstract: Liquid metal electrodes are one of the key components of different electrical energy storage technologies. The understanding of transport phenomena in liquid electrodes is mandatory in order to ensure efficient operation. In the present study we focus our attention on the positive electrode of the Li||Bi liquid metal battery. Starting from a real experimental setup, we numerically investigate the… ▽ More Liquid metal electrodes are one of the key components of different electrical energy storage technologies. The understanding of transport phenomena in liquid electrodes is mandatory in order to ensure efficient operation. In the present study we focus our attention on the positive electrode of the Li||Bi liquid metal battery. Starting from a real experimental setup, we numerically investigate the charge transfer in a molten salt electrolyte and the mass transport in the positive electrode. The two phenomena are tightly coupled, because the current distribution influences the concentration field in the positive electrode. The cell is studied during charging when compositional convection becomes apparent. First results of compositional convection from a non-uniform current distribution are presented, highlighting its capability to affect the flow in the positive electrode and the cell performance. △ Less

Submitted 31 March, 2021; originally announced April 2021.

Journal ref: Magnetohydrodynamics Vol. 56 (2020), No. 2-3, pp. 247-254

Addressing Research Software Sustainability via Institutes

Authors: Daniel S. Katz, Jeffrey C. Carver, Neil P. Chue Hong, Sandra Gesing, Simon Hettrick, Tom Honeyman, Karthik Ram, Nicholas Weber

Abstract: Research software is essential to modern research, but it requires ongoing human effort to sustain: to continually adapt to changes in dependencies, to fix bugs, and to add new features. Software sustainability institutes, amongst others, develop, maintain, and disseminate best practices for research software sustainability, and build community around them. These practices can both reduce the amou… ▽ More Research software is essential to modern research, but it requires ongoing human effort to sustain: to continually adapt to changes in dependencies, to fix bugs, and to add new features. Software sustainability institutes, amongst others, develop, maintain, and disseminate best practices for research software sustainability, and build community around them. These practices can both reduce the amount of effort that is needed and create an environment where the effort is appreciated and rewarded. The UK SSI is such an institute, and the US URSSI and the Australian AuSSI are planning to become institutes, and this extended abstract discusses them and the strengths and weaknesses of this approach. △ Less

Submitted 5 March, 2021; originally announced March 2021.

Comments: accepted by ICSE 2021 BokSS Workshop (https://bokss.github.io/bokss2021/)

Deep learning based quantum vortex detection in atomic Bose-Einstein condensates

Authors: Friederike Metz, Juan Polo, Natalya Weber, Thomas Busch

Abstract: Quantum vortices naturally emerge in rotating Bose-Einstein condensates (BECs) and, similarly to their classical counterparts, allow the study of a range of interesting out-of-equilibrium phenomena like turbulence and chaos. However, the study of such phenomena requires to determine the precise location of each vortex within a BEC, which becomes challenging when either only the condensate density… ▽ More Quantum vortices naturally emerge in rotating Bose-Einstein condensates (BECs) and, similarly to their classical counterparts, allow the study of a range of interesting out-of-equilibrium phenomena like turbulence and chaos. However, the study of such phenomena requires to determine the precise location of each vortex within a BEC, which becomes challenging when either only the condensate density is available or sources of noise are present, as is typically the case in experimental settings. Here, we introduce a machine learning based vortex detector motivated by state-of-the-art object detection methods that can accurately locate vortices in simulated BEC density images. Our model allows for robust and real-time detection in noisy and non-equilibrium configurations. Furthermore, the network can distinguish between vortices and anti-vortices if the condensate phase profile is also available. We anticipate that our vortex detector will be advantageous both for experimental and theoretical studies of the static and dynamical properties of vortex configurations in BECs. △ Less

Submitted 13 April, 2021; v1 submitted 23 December, 2020; originally announced December 2020.

