The focus of this thesis is on the technical methods which help promote the movement towards Trustworthy AI, specifically within the Inspectorate of the Netherlands.The goal is develop and assess... Show moreThe focus of this thesis is on the technical methods which help promote the movement towards Trustworthy AI, specifically within the Inspectorate of the Netherlands.The goal is develop and assess the technical methods which are required to shift the actions of the Inspectorate to a data-driven paradigm, concretely under a supervised classification framework of machine learning.The aspect of reliability is addressed as a data quality concern, viz. missingness and noise.The aspect of fairness is addressed as a counter to bias in the selection process of inspections.The conclusion is that, whilst no complete solution has yet been suggested, it is possible to address the concerns related to data quality and data bias, culminating in well-performing classification models which are reliable and fair. Show less
This doctoral dissertation explores classical music performance from a curatorial perspective, reflecting upon and challenging the traditional configuration of performance environments. Beginning... Show moreThis doctoral dissertation explores classical music performance from a curatorial perspective, reflecting upon and challenging the traditional configuration of performance environments. Beginning with a consideration for the historical origins of absorbed attention and silence as the dominant mode of performing and hearing classical music, the subsequent chapters of this dissertation investigate alternatives to this mode by exploring artistic creations developed during this research. Informed by my combined experience as a curator and performer in the contemporary music field, these artistic creations use what I call ‘metaxical amplification’: the amplification of environmental sounds that are generally considered noise in the context of classical music performances, and that are therefore rarely considered in relation to the artistic experiences generated by these performances. Metaxical amplification proposes a reconfiguration of the performance environment and the ways in which attention unfolds within it. It also challenges a work-centred performance culture, since the performance mode emerging from this form of amplification is oriented towards the sonic exploration of musical environments through these works. More broadly, it propels the development of a practice in which musical interpretation, improvisation, and curatorial thinking are tightly interwoven. Findings are discussed in close dialogue with literature from various fields including sociology, philosophy, media theory, as well as through related examples from the fields of music, theatre, and the visual arts. Show less
Objectives: Speech understanding in noise is difficult for patients with a cochlear implant. One common and disruptive type of noise is transient noise. We have tested transient noise reduction ... Show moreObjectives: Speech understanding in noise is difficult for patients with a cochlear implant. One common and disruptive type of noise is transient noise. We have tested transient noise reduction (TNR) algorithms in cochlear implant users to investigate the merits of personalizing the noise reduction settings based on a subject's own preference. Design: The effect of personalizing two parameters of a broadband and a multiband TNR algorithm (TNRbb and TNRmb, respectively) on speech recognition was tested in a group of 15 unilaterally implanted subjects in cafeteria noise. The noise consisted of a combination of clattering dishes and babble noise. Each participant could individually vary two parameters, namely the scaling factor of the attenuation and the release time (tau). The parameter tau represents the duration of the attenuation applied after a transient is detected. As a reference, the current clinical standard TNR "SoundRelax" from Advanced Bionics was tested (TNRbb-std). Effectiveness of the algorithms on speech recognition was evaluated adaptively by determining the speech reception threshold (SRT). Possible subjective benefits of the algorithms were assessed using a rating task at a fixed signal-to-noise ratio (SNR) of SRT + 3 dB. Rating was performed on four items, namely speech intelligibility, speech naturalness, listening effort, and annoyance of the noise. Word correct scores were determined at these fixed speech levels as well. Results: The personalized TNRmb improved the SRT statistically significantly with 1.3 dB, while the personalized TNRbb degraded it significantly by 1.7 dB. For TNRmb, we attempted to further optimize its settings by determining a group-based setting, leaving out those subjects that did not experience a benefit from it. Using these group-based settings, however, TNRmb did not have a significant effect on the SRT any longer. TNRbb-std did not affect speech recognition