83 research outputs found
Informationsextraktion für datengetriebenes Indoor-Tracking mit Ultra-Breitbandsignalen
No full textIn indoor environments, emerging applications such robot navigation or industrial process
surveillance rely on accurate radio frequency ( RF)-based tracking. It employs the commu-
nication links between mobile agents and stationary anchors to infer their position. A major
challenge in classical multilateration-based approaches is the influence of the environment on
signal propagation, as furniture and structural components cause multipath propagation like
scattering and reflections. As a solution, ultra-wideband (UWB )-signalling resolves multipath
components ( MPCs) by collecting channel measurements (CMs) that contain additional spa-
tial spatial information on the propagation. This has the potential to reduce deployment costs
for RF-based tracking systems as it lowers the necessary anchor infrastructure. To exploit
this information for tracking, however, it first has to be extracted from the CMs. The limits of
the channel hereby cause overlap between MPCs and non-linear distortions. State-of-the-art
data-driven machine learning ( ML ) methods, especially neural networks ( NNs), learn efficient
function approximations from training data that can handle these non-linear effects. However,
for cost-efficient deployment of these data-driven information extraction models to a target en-
vironment, they need to generalize from environment-representative training data. This thesis
proposes different NNs that learn to extract spatial information in industrial environments
with harsh propagation conditions from data obtained in different environments. Specifically,
these models learn how to estimate the line-of-sight ( LOS) presence based on data from
artificial labeling environments using convolutional neural networks (CNNs) and variational
autoencoders (VAE s)-based anomality detection. Because CM contains an arbitrary num-
ber of MPC, MPC delay extraction is not a straightforward classification or regression task
and state-of-the-art methods rely on computationally expensive iterative statistical signal
processing. Instead, this thesis proposes a method that learns to accurately extract MPC
delays from simulated data with a time-series segmentation approach relying on a compu-
tationally efficient UNET CNN. A lower-dimensional representation of CM is achieved
with propagation-model inspired features in the state-of-the-art. As a data-driven alternative,
this thesis proposes a representation learning approach that learns to compress the spatial
information in CM into a lower-dimensional latent-space representation from data obtained
in similar industrial environments with an autoencoder (AE). Tracking results prove that this
representation results in more accurate tracking than state-of-the-art features. Thus, compared
to the state-of-the-art, the proposed spatial information extraction learn to accurately represent
the spatial information in CM from easily obtainable data. The generalization abilities of the
proposed methods reduce the need for cost- and labor-intensive data collection and enable
the deployment of trained models to an application environments with harsh propagation
conditions. Furthermore, this information extraction compresses high-dimensional CMs into a lower-dimensional representation that reduces data storage and transmission requirements
in a CM-based tracking system. To exploit this extracted information for tracking, this thesis
proposes a particle filter for fingerprinting (FP) that fuses the extracted LOS presence indicator
and a Gaussian process regression (GPR)-based observation likelihood model of the extracted
latent-space representation. As it employs this additional, compressed spatial information
it enables accurate tracking in an indoor environments with harsh propagation conditions,
unlike state-of-the-art LOS-focused tracking methods. FP is trained on environment-specific
data that are hard to acquire and maintain. The proposed tracking methods lowers the data
collection and maintenance effort over state-of-the-art CM-based FP because the learned