Comments: 13+8 pages, 5+4 figures

Journal ref: Mach. Learn.: Sci. Technol. 2 035019 (2021)

Polaronic Contributions to Friction in a Manganite Thin Film

Authors: Niklas A. Weber, Hendrik Schmidt, Tim Sievert, Christian Jooss, Friedrich Güthoff, Vasily Moshneaga, Konrad Samwer, Matthias Krüger, Cynthia A. Volkert

Abstract: Despite the huge importance of friction in regulating movement in all natural and technological processes, the mechanisms underlying dissipation at a sliding contact are still a matter of debate. Attempts to explain the dependence of measured frictional losses at nanoscale contacts on the electronic degrees of freedom of the surrounding materials have so far been controversial. Here, it is propose… ▽ More Despite the huge importance of friction in regulating movement in all natural and technological processes, the mechanisms underlying dissipation at a sliding contact are still a matter of debate. Attempts to explain the dependence of measured frictional losses at nanoscale contacts on the electronic degrees of freedom of the surrounding materials have so far been controversial. Here, it is proposed that friction can be explained by considering damping of stick-slip pulses in a sliding contact. Based on friction force microscopy studies of La$_{(1-x)}$Sr$_x$MnO$_3$ films at the ferromagnetic-metallic to paramagnetic-polaronic conductor phase transition, it is confirmed that the sliding contact generates thermally-activated slip pulses in the nanoscale contact, and argued that these are damped by direct coupling into phonon bath. Electron-phonon coupling leads to the formation of Jahn-Teller polarons and a clear increase in friction in the high temperature phase. There is no evidence for direct electronic drag on the atomic force microscope tip nor any indication of contributions from electrostatic forces. This intuitive scenario, that friction is governed by the damping of surface vibrational excitations, provides a basis for reconciling controversies in literature studies as well as suggesting possible tactics for controlling friction. △ Less

Submitted 25 September, 2020; originally announced September 2020.

Thermal neutron radiography of a passive proton exchange membrane fuel cell for portable hydrogen energy systems

Authors: Antonio M. Chaparro, P. Ferreira-Aparicio, M. Antonia Folgado, Rico Hübscher, Carsten Lange, Norbert Weber

Abstract: A proton exchange membrane fuel cell (PEMFC) for low power and portable applications is studied with thermal neutron radiography. The PEMFC operates under full passive conditions, with an air-breathing cathode and a dead-end anode supplied with static ambient air and dry hydrogen, respectively. A columnar cathodic plate favors the mobility of water drops over the cathode surface and their eliminat… ▽ More A proton exchange membrane fuel cell (PEMFC) for low power and portable applications is studied with thermal neutron radiography. The PEMFC operates under full passive conditions, with an air-breathing cathode and a dead-end anode supplied with static ambient air and dry hydrogen, respectively. A columnar cathodic plate favors the mobility of water drops over the cathode surface and their elimination. Thermal neutron images show liquid water build up during operation with the cell in vertical and horizontal positions, i.e. with its main plane aligned parallel and perpendicular to the gravity field, respectively. Polarization curves and impedance spectroscopy show cell orientation dependent response that can be related with the water accumulation profiles. In vertical position, lower water contents in the cathode electrode is favored by the elimination of water drops rolling over the cathode surface and the onset of natural convection. As a consequence, oxygen transport is improved in the vertical cell, that can be operated full passive for hours under ambient conditions, providing steady peak power densities above 100 mW/cm2. Gravity and natural convection are less effective in horizontal position, leading to a 17% decrease in peak power density due to oxygen transport losses. The horizontal position is especially adverse if the upper electrode is the cathode, because of anode flooding causing cell failure after production of a small amount of water (5 mg/cm2). The combined information from thermal neutron radiography and cell response characteristics explains the important influence of cell orientation on the performance of a fully passive air-breathing PEMFC for portable applications. △ Less

Submitted 14 September, 2020; originally announced September 2020.

