Bielefeld University

Liquid state machines (LSMs) are biologically more plausible than feedforward spiking neural networks for brain-inspired computing and neuromorphic engineering. However, optimizing and training complex recurrent network architectures in the reservoir of LSMs remains challenging. Most existing algorithms aim to adjust the synaptic strength only, without fundamentally modifying the reservoir architecture of LSMs. Recently, it has become popular to simultaneously optimize the architecture and parameters of the reservoir in LSMs. However, most existing architecture representation schemes are too restricted to discover more powerful architectures of LSMs. To address the above issue, this paper proposes a generative liquid state machine, whose reservoir architecture is evolved using a cooperative co-evolutionary algorithm whose weights are tuned by synaptic plasticity rules. To reduce the computation time for evolving the reservoir, random forest is adopted to assist the cooperative co-evolutionary algorithm, together with a data parallelism strategy. The proposed algorithm is assessed on three sequence classification benchmarks and our experimental results show that the proposed algorithm outperforms the state-of-the-art on the benchmark problems. Meanwhile, our analysis shows that the data parallelism strategy is effective in speeding up the evaluation process.

Optimization problems whose evaluations of the objective and constraints involve costly numerical simulations or physical experiments are referred to as expensive constrained optimization (ECO) problems. Such problems can be solved by evolutionary algorithms (EAs) in conjunction with computationally cheap surrogates that separately approximate the expensive objective and constraint functions. During the process of the ECO, the interested regions of surrogate models for the objective and constraints usually have a small overlap only. Specifically, the surrogate model for the objective function should focus on the prediction accuracy in the promising region, while the models for constraint functions should concentrate on the accuracy at the boundary of the feasible region. However, most existing methods neglect such differences and train those different models using the same training data, barely resulting in satisfactory performance. Therefore, we propose a general framework for solving expensive optimization problems with inequality constraints. In the proposed framework, the objective and constraints are separately trained with two different sets of training data to enhance the prediction accuracy and reliability in the interested regions. A novel infill sampling criterion is tailored to decide whether potentially better or more uncertain solutions should be sampled. Moreover, a new strategy, termed search intensity adjustment, is designed for adjusting the number of search generations on new surrogate models. We attempt to embed three competitive constrained EAs into our framework to verify its generality. The experimental results obtained on numerous benchmark functions from CEC2006, CEC2010, and CEC2017 have demonstrated the superiority of our approach over three state-of-the-art surrogate-assisted EAs.

Implementing key biochemical engineering principles based on the kinetics and stoichiometry of growth unlocks the full potential of microfluidic single-cell analysis. We introduce a unique integrative approach, using non-invasive advanced microfluidic cultivation and analysis technologies to integrate physiologic single-cell data. Our groundwork enables microscale material balancing beyond population-based average values and advances modern bioprocess modeling [1].

This article analyzes the social and spatial dynamics of the mobile trade in low‐cost goods by rural people from a mountainous region of China's Zhejiang province and how these interact with the mobility and social reproductive patterns of the rural–urban migrant workers they cater to. Also formally categorized as peasants, the traders not only supply the goods necessary for the maintenance of the workers but also of their spatially divided household, thus contributing to the reproduction of migrant labor power more generally. In doing so, they assume mobility trajectories that align with those of factory production and experience familial trade‐offs commonly experienced by migrant workers. Meanwhile, the provision of low‐cost goods to migrant workers has enabled a thriving economy employing peasant families for whom agricultural livelihoods slowly disappear. These dynamics indicate the mutual connection between waged and self‐employed labor that works in the interest of capital accumulation at the same time with the differentiation of migrant labor. As in other comparable Asian contexts, their connection lies at the heart of the state‐sponsored production regime premised on the low‐cost reproduction of flexible migrant labor.

