Ministry for Environment, Agriculture, Conservation and Consumer Protection
Digital Peer PublishingNot a member yet
2777 research outputs found
Sort by
Imbalanced data as risk factor of discriminating automated decisions: a measurement-based approach
No full textOver the last two decades, the number of organizations -both in the public and private sector- which have automated decisional processes has grown notably. The phenomenon has been enabled by the availability of massive amounts of personal data and the development of software systems that use those data to optimize decisions with respect to certain optimization goals. Today, software systems are involved in a wide realm of decisions that are relevant for the lives of people and the exercise of their rights and freedoms. Illustrative examples are systems that score individuals for their possibility to pay back a debt, recommenders of the best candidates for a job or a house rent advertisement, or tools for automatic moderation of online debates.
While advantages for using algorithmic decision making concern mainly scalability and economic affordability, on the other hand, several critical aspects have emerged, including systematic adverse impact for individuals belonging to minorities and disadvantaged groups. In this context, the terms data and algorithm bias have become familiar to researchers, industry leaders and policy makers, and much ink has been spelled on the concept of algorithm fairness, in order to produce more equitable results and to avoid discrimination. Our approach is different from the main corpus of research on algorithm fairness because we shift the focus from the outcomes of automated decision making systems to its inputs and processes. Instead, we lay the foundations of a risk assessment approach based on a measurable characteristic of input data, i.e. imbalance, which can lead to discriminating automated decisions. We then relate the imbalance to existing standards and risk assessment procedures.
We believe that the proposed approach can be useful to a variety of stakeholders, e.g. producers and adopters of automated decision making software, policy makers, certification or audit authorities. This would allow for the assessment of the risk level of discriminations when using imbalanced data in decision making software. This assessment should prompt all the involved stakeholders to take appropriate actions to prevent adverse effects. Such discriminations, in fact, pose a significant obstacle to human rights and freedoms, as our societies increasingly rely on automated decision making. This work is intended to help mitigate this problem, and to contribute to the development of software systems that are socially sustainable and are in line with the shared values of our democratic societies.
This paper is also available at https://doi.org/10.5281/zenodo.579518
Possession adnominale dans différentes variétés de wolof
No full textZiel des vorliegenden Aufsatzes ist es, die Unterschiedlichkeit verschiedener noch wenig bekannter und kaum untersuchter Varietäten des Wolof auf morphologischer und syntaktischer Ebene aufzuzeigen. Dies geschieht anhand zweier adnominaler Possessionskonstruktionen: die besitzanzeigende und die genitivische. Die Beschreibung und der Vergleich dieser Konstruktionen weisen in den verschiedenen Varietäten bedeutende Unterschiede auf, sowohl in Bezug auf die verwendeten Morpheme als auch in Bezug auf den Gebrauch anscheinend identischer Morpheme in Verbindung mit je nach Dialekt unterschiedlichen Konstruktionsregeln.The aim of this article is to show the diversity of several varieties of Wolof, which are still little known and very poorly exploited, by showing clear differences between certain dialectal varieties, at the morphological and syntactic levels. The field in which this diversity will be dealt with is the adnominal possession through two structures: the possessive construction and the genitive construction. The description and comparison of these constructions in several varieties show quite important differences in the morphemes used or in the use of apparently identical morphemes leading to quite different rules of construction according to the dialects.L’objectif de cet article est de faire état de la réelle diversité de la langue wolof, encore trop méconnue et très mal exploitée en montrant des différences claires entre certaines variétés dialectales, au niveau morphologique et syntaxique. Le domaine dans lequel cette diversité sera traitée est celui de la possession adnominale à travers deux structures ; la construction à possessif et la construction génitivale. La description et la comparaison de ces constructions dans plusieurs variétés montrent des différences assez importantes dans les morphèmes utilisés ou dans l’utilisation de morphèmes en apparence identique aboutissant à des règles de constructions assez différentes selon les dialectes
From Theory To Practice: Exercising The Right Of Access Under The Law Enforcement And PNR Directives
No full textThe right of access is often considered as the most important prerogative in the data subject’s toolkit because it grants individuals the possibility to complement the information made available through privacy notices, but also because it paves the way for the exercise of other rights enshrined in data protection law, such as the rights to erasure or rectification. While the efficiency of the right of access under the General Data Protection Regulation has already been abundantly documented, there is a lack of empirical evidence as to its counterparts in the area of law enforcement and security. This contribution aims to fill that gap and provide insight into the practical exercise of the right of access in the Law Enforcement and Passenger Name Record Directives. Through both traditional desktop research and a legal-empirical study, the present paper delves into the national transpositions of those texts in a selection of Member States, and highlights the issues encountered when practically exercising the right of access against competent authorities and Passenger Information Units. It also draws upon the lessons learned from that exercise and suggests solutions and ways forward in order to overcome the obstacles faced along the way
International Law Association’s Guidelines on Intellectual Property and Private International Law (“Kyoto Guidelines”): Jurisdiction