Journal ref: Journal of Power Sources 480 (2020) 228668

Switching friction at a manganite surface using electric fields

Authors: H. Schmidt, J. -O. Krisponeit, N. Weber, K. Samwer, C. A. Volkert

Abstract: We report active control of the friction force at the contact between a nanoscale asperity and a La$_{0.55}$Ca$_{0.45}$MnO$_3$ (LCMO) thin film using electric fields. We use friction force microscopy under ultrahigh vacuum conditions to measure the friction force as we change the film resistive state by electric field-induced resistive switching. Friction forces are high in the insulating state an… ▽ More We report active control of the friction force at the contact between a nanoscale asperity and a La$_{0.55}$Ca$_{0.45}$MnO$_3$ (LCMO) thin film using electric fields. We use friction force microscopy under ultrahigh vacuum conditions to measure the friction force as we change the film resistive state by electric field-induced resistive switching. Friction forces are high in the insulating state and clearly change to lower values when the probed local region is switched to the conducting state. Upon switching back to an insulating state, the friction forces increase again. Thus, we demonstrate active control of friction without having to change the contact temperature or pressure. By comparing with measurements of friction at the metal-to-insulator transition and with the effect of applied voltage on adhesion, we rule out electronic excitations, electrostatic forces and changes in contact area as the reasons for the effect of resistive switching on friction. Instead, we argue that friction is limited by phonon relaxation times which are strongly coupled to the electronic degrees of freedom through distortions of the MnO6 octahedra. The concept of controlling friction forces by electric fields should be applicable to any materials where the field produces strong changes in phonon lifetimes. △ Less

Submitted 16 May, 2020; originally announced May 2020.

Comments: Manuscript: 14 pages, 5 figures. Supporting information: 5 pages, 3 figures. arXiv admin note: substantial text overlap with arXiv:1611.02684

Generating Narrative Text in a Switching Dynamical System

Authors: Noah Weber, Leena Shekhar, Heeyoung Kwon, Niranjan Balasubramanian, Nathanael Chambers

Abstract: Early work on narrative modeling used explicit plans and goals to generate stories, but the language generation itself was restricted and inflexible. Modern methods use language models for more robust generation, but often lack an explicit representation of the scaffolding and dynamics that guide a coherent narrative. This paper introduces a new model that integrates explicit narrative structure w… ▽ More Early work on narrative modeling used explicit plans and goals to generate stories, but the language generation itself was restricted and inflexible. Modern methods use language models for more robust generation, but often lack an explicit representation of the scaffolding and dynamics that guide a coherent narrative. This paper introduces a new model that integrates explicit narrative structure with neural language models, formalizing narrative modeling as a Switching Linear Dynamical System (SLDS). A SLDS is a dynamical system in which the latent dynamics of the system (i.e. how the state vector transforms over time) is controlled by top-level discrete switching variables. The switching variables represent narrative structure (e.g., sentiment or discourse states), while the latent state vector encodes information on the current state of the narrative. This probabilistic formulation allows us to control generation, and can be learned in a semi-supervised fashion using both labeled and unlabeled data. Additionally, we derive a Gibbs sampler for our model that can fill in arbitrary parts of the narrative, guided by the switching variables. Our filled-in (English language) narratives outperform several baselines on both automatic and human evaluations. △ Less

Submitted 7 April, 2020; originally announced April 2020.

Causal Inference of Script Knowledge

Authors: Noah Weber, Rachel Rudinger, Benjamin Van Durme

Abstract: When does a sequence of events define an everyday scenario and how can this knowledge be induced from text? Prior works in inducing such scripts have relied on, in one form or another, measures of correlation between instances of events in a corpus. We argue from both a conceptual and practical sense that a purely correlation-based approach is insufficient, and instead propose an approach to scrip… ▽ More When does a sequence of events define an everyday scenario and how can this knowledge be induced from text? Prior works in inducing such scripts have relied on, in one form or another, measures of correlation between instances of events in a corpus. We argue from both a conceptual and practical sense that a purely correlation-based approach is insufficient, and instead propose an approach to script induction based on the causal effect between events, formally defined via interventions. Through both human and automatic evaluations, we show that the output of our method based on causal effects better matches the intuition of what a script represents △ Less

Submitted 2 April, 2020; originally announced April 2020.