We propose highly efficient photon pair generation via down-conversion in a polaritonic cyclic three-level system formed via coupling an exciton and a localized plasmon polariton. Pure dephasing of the exciton enables optical transitions from the upper to lower polariton state via symmetry breaking and thus opens a pathway for the cascaded emission of two photons. The strength of this down-conversion pathway is determined by the coupling strength between exciton and plasmon polariton and the exciton dephasing rate. In the case of strong up to ultrastrong coupling and strong dephasing, this results in a nanoscopic χ(2)-system capable of few-photon down-conversion. A semiconductor quantum dot embedded in a gap plasmon, which supports strong coupling between plasmon and exciton, in combination with fast pure dephasing in the quantum dot exciton, allows one to reach conversion efficiencies down to the single-photon limit under experimentally achievable parameters at ambient conditions.

The saponin β\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta$$\end{document}-aescin is well known for its self-aggregation above the critical micelle concentration (cmc) and its interaction with model membranes made of zwitterionic phospholipids including the formation of mixed bicelle systems. In this study, we investigate the interaction of β\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta$$\end{document}-aescin with small unilamellar vesicles (SUVs) made of the negatively charged lipid 1,2-dioleoyl-sn-glycero-3-phosphoglycerol (DOPG). The study is conducted at a pH value at which aescin is negatively charged as well, and mixtures up to an aescin content of 50 mol% (equivalent to a molecular ratio of 1:1) were investigated, so that the cmc of aescin is exceeded by far. Analysis of the system by scattering and NMR methods was performed with respect to two reference systems made of the bare components: DOPG SUVs and aescin micelles. Wide-angle X-ray scattering (WAXS) was used to determine molecular correlation distances for both kinds of molecules, and small-angle neutron and X-ray scattering (SANS and SAXS) revealed a structural picture of the system, which was further confirmed by diffusion-ordered nuclear magnetic resonance spectroscopy (DOSY-NMR). Contrary to the expected solubilization of the DOPG membrane, most probably none- or only weakly-interacting, separated DOPG SUVs and aescin micelles were found. The study additionally highlights the importance of using independent methods to characterize a rather complex colloidal system in order to obtain a complete picture of the structures formed.

Many scientists share preprints on social media platforms to gain attention from academic peers, policy-makers, and journalists. In this study we shed light on an unintended but highly consequential effect of sharing preprints: Their contribution to conspiracy theories. Although the scientific community might quickly dismiss a preprint as insubstantial and ‘clickbaity’, its uncertain epistemic status nevertheless allows conspiracy theorists to mobilize the text as scientific support for their own narratives. To better understand the epistemic politics of preprints on social media platforms, we studied the case of a biomedical preprint, which was shared widely and discussed controversially on Twitter in the wake of the coronavirus disease 2019 pandemic. Using a combination of social network analysis and qualitative content analysis, we compared the structures of engagement with the preprint and the discursive practices of scientists and conspiracy theorists. We found that despite substantial engagement, scientists were unable to dampen the conspiracy theorists’ enthusiasm for the preprint. We further found that members from both groups not only tried to reduce the preprint's epistemic uncertainty but sometimes deliberately maintained it. The maintenance of epistemic uncertainty helped conspiracy theorists to reinforce their group's identity as skeptics and allowed scientists to express concerns with the state of their profession. Our study contributes to research on the intricate relations between scientific knowledge and conspiracy theories online, as well as the role of social media platforms for new genres of scholarly communication.

In addition to erectile dysfunction, urinary incontinence is the most common functional limitation after radical prostatectomy (RPE) for prostate cancer (PCa). The German S3 guideline recommends informing patients about possible effects of the therapy options, including incontinence. However, only little data on continence from routine care in German-speaking countries after RPE are currently available, which makes it difficult to inform patients. The aim of this work is to present data on the frequency and severity of urinary incontinence after RPE from routine care. Information from the PCO (Prostate Cancer Outcomes) study is used, which was collected between 2016 and 2022 in 125 German Cancer Society (DKG)-certified prostate cancer centers in 17,149 patients using the Expanded Prostate Cancer Index Composite Short Form (EPIC-26). Changes in the “incontinence” score before (T0) and 12 months after RPE (T1) and the proportion of patients who used pads, stratified by age and risk group, are reported. The average score for urinary incontinence (value range: 0—worst possible to 100—best possible) was 93 points at T0 and 73 points 12 months later. At T0, 97% of the patients did not use a pad, compared to 56% at T1. 43% of the patients who did not use a pad before surgery used at least one pad a day 12 months later, while 13% use two or more. The proportion of patients using pads differs by age and risk classification. The results provide a comprehensive insight into functional outcome 12 months after RPE and can be taken into account when informing patients.