No full textThe chapter “Jurisdiction” of the International Law Association’s Guidelines on Intellectual Property and Private International Law (“Kyoto Guidelines”) provides where international intellectual property claims can be brought to court. It defines the basic forum at the defendants domicile and alternative fora for contractual, infringement and other kinds of claims. It also provides grounds that should be taken as insufficient for the granting of jurisdiction. The chapter states to what extent validity and registration claims should be subject to exclusive jurisdiction. Finally, it comprises Guidelines for the coordination of claims pending before different courts
Cyber-Physical Twin Framework for Generating Human Movement Data in Intralogistics
No full textMenschliche Bewegungen zu erkennen, sie zu deuten und für die Analyse manueller Prozesse relevanten Aktivitäten zuzuordnen sind zentrale Herausforderungen der Human Activity Recognition (HAR). Diesen Herausforderungen geht das Trainieren eines Klassifikators mit Daten voraus. Die Erstellung dieser Trainingsdatensätze, bestehend aus Datenaufnahme, Annotation und Revision von Zeitreihen, bedingt einen immensen Aufwand. Aus diesem Grund werden HAR-Methoden überwiegend an simplen Alltagssituationen getestet. Um HAR-Methoden auch für komplexe Umgebungen wie die Intralogistik entwickeln zu können, ist eine neue Form der Datensatzerstellung notwendig. Dieser Beitrag schlägt ein Framework vor, den Aufwand der Datenaufnahme durch Zuhilfenahme cyber-physischer Zwillinge von Menschen zu reduzieren.Recognizing human movements, interpreting them and assigning relevant activities for the analysis of manual processes are central challenges of Human Activity Recognition (HAR). These challenges are preceded by training a classifier with data. The creation of these training data sets, consisting of data acquisition, annotation and revision of time series, requires immense effort. For this reason, HAR methods are mainly tested on simple everyday situations. A new form of data set creation is necessary to develop HAR methods for complex environments such as intralogistics. This contribution proposes a framework to reduce the effort of data acquisition by using cyber-physical twins
Verwendung von tiefen neuronalen Netzen zur Messung von Pufferpegeln in Echtzeit mit Edge-Computing
No full textTechnological advances and increasing data traffic in the IoT environment lead to the relocation of sophisticated data processing to the edge of networks. At the same time, powerful object detection approaches based on deep neural networks have been developed in recent years. In this paper, an intelligent camera based on deep learning algorithms and consisting of low-cost hardware with limited computational and storage capacity is presented. The developed object detection solution enables real-time monitoring of the inventory of filled and empty small load carriers in a buffer zone.Der technologische Fortschritt und zunehmende Datenströme im IoT-Umfeld führen dazu, dass anspruchsvolle Datenverarbeitungsprozesse an den Rand von Netzwerken verlagert werden. Gleichzeitig wurden in den letzten Jahren leistungsfähige Objekterkennungsansätze entwickelt, die auf tiefen neuronalen Netzen basieren. Im Rahmen dieser Arbeit wird eine intelligente Kamera vorgestellt, welche auf Deep-Learning-Algorithmen basiert und aus kostengünstiger Hardware mit beschränkter Rechen- und Speicherkapazität besteht. Die entwickelte Objekterkennungslösung ermöglicht die Überwachung des Bestands von gefüllten und leeren Kleinladungsträgern in einer Pufferzone in Echtzeit
Databased surgery planning in hospitals under development and integration of a machine learning platform
No full textDie vorliegende Arbeit gibt einen Überblick zur Ergänzung von Planungs- und Steuerungsprozessen im OP-Bereich von Krankenhäusern. Der Fokus liegt hier auf der automatisierten Vorhersage von Operationszeiten mittels maschineller Lernmethoden auf Basis von retrospektiven Daten. Der vorliegende Artikel befasst sich mit der Analyse der zu beziehenden Daten, der Aufbereitung und Vorverarbeitung dieser und der Empfehlung an geeigneten Algorithmen für das M Maschinelle Lernen. Die beschriebene Arbeit ist im Rahmen des SAB geförderten Projektes; Platform for Operation Scheduling and Prediction using machine learning (PROSPER) entstanden.This paper gives an overview of the enhancement of planning and control processes in the operating theater of hospitals. The focus here is on the automated prediction of operating times using machine learning methods based on retrospective data. This article deals with the analysis of the data to be obtained, the preparation and preprocessing of this data and the recommendation of suitable algorithms for machine learning. This work was carried out as part of the funded project Platform for Operation Scheduling and Prediction using machine learning (PROSPER)
Technologies for tracking of load carriers and tools in project manufacturing
No full textDie ortsgebundene Fertigung zeichnet sich durch den Transport von Materialien, Arbeitskräften und Betriebsmitteln zum zu produzierenden Objekt aus. Damit verbunden sind nicht nur komplexe logistische Prozesse, sondern auch ein hoher Koordinationsaufwand. Zusätzlich besteht die Gefahr, dass es entlang der Prozesskette von der Kommissionierung über den Transport bis hin zur Bereitstellung der Materialien für die Fertigung zu Fehlern kommt, was zu einer mangelnden Versorgung der Fertigung führt. Dies kann aufwendige Suchprozessen zur Folge haben, wodurch Folgeprozesse beeinträchtigt werden können. Aufgrund der zahlreichen Tracking-Technologien und stetig sinkenden Investitionskosten ergeben sich neue Anwendungspotenziale, wie zum Beispiel im Bereich der ortsgebundenen Fertigung. Dieser Beitrag stellt die Ergebnisse der Untersuchung verschiedener Technologien zum Tracking von Ladungsträgern und Werkzeugen in der ortsgebundenen Fertigung vor.The project manufacturing is characterized by the transport of materials, workers and resources to the producing object. Associated with this are not only complex logistical processes, but also a higher coordination effort. In addition, there is a risk that work packages will be incorrectly assigned or tools will be relocated along the process chain, from order picking to transport to the provision of materials for production. This can lead to complex search processes, which can impair subsequent processes. Due to numerous tracking technologies and constantly falling investment costs, there are new application potentials, for example in the area of project manufacturing. This article presents the results of the investigations of various technologies for tracking load carriers and tools in project manufacturing