Comments: Pre-Print

SOL: Effortless Device Support for AI Frameworks without Source Code Changes

Authors: Nicolas Weber, Felipe Huici

Abstract: Modern high performance computing clusters heavily rely on accelerators to overcome the limited compute power of CPUs. These supercomputers run various applications from different domains such as simulations, numerical applications or artificial intelligence (AI). As a result, vendors need to be able to efficiently run a wide variety of workloads on their hardware. In the AI domain this is in part… ▽ More Modern high performance computing clusters heavily rely on accelerators to overcome the limited compute power of CPUs. These supercomputers run various applications from different domains such as simulations, numerical applications or artificial intelligence (AI). As a result, vendors need to be able to efficiently run a wide variety of workloads on their hardware. In the AI domain this is in particular exacerbated by the existence of a number of popular frameworks (e.g, PyTorch, TensorFlow, etc.) that have no common code base, and can vary in functionality. The code of these frameworks evolves quickly, making it expensive to keep up with all changes and potentially forcing developers to go through constant rounds of upstreaming. In this paper we explore how to provide hardware support in AI frameworks without changing the framework's source code in order to minimize maintenance overhead. We introduce SOL, an AI acceleration middleware that provides a hardware abstraction layer that allows us to transparently support heterogeneous hardware. As a proof of concept, we implemented SOL for PyTorch with three backends: CPUs, GPUs and vector processors. △ Less

Submitted 24 March, 2020; originally announced March 2020.

Comments: HPML Workshop 2020

Parameter learning and fractional differential operators: application in image regularization and decomposition

Authors: Sören Bartels, Nico Weber

Abstract: In this paper, we focus on learning optimal parameters for PDE-based image regularization and decomposition. First we learn the regularization parameter and the differential operator for gray-scale image denoising using the fractional Laplacian in combination with a bilevel optimization problem. In our setting the fractional Laplacian allows the use of Fourier transform, which enables the optimiza… ▽ More In this paper, we focus on learning optimal parameters for PDE-based image regularization and decomposition. First we learn the regularization parameter and the differential operator for gray-scale image denoising using the fractional Laplacian in combination with a bilevel optimization problem. In our setting the fractional Laplacian allows the use of Fourier transform, which enables the optimization of the denoising operator. We prove stable and explainable results as an advantage in comparison to other machine learning approaches. The numerical experiments correlate with our theoretical model setting and show a reduction of computing time in contrast to the ROF model. Second we introduce a new image decomposition model with the fractional Laplacian and the Riesz potential. We provide an explicit formula for the unique solution and the numerical experiments illustrate the efficiency. △ Less

Submitted 10 January, 2020; originally announced January 2020.

Numerical and Experimental Investigation of Electro-Vortex Flow in a Cylindrical Container

Authors: K. Liu, F. Stefani, N. Weber, T. Weier, B. W. Li

Abstract: In a cylindrical container filled with an eutectic GaInSn alloy, an electro-vortex flow (EVF) is generated by the interaction of a non-uniform current with its own magnetic field. In this paper, we investigate the EVF phenomenon numerically and experimentally. Ultrasound Doppler Velocimetry (UDV) is applied to measure the velocity field in a cylindrical vessel. Second, we enhance an old numerical… ▽ More In a cylindrical container filled with an eutectic GaInSn alloy, an electro-vortex flow (EVF) is generated by the interaction of a non-uniform current with its own magnetic field. In this paper, we investigate the EVF phenomenon numerically and experimentally. Ultrasound Doppler Velocimetry (UDV) is applied to measure the velocity field in a cylindrical vessel. Second, we enhance an old numerical solver by taking into account the effect of Joule heating, and employ it for the numerical simulation of the EVF experiment. Special focus is laid on the role of the magnetic field, which is the combination of the current induced magnetic field and the external geomagnetic field. For getting a higher computational efficiency, the so-called parent-child mesh technique is applied in OpenFOAM when computing the electric potential, the current density and the temperature in the coupled solid-liquid conductor system. The results of the experiment are in good agreement with those of the simulation. This study may help to identify the factors that are essential for the EVF phenomenon, and for quantifying its role in liquid metal batteries. △ Less