Understanding the evolutionary mechanisms underlying the maintenance of individual differences in behavior and physiology is a fundamental goal in ecology and evolution. The Pace-of-life syndrome hypothesis is often invoked to explain the maintenance of such within-population variation. This hypothesis predicts that behavioral traits are part of a suite of correlated traits that collectively determine an individual’s propensity to prioritize reproduction or survival. A key assumption of this hypothesis is that these traits are underpinned by genetic trade-offs among life-history traits: genetic variants that increase fertility, reproduction and growth might also reduce lifespan. We performed a systematic literature review and meta-analysis to summarize the evidence for the existence of genetic trade-offs between five key life-history traits: survival, growth rate, body size, maturation rate, and fertility. Counter to our predictions, we found an overall positive genetic correlation between survival and other life-history traits and no evidence for any genetic correlations between the non-survival life-history traits. This finding was generally consistent across pairs of life-history traits, sexes, life stages, lab vs field studies, and narrow- vs broad-sense correlation estimates. Our study highlights that genetic trade-offs may not be as common, or at least not as easily quantifiable, in animals as often assumed.

With the growing digitalization all over the globe, the relevance of network security becomes increasingly important. Machine learning-based intrusion detection constitutes a promising approach for improving security, but it bears several challenges. These include the requirement to detect novel and unseen network events, as well as specific data properties, such as events over time together with the inherent graph structure of network communication. In this work, we propose a novel intrusion detection method, TGN-SVDD, which builds upon modern dynamic graph modelling and deep anomaly detection. We demonstrate its superiority over several baselines for realistic intrusion detection data and suggest a more challenging variant of the latter.

When encountering social robots, potential users are often facing a dilemma between privacy and utility. That is, high utility often comes at the cost of lenient privacy settings, allowing the robot to store personal data and to connect to the internet permanently, which brings in associated data security risks. However, to date, it still remains unclear how this dilemma affects attitudes and behavioral intentions towards the respective robot. To shed light on the influence of a social robot’s privacy settings on robot-related attitudes and behavioral intentions, we conducted two online experiments with a total sample of N = 320 German university students. We hypothesized that strict privacy settings compared to lenient privacy settings of a social robot would result in more favorable attitudes and behavioral intentions towards the robot in Experiment 1. For Experiment 2, we expected more favorable attitudes and behavioral intentions for choosing independently the robot’s privacy settings in comparison to evaluating preset privacy settings. However, those two manipulations seemed to influence attitudes towards the robot in diverging domains: While strict privacy settings increased trust, decreased subjective ambivalence and increased the willingness to self-disclose compared to lenient privacy settings, the choice of privacy settings seemed to primarily impact robot likeability, contact intentions and the depth of potential self-disclosure. Strict compared to lenient privacy settings might reduce the risk associated with robot contact and thereby also reduce risk-related attitudes and increase trust-dependent behavioral intentions. However, if allowed to choose, people make the robot ‘their own’, through making a privacy-utility tradeoff. This tradeoff is likely a compromise between full privacy and full utility and thus does not reduce risks of robot-contact as much as strict privacy settings do. Future experiments should replicate these results using real-life human robot interaction and different scenarios to further investigate the psychological mechanisms causing such divergences.