observation likelihood model implies a reliability measure, and thus can learn from spatially
sparse datasets that only contain data instances in areas within the environment that require
FP for accurate tracking. Real-world evaluation shows that the proposed tracking method
achieves more reliable information extraction and tracking than state-of-the-art methods,
especially on a sparse FP database and with a small anchor infrastructure. The proposed
information extraction reduces transmission and data storage requirements and generalizes
to target environments at low data acquisition effort. So, is enables cost- and energy effi-
cient deployment of data-driven CM-based positioning in indoor environments with harsh
propagation conditions
Ghosts of Manila the fateful blood feud between Muhammad Ali and Joe Frazier
No full textThe author tells the story of the Ali-Frazier matchup in Manila. "When the 'Thrilla in Manila' was over, the hype no longer mattered: one man was left with a ruin of a life; the other was battered to his soul."--Jacket
The effective mentor, mentee and mentoring relationship
No full textThe aim of this chapter is to examine the various competency frameworks for mentors and mentees and consider the requirements for an effective mentoring relationship, exploring theoretical and empirical studies as well as conceptual models and frameworks. The chapter begins by outlining the behaviours, capabilities and characteristics of mentors and mentees drawing on current literature (Cooper & Palmer, 2000; Clutterbuck, 2004, 2011; Brockbank & McGill, 2006, Allen & Eby, 2011). These are compared and contrasted, taking into account methodological issues such as the significance of context (Kram, 1988; Bierema & Merriam, 2002; Fowler & O’Gorman, 2005; Ghosh, 2012), purpose and type of mentoring (Kram, 1980, 1985; Ragins & Cotton, 1999; Clutterbuck, 1998, 2015) and that competences may evolve through the different phases of the mentor-mentee relationship (Missiran, 1982; Kram, 1983; Clutterbuck, 1995, 1998).
In addition, the author recognises the need to consider the complex adaptive system (Mitleton-Kelly, 1997; Lansing, 2003; Clutterbuck, 2012) in which the mentor-mentee relationship is established and developed. Next, the author examines the measures for the effectiveness of a mentoring relationship, with particular reference to how this might be useful in the initiation, support and measurement of mentoring outcomes. Finally, the author offers recommendations for future research
Robust Concept Erasure via Kernelized Rate-Distortion Maximization
No full textDistributed representations provide a vector space that captures meaningful
relationships between data instances. The distributed nature of these
representations, however, entangles together multiple attributes or concepts of
data instances (e.g., the topic or sentiment of a text, characteristics of the
author (age, gender, etc), etc). Recent work has proposed the task of concept
erasure, in which rather than making a concept predictable, the goal is to
remove an attribute from distributed representations while retaining other
information from the original representation space as much as possible. In this
paper, we propose a new distance metric learning-based objective, the
Kernelized Rate-Distortion Maximizer (KRaM), for performing concept erasure.
KRaM fits a transformation of representations to match a specified distance
measure (defined by a labeled concept to erase) using a modified
rate-distortion function. Specifically, KRaM's objective function aims to make
instances with similar concept labels dissimilar in the learned representation
space while retaining other information. We find that optimizing KRaM
effectively erases various types of concepts: categorical, continuous, and
vector-valued variables from data representations across diverse domains. We
also provide a theoretical analysis of several properties of KRaM's objective.
To assess the quality of the learned representations, we propose an alignment
score to evaluate their similarity with the original representation space.