Submitted 23 October, 2019; originally announced October 2019.

Comments: 16 pages, 8 figures

Modeling discontinuous potential distributions using the finite volume method, and application to liquid metal batteries

Authors: Norbert Weber, Steffen Landgraf, Kashif Mushtaq, Michael Nimtz, Paolo Personnettaz, Tom Weier, Ji Zhao, Donald Sadoway

Abstract: The electrical potential in a battery jumps at each electrode-electrolyte interface. We present a model for computing three-dimensional current and potential distributions, which accounts for such internal voltage jumps. Within the framework of the finite volume method we discretize the Laplace and gradient operators such that they account for internal jump boundary conditions. After implementing… ▽ More The electrical potential in a battery jumps at each electrode-electrolyte interface. We present a model for computing three-dimensional current and potential distributions, which accounts for such internal voltage jumps. Within the framework of the finite volume method we discretize the Laplace and gradient operators such that they account for internal jump boundary conditions. After implementing a simple battery model in OpenFOAM we validate it using an analytical test case, and show its capabilities by simulating the current distribution and discharge curve of a Li||Bi liquid metal battery. △ Less

Submitted 21 June, 2019; originally announced June 2019.

Journal ref: Electrochimica Acta 318 (2019) 857-864

Hierarchical Quantized Representations for Script Generation

Authors: Noah Weber, Leena Shekhar, Niranjan Balasubramanian, Nathanael Chambers

Abstract: Scripts define knowledge about how everyday scenarios (such as going to a restaurant) are expected to unfold. One of the challenges to learning scripts is the hierarchical nature of the knowledge. For example, a suspect arrested might plead innocent or guilty, and a very different track of events is then expected to happen. To capture this type of information, we propose an autoencoder model with… ▽ More Scripts define knowledge about how everyday scenarios (such as going to a restaurant) are expected to unfold. One of the challenges to learning scripts is the hierarchical nature of the knowledge. For example, a suspect arrested might plead innocent or guilty, and a very different track of events is then expected to happen. To capture this type of information, we propose an autoencoder model with a latent space defined by a hierarchy of categorical variables. We utilize a recently proposed vector quantization based approach, which allows continuous embeddings to be associated with each latent variable value. This permits the decoder to softly decide what portions of the latent hierarchy to condition on by attending over the value embeddings for a given setting. Our model effectively encodes and generates scripts, outperforming a recent language modeling-based method on several standard tasks, and allowing the autoencoder model to achieve substantially lower perplexity scores compared to the previous language modeling-based method. △ Less

Submitted 28 August, 2018; originally announced August 2018.

Comments: EMNLP 2018

Thermally driven convection in Li||Bi liquid metal batteries

Authors: Paolo Personnettaz, Pascal Beckstein, Steffen Landgraf, Thomas Köllner, Michael Nimtz, Norbert Weber, Tom Weier