Zusammenfassung Patient*innen mit Migrationsgeschichte stoßen im deutschen Gesundheitssystem vielfach auf Zugangsbarrieren, die die Qualität der ihnen zugänglichen Versorgung mindern und ihre Gesundheit beeinträchtigen. Diese Barrieren haben einerseits politische Ursachen, sind jedoch auch auf einen Mangel an migrations- und diversitätsbezogenen Inhalten im Medizinstudium und in anderen gesundheitsbezogenen Studien- und Ausbildungsgängen zurückzuführen. Obwohl die Versorgung von Patientinnen und Patienten mit eigener oder familiärer Migrationsgeschichte zum Alltag gehört, sind dafür relevante Inhalte bislang nicht in den Curricula verankert und werden bestenfalls in Form von Wahlpflichtfächern oder anderen fakultativen Lehrangeboten vermittelt. Um diese Situation zu verbessern und eine menschenrechtsbasierte, diversitätssensible und Equity-orientierte Weiterentwicklung der Curricula voranzutreiben, hat sich das „Lehrnetzwerk Migration und Gesundheit“ gegründet. Es zielt darauf ab, 1) in der Lehre aktive Personen miteinander zu vernetzen und den Austausch sowie die gemeinsame Weiterentwicklung von Lehrmaterial zu fördern, 2) darauf aufbauend einen Modellkurs „Migration und Gesundheit“ zu entwickeln und 3) Strategien für die longitudinale Implementierung entsprechender Inhalte in Pflichtcurricula zu erarbeiten. Diese Bestrebungen werden von Lehrforschung flankiert. An Mitarbeit im Lehrnetzwerk Interessierte sind herzlich eingeladen, die Autor*innen zu kontaktieren und an diesen Vorhaben mitzuwirken.

Zusammenfassung Rassismus und Diskriminierung als soziale Determinanten der Gesundheit stehen auch in Deutschland zunehmend im Fokus der Public-Health-Forschung. Studien zeigen Zusammenhänge mit physischer und psychischer Gesundheit bis hin zu Veränderungen auf zellulärer Ebene auf. Neben den gesundheitsschädigenden Effekten interpersoneller und direkter Diskriminierung ist die Relevanz des strukturellen und institutionellen Rassismus für die gesundheitliche Ungleichheit bislang nur wenig beleuchtet. Im Rahmen einer narrativen Übersichtsarbeit werden relevante und aktuelle Forschungsergebnisse zusammengestellt und kritisch diskutiert sowie Handlungsempfehlungen für Forschung und Praxis abgeleitet. Strukturelle und institutionelle Aspekte von Diskriminierung und Rassismus stehen in engem Zusammenhang mit der gesundheitlichen Lage. So steht die systematische Benachteiligung in den Bereichen Bildung, Arbeit, Wohnen sowie Gesundheitsversorgung im Zusammenhang mit der allgemeinen, psychischen und physischen Gesundheit, mit der Inanspruchnahme von Präventions- und Versorgungsleistungen sowie mit dem Gesundheitsverhalten. Eine Analyse der Verschränkung von Lebens‑, Wohn- und Arbeitsbedingungen mit der gesundheitlichen Lage von Menschen mit (und ohne) Migrationsgeschichte – generell und in Verbindung mit Rassismus und Diskriminierung – erscheint notwendig, um gezielte Maßnahmen im Hinblick auf Verhältnisprävention abzuleiten, statt auf bloße Verhaltensprävention zu fokussieren. Neben praktischen Interventionen (Trainings, Aufklärungsarbeit, communitybasierten Ansätzen) ist die Weiterentwicklung methodischer Aspekte im Bereich der Erhebung und Analyse von Daten wichtig, um dieser Problemlage umfassend in Forschung und Praxis zu begegnen.

The main strength of systems theory is that it can make the different types of logic in different kinds of social systems comprehensible and implement them in its analyses. In systems theory, only two forms of differentiation are ever dealt with in the treatment of modern societies: the functional differentiation between different social fields such as the economy, politics, law, science or mass media, and the level differentiation between interactions, organizations and society.

Explainable artificial intelligence has mainly focused on static learning scenarios so far. We are interested in dynamic scenarios where data is sampled progressively, and learning is done in an incremental rather than a batch mode. We seek efficient incremental algorithms for computing feature importance (FI). Permutation feature importance (PFI) is a well-established model-agnostic measure to obtain global FI based on feature marginalization of absent features. We propose an efficient, model-agnostic algorithm called iPFI to estimate this measure incrementally and under dynamic modeling conditions including concept drift. We prove theoretical guarantees on the approximation quality in terms of expectation and variance. To validate our theoretical findings and the efficacy of our approaches in incremental scenarios dealing with streaming data rather than traditional batch settings, we conduct multiple experimental studies on benchmark data with and without concept drift.