Additionally, we conduct experiments to showcase KRaM's efficacy in various
settings, from erasing binary gender variables in word embeddings to
vector-valued variables in GPT-3 representations.Comment: NeurIPS 202
As Time Goes by: EU Climate Change Actorness from Rio to Copenhagen. Bruges Regional Integration & Global Governance Paper 3/2011, September 2011
No full textDuring the past two decades the European Union (EU) has increasingly come to be recognised as an important international actor in environmental politics. The failure of the EU to instigate an ambitious post-2012 environmental framework agreement at the Fifteenth Conference of the Parties (COP15) to the United Nations Framework Convention on Climate Change (UNFCCC) in Copenhagen in 2009 may, however, signal a change in the EU’s status as an international climate change actor. It raises the question of which conditions allowed the EU to be an actor in the first place. Drawing on the theoretical concept of actorness, the paper analyses the conditions for EU actorness in the area of climate change. It will be argued that for the EU to be an actor, all four criteria of actorness – recognition, authority, cohesion and autonomy – need to be present. While these criteria were present at the 1992 Rio Summit and the COP3 in Kyoto in 1997, a lack of autonomy and cohesion prevented the EU from being an international actor in Copenhagen
The RADAR NAMs Toolbox: In Vitro Methods for Efficient Identification of Safe and Sustainable Aromatic Compounds as Substitutes for Chemi- cals of Concern
No full textHarmful fossil-based chemicals are widespread in consumer products and can pose significant risks to human health and the environment due to their endocrine-disrupting and carcinogenic properties. In response, the EU promotes the development of safer, bio-based alternatives. The RADAR project (“Renewable and safe Aromatic CompounDs As Replacements for substances of concern”) aims to address this issue by valorizing wood waste to produce ligninderived aromatic compounds as sustainable replacements for harmful substances. This approach focuses on synthesizing bio-based phenolics from purified lignin oil for use in applications such as can coatings, flame retardants, and surfactants. Our objective is to evaluate the safety of these novel compounds, specifically their toxicity and endocrine effects, using a comprehensive toolbox of screening-compatible New Approach Methodologies (NAMs). The results will be compared with those of currently used chemicals, with particular emphasis on cell toxicity and effects on steroid and thyroid hormones. A range of imaging and molecular biological techniques will be employed to investigate these potential health risks
Approximation of enzyme kinetics for high enzyme concentration by a first order perturbation approach
No full textUWB Channel Impulse Responses for Positioning in Complex Environments: A Detailed Feature Analysis
No full textRadio signal-based positioning in environments with complex propagation paths is a challenging task for classical positioning methods. For example, in a typical industrial environment, objects such as machines and workpieces cause reflections, diffractions, and absorptions, which are not taken into account by classical lateration methods and may lead to erroneous positions. Only a few data-driven methods developed in recent years can deal with these irregularities in the propagation paths or use them as additional information for positioning. These methods exploit the channel impulse responses (CIR) that are detected by ultra-wideband radio systems for positioning. These CIRs embed the signal properties of the underlying propagation paths that represent the environment. This article describes a feature-based localization approach that exploits machine-learning to derive characteristic information of the CIR signal for positioning. The approach is complete without highly time-synchronized receiver or arrival times. Various features were investigated based on signal propagation models for complex environments. These features were then assessed qualitatively based on their spatial relationship to objects and their contribution to a more accurate position estimation. Three datasets collected in environments of varying degrees of complexity were analyzed. The evaluation of the experiments showed that a clear relationship between the features and the environment indicates that features in complex propagation environments improve positional accuracy. A quantitative assessment of the features was made based on a hierarchical classification of stratified regions within the environment. Classification accuracies of over 90% could be achieved for region sizes of about 0.1 m 2 . An application-driven evaluation was made to distinguish between different screwing processes on a car door based on CIR measures. While in a static environment, even with a single infrastructure tag, nearly error-free classification could be achieved, the accuracy of changes in the environment decreases rapidly. To adapt to changes in the environment, the models were retrained with a small amount of CIR data. This increased performance considerably. The proposed approach results in highly accurate classification, even with a reduced infrastructure of one or two tags, and is easily adaptable to new environments. In addition, the approach does not require calibration or synchronization of the positioning system or the installation of a reference system
Rare plant conservation planning workshop results: Middle Park
No full textThis document identifies conservation strategies for Penland penstemon and Kremmling milkvetch, based on an assessment of the plants' viability and threats by participants of a June 2008 workshop. The primary audience is intended to be the workshop participants and other stakeholders interested in helping to implement the strategies.Sponsored by the Colorado Rare Plant Conservation Initiative, June 26, 2008
Rare plant conservation planning workshop results: North Park
No full textThis document identifies conservation strategies for the North Park facelift, based on an assessment of the plants' viability and threats by participants of workshops in Summer 2008. The primary audience is intended to be the workshop participants and other stakeholders interested in helping to implement the strategies.Sponsored by the Colorado Rare Plant Conservation Initiative, August 21, 2008
- …