Abstract: Liquid Metal Batteries (LMBs) are a promising concept for cheap electrical energy storage at grid level. These are built as a stable density stratification of three liquid layers, with two liquid metals separated by a molten salt. In order to ensure a safe and efficient operation, the understanding of transport phenomena in LMBs is essential. With this motivation we study thermal convection induce… ▽ More Liquid Metal Batteries (LMBs) are a promising concept for cheap electrical energy storage at grid level. These are built as a stable density stratification of three liquid layers, with two liquid metals separated by a molten salt. In order to ensure a safe and efficient operation, the understanding of transport phenomena in LMBs is essential. With this motivation we study thermal convection induced by internal heat generation. We consider the electrochemical nature of the cell in order to define the heat balance and the operating parameters. Moreover we develop a simple 1D heat conduction model as wellas a fully 3D thermo-fluid dynamics model. The latter is implemented in the CFD library OpenFOAM, extending the volume of fluid solver, and validated against a pseudo-spectral code. Both models are used to study a rectangular 10x10 cm Li||Bi LMB cell at three different states of charge. △ Less

Submitted 29 May, 2018; originally announced May 2018.

Journal ref: Journal of Power Sources 401 (2018), 362-374

The Fine Line between Linguistic Generalization and Failure in Seq2Seq-Attention Models

Authors: Noah Weber, Leena Shekhar, Niranjan Balasubramanian

Abstract: Seq2Seq based neural architectures have become the go-to architecture to apply to sequence to sequence language tasks. Despite their excellent performance on these tasks, recent work has noted that these models usually do not fully capture the linguistic structure required to generalize beyond the dense sections of the data distribution \cite{ettinger2017towards}, and as such, are likely to fail o… ▽ More Seq2Seq based neural architectures have become the go-to architecture to apply to sequence to sequence language tasks. Despite their excellent performance on these tasks, recent work has noted that these models usually do not fully capture the linguistic structure required to generalize beyond the dense sections of the data distribution \cite{ettinger2017towards}, and as such, are likely to fail on samples from the tail end of the distribution (such as inputs that are noisy \citep{belkinovnmtbreak} or of different lengths \citep{bentivoglinmtlength}). In this paper, we look at a model's ability to generalize on a simple symbol rewriting task with a clearly defined structure. We find that the model's ability to generalize this structure beyond the training distribution depends greatly on the chosen random seed, even when performance on the standard test set remains the same. This suggests that a model's ability to capture generalizable structure is highly sensitive. Moreover, this sensitivity may not be apparent when evaluating it on standard test sets. △ Less

Submitted 8 May, 2018; v1 submitted 3 May, 2018; originally announced May 2018.

Comments: Workshop on New Forms of Generalization in Deep Learning and NLP (NAACL 2018), revised to update some references

BrainSlug: Transparent Acceleration of Deep Learning Through Depth-First Parallelism

Authors: Nicolas Weber, Florian Schmidt, Mathias Niepert, Felipe Huici

Abstract: Neural network frameworks such as PyTorch and TensorFlow are the workhorses of numerous machine learning applications ranging from object recognition to machine translation. While these frameworks are versatile and straightforward to use, the training of and inference in deep neural networks is resource (energy, compute, and memory) intensive. In contrast to recent works focusing on algorithmic en… ▽ More Neural network frameworks such as PyTorch and TensorFlow are the workhorses of numerous machine learning applications ranging from object recognition to machine translation. While these frameworks are versatile and straightforward to use, the training of and inference in deep neural networks is resource (energy, compute, and memory) intensive. In contrast to recent works focusing on algorithmic enhancements, we introduce BrainSlug, a framework that transparently accelerates neural network workloads by changing the default layer-by-layer processing to a depth-first approach, reducing the amount of data required by the computations and thus improving the performance of the available hardware caches. BrainSlug achieves performance improvements of up to 41.1% on CPUs and 35.7% on GPUs. These optimizations come at zero cost to the user as they do not require hardware changes and only need tiny adjustments to the software. △ Less

Submitted 23 April, 2018; originally announced April 2018.