significantly. No significant effects on subjective ratings were found for any of the items investigated. In addition, at a constant speech level of SRT + 3 dB, no effect of any of the algorithms was found on word correct scores, including TNRmb with personalized settings. Conclusions: Our study results indicate that personalizing noise reduction settings of a multiband TNR algorithm can significantly improve speech intelligibility in transient noise, but only under challenging listening conditions around the SRT. At more favorable SNRs (SRT + 3 dB), this benefit was lost. We hypothesize that TNRmb was beneficial at lower SNRs, because of more effective artifact detection under those conditions. Group-averaged settings of the multiband algorithm did not significantly affect speech recognition. TNRbb decreased speech recognition significantly using personalized parameter settings. Rating scores were not significantly affected by the algorithms under any condition tested. The currently available TNR algorithm for Advanced Bionics systems (SoundRelax) is a broadband filter that does not support personalization of its settings. Future iterations of this algorithm might benefit from upgrading it to a multiband variant with the option to personalize its parameter settings. Show less
Introduction: Contralateral routing of signals (CROS) can be used to eliminate the head shadow effect. In unilateral cochlear implant (CI) users, CROS can be achieved with placement of a microphone... Show moreIntroduction: Contralateral routing of signals (CROS) can be used to eliminate the head shadow effect. In unilateral cochlear implant (CI) users, CROS can be achieved with placement of a microphone on the contralateral ear, with the signal streamed to the CI ear. CROS was originally developed for unilateral CI users without any residual hearing in the nonimplanted ear. However, the criteria for implantation are becoming progressively looser, and the nonimplanted ear can have substantial residual hearing. In this study, we assessed how residual hearing in the contralateral ear influences CROS effectiveness in unilateral CI users. Methods: In a group of unilateral CI users (N = 17) with varying amounts of residual hearing, we deployed free-field speech tests to determine the effects of CROS on the speech reception threshold (SRT) in amplitude-modulated noise. We compared 2 spatial configurations: (1) speech presented to the CROS ear and noise to the CI ear (SCROSNCI) and (2) the reverse (SCINCROS). Results: Compared with the use of CI only, CROS improved the SRT by 6.4 dB on average in the SCROSNCI configuration. In the SCINCROS configuration, however, CROS deteriorated the SRT by 8.4 dB. The benefit and disadvantage of CROS both decreased significantly with the amount of residual hearing. Conclusion: CROS users need careful instructions about the potential disadvantage when listening in conditions where the CROS ear mainly receives noise, especially if they have residual hearing in the contralateral ear. The CROS device should be turned off when it is on the noise side (SCINCROS). CI users with residual hearing in the CROS ear also should understand that contralateral amplification (i.e., a bimodal hearing solution) will yield better results than a CROS device. Unilateral CI users with no functional contralateral hearing should be considered the primary target population for a CROS device. Show less
Objectives:The ability to perceive soft speech by cochlear implant (CI) users is restricted in part by the inherent system noise produced by the speech processor, and in particular by the... Show moreObjectives:The ability to perceive soft speech by cochlear implant (CI) users is restricted in part by the inherent system noise produced by the speech processor, and in particular by the microphone(s). The algorithm "SoftVoice" (SV) was developed by Advanced Bionics to enhance the perception of soft speech by reducing the system noise in speech processors. The aim of this study was to examine the effects of SV on speech recognition and listening effort.Design:Seventeen adult Advanced Bionics CI recipients were recruited and tested in two sessions. The effect of SV on speech recognition was tested by determining the SRT in quiet using the Matrix test. Based on the individual subjects' SRTs, we investigated speech-recognition scores at fixed speech levels, namely SRT -5 dB, SRT +0 dB, SRT +5 dB, and SRT +10 dB, again in quiet and using the Matrix test. Listening effort was measured at each of these speech levels subjectively by using a rating scale, and objectively by determining pupil dilation with pupillometry. To verify whether SoftVoice had any negative effects on speech perception in noise, we determined the SRT in steady state, speech-weighted noise of 60 dBA.Results:Our results revealed a significant improvement of 2.0 dB on the SRT in quiet with SoftVoice. The average SRT in quiet without SoftVoice was 38 dBA. SoftVoice