Comments: Technical Report, 13 pages

Detail-Preserving Pooling in Deep Networks

Authors: Faraz Saeedan, Nicolas Weber, Michael Goesele, Stefan Roth

Abstract: Most convolutional neural networks use some method for gradually downscaling the size of the hidden layers. This is commonly referred to as pooling, and is applied to reduce the number of parameters, improve invariance to certain distortions, and increase the receptive field size. Since pooling by nature is a lossy process, it is crucial that each such layer maintains the portion of the activation… ▽ More Most convolutional neural networks use some method for gradually downscaling the size of the hidden layers. This is commonly referred to as pooling, and is applied to reduce the number of parameters, improve invariance to certain distortions, and increase the receptive field size. Since pooling by nature is a lossy process, it is crucial that each such layer maintains the portion of the activations that is most important for the network's discriminability. Yet, simple maximization or averaging over blocks, max or average pooling, or plain downsampling in the form of strided convolutions are the standard. In this paper, we aim to leverage recent results on image downscaling for the purposes of deep learning. Inspired by the human visual system, which focuses on local spatial changes, we propose detail-preserving pooling (DPP), an adaptive pooling method that magnifies spatial changes and preserves important structural detail. Importantly, its parameters can be learned jointly with the rest of the network. We analyze some of its theoretical properties and show its empirical benefits on several datasets and networks, where DPP consistently outperforms previous pooling approaches. △ Less

Submitted 11 April, 2018; originally announced April 2018.

Comments: To appear at CVPR 2018

Controlling Decoding for More Abstractive Summaries with Copy-Based Networks

Authors: Noah Weber, Leena Shekhar, Niranjan Balasubramanian, Kyunghyun Cho

Abstract: Attention-based neural abstractive summarization systems equipped with copy mechanisms have shown promising results. Despite this success, it has been noticed that such a system generates a summary by mostly, if not entirely, copying over phrases, sentences, and sometimes multiple consecutive sentences from an input paragraph, effectively performing extractive summarization. In this paper, we veri… ▽ More Attention-based neural abstractive summarization systems equipped with copy mechanisms have shown promising results. Despite this success, it has been noticed that such a system generates a summary by mostly, if not entirely, copying over phrases, sentences, and sometimes multiple consecutive sentences from an input paragraph, effectively performing extractive summarization. In this paper, we verify this behavior using the latest neural abstractive summarization system - a pointer-generator network. We propose a simple baseline method that allows us to control the amount of copying without retraining. Experiments indicate that the method provides a strong baseline for abstractive systems looking to obtain high ROUGE scores while minimizing overlap with the source article, substantially reducing the n-gram overlap with the original article while keeping within 2 points of the original model's ROUGE score. △ Less

Submitted 19 March, 2018; v1 submitted 19 March, 2018; originally announced March 2018.

Mining Open Government Data Used in Scientific Research

Authors: An Yan, Nicholas Weber

Abstract: In the following paper, we describe results from mining citations, mentions, and links to open government data (OGD) in peer-reviewed literature. We inductively develop a method for categorizing how OGD are used by different research communities, and provide descriptive statistics about the publication years, publication outlets, and OGD sources. Our results demonstrate that, 1. The use of OGD in… ▽ More In the following paper, we describe results from mining citations, mentions, and links to open government data (OGD) in peer-reviewed literature. We inductively develop a method for categorizing how OGD are used by different research communities, and provide descriptive statistics about the publication years, publication outlets, and OGD sources. Our results demonstrate that, 1. The use of OGD in research is steadily increasing from 2009 to 2016; 2. Researchers use OGD from 96 different open government data portals, with data.gov.uk and data.gov being the most frequent sources; and, 3.Contrary to previous findings, we provide evidence suggesting that OGD from developing nations, notably India and Kenya, are being frequently used to fuel scientific discoveries. The findings of this paper contribute to ongoing research agendas aimed at tracking the impact of open government data initiatives, and provides an initial description of how open government data are valuable to diverse scientific research communities. △ Less

Submitted 24 March, 2018; v1 submitted 8 February, 2018; originally announced February 2018.

Comments: Accepted to iConference 2018

Journal ref: Transforming Digital Worlds. iConference 2018. Lecture Notes in Computer Science, vol 10766. Springer, Cham