did not affect the SRT in steady state, speech-weighted noise of 60 dB. At an average speech level of 33 dBA (SRT -5 dB) and 38 dBA (SRT +0 dB) in quiet, significant improvements of 17% and 9% on speech-recognition scores were found with SoftVoice, respectively. At higher speech levels, SoftVoice did not significantly affect speech recognition. Pupillometry did not show significant effects of SoftVoice at any speech level. However, subjective ratings of listening effort indicated a decrease of listening effort with SoftVoice at a speech level of 33 dBA.Conclusions:We conclude that SoftVoice substantially improves recognition of soft speech and lowers subjective listening effort at low speech levels in quiet. However, no significant effect of SoftVoice was found on pupil dilation. As SRTs in noise were not statistically significantly affected by SoftVoice, we conclude that SoftVoice can be used in noisy listening conditions with little negative impact on speech recognition, if any. The increased power demands of the algorithm are considered to be negligible. It is expected that SoftVoice will reduce power consumption at low ambient sound levels. These results support the use of SoftVoice as a standard feature of Advanced Bionics CI fittings for everyday use. Show less
This conversation takes Katrine Dirckinck-Holmfeld’s installation and perfor- mative presentation The Christmas Report & Other Fragments (2017) as a start- ing point to discuss legibility in... Show moreThis conversation takes Katrine Dirckinck-Holmfeld’s installation and perfor- mative presentation The Christmas Report & Other Fragments (2017) as a start- ing point to discuss legibility in relation to the mass digitization of the colonial archives in Denmark. To gain access to the archive, Dirckinck-Holmfeld draws on the figure of the Data Thief, inspired by The Black Audio Film Collective, in an attempt to unearth and excel the vulnerabilities and ethical dilemmas at the heart of today’s data desire. The Data Thief, Dirckinck-Holmfeld claims in conversation with Pepita Hesselberth, teaches us to attune to the noise, to the sonorous, affective and textural dimensions of the archive. It compels us to create assemblages of enunciation that cut across semiotic and machinic flows, and invites us to nourish a relationship to time where the past keeps enfolding on itself in the present. This way, she concludes, it demands us to stay in and with the discomfort, and to stay in the cybernetic fold of radical, creative, decolonial and technological reimagination. Show less
Werner, C.D.; Linting, M.; Vermeer, H.J.; IJzendoorn, M.H. van 2015
Real-world (black-box) optimization problems often involve various types of uncertainties and noise emerging in different parts of the optimization problem. When this is not accounted for,... Show moreReal-world (black-box) optimization problems often involve various types of uncertainties and noise emerging in different parts of the optimization problem. When this is not accounted for, optimization may fail or may yield solutions that are optimal in the classical strict notion of optimality, but fail in practice. Robust optimization is the practice of optimization that actively accounts for uncertainties and/or noise. Evolutionary Algorithms form a class of optimization algorithms that use the principle of evolution to find good solutions to optimization problems. Because uncertainty and noise are indispensable parts of nature, this class of optimization algorithms seems to be a logical choice for robust optimization scenarios. This thesis provides a clear definition of the term robust optimization and a comparison and practical guidelines on how Evolution Strategies, a subclass of Evolutionary Algorithms for real-parameter optimization problems, should be adapted for such scenarios. Show less
Stochastic differential equations with delay are the inspiration for this thesis. Examples of such equations arise in population models, control systems with delay and noise, lasers, economical... Show moreStochastic differential equations with delay are the inspiration for this thesis. Examples of such equations arise in population models, control systems with delay and noise, lasers, economical models, neural networks, environmental pollution and in many other situations. In such models we are often interested in the evolution of a particular quantity, for example the size of a population, or the amount of pollution in a particular area, changing in time. A differential equation with delay, or delay equation, is a differential equation in which the change in time of such a quantity is expressed as a function of the value of that quantity at different points in time, in the past as well as in the present. This is in contrast with an ordinary differential equation, in which the change in time of the quantity at a specific time is expressed as a function of that quantity at that specific time only. Show less
