26838 research outputs found
Sort by
Joint demapping of QAM and APSK constellations using machine learning
Get PDFAs telecommunication systems evolve to meet increasing demands, integrating deep neural networks (DNNs) has shown promise in enhancing performance. However, the trade-off between accuracy and flexibility remains challenging when replacing traditional receivers with DNNs. This paper introduces a novel probabilistic framework that allows a single DNN demapper to demap multiple QAM and APSK constellations simultaneously. It is demonstrated that the framework can exploit hierarchical relationships in families of constellations. The consequence is that we need fewer neural network outputs to encode the same function without an increase in Bit Error Rate (BER). The simulation results confirm that the framework approaches the optimal demodulation error bound under an Additive White Gaussian Noise (AWGN) channel for multiple constellations. Under 3GPP-compliant OFDM fading channels, it is as accurate as a neural receiver operating on just one modulation type. Thereby, the framework addresses multiple important issues in practical neural receiver design. These include improvements in computational efficiency, a reduction in memory overhead, and an improved adaptability in dynamic environments
Comprehensive monitoring and prediction of seismicity within the Groningen gas field using large-scale field observations
Get PDFThis dissertation addresses the critical issue of induced seismicity within the Groningen gas field, one of the largest natural gas reservoirs globally. Since its discovery in the 1960s, the Groningen field has played a pivotal role in meeting the energy needs of the Netherlands. However, decades of sustained gas extraction have resulted in significant seismic activity in what was previously a tectonically inactive region. These induced earthquakes have raised pressing public safety concerns, necessitating a comprehensive understanding of the underlying processes, alongside the development of robust methodologies for monitoring and predicting seismic events. The aim of this research is to advance the monitoring, analysis, and prediction of seismicity by integrating statistical models with geomechanical insights and decision-making frameworks. Specifically, this study seeks to analyse the spatial and temporal patterns of seismic activity, predict future earthquake occurrences using spatio-temporal models, and optimise production strategies while mitigating seismic risk. Conducted within the framework of the DeepNL programme by the Dutch Research Council (NWO), this work forms part of a broader effort to deepen understanding of subsurface dynamics influenced by anthropogenic activities.
A multidisciplinary methodological approach was employed to achieve the study objectives. The research relies on an extensive dataset comprising seismic event records provided by KNMI (covering magnitudes ≥ 1.5 from 1995 to 2023), historical gas production data, pore pressure measurements from boreholes, and NAM’s dynamic reservoir model output. Rigorous data cleaning and integration procedures ensured the datasets’ reliability, which formed the basis for subsequent statistical analyses. The study begins with a detailed temporal and spatial analysis of seismic activity. Temporal trends were examined using Poisson point processes, which allowed for the identification of earthquake timing patterns. Moreover, adaptive kernel estimation techniques were employed to detect spatial clustering of seismic events, particularly in regions experiencing significant pore pressure depletion. Building upon these initial analyses, spatio-temporal Cox processes were applied to model earthquake intensity more comprehensively. This approach was enhanced by incorporating covariates derived from pore pressure estimates using polynomial kriging and Bayesian inference methods, thereby improving the predictive accuracy of seismic hazard assessments. An important advancement of this research lies in integrating NAM’s dynamic reservoir model with the spatio-temporal Cox process framework. This integration enabled the incorporation of physical processes, such as stress build-up and reservoir compaction, into the statistical models, thus bridging the gap between geomechanical insights and statistical predictions. To address the practical challenge of balancing gas production targets with seismic hazard mitigation, this dissertation introduces a decision-making framework based on Markov Decision Processes (MDPs). This framework allows for optimising production policies by accounting for seismic risk and resource extraction goals. By formulating and evaluating various production scenarios, the MDP framework provides a structured and data-driven approach to decision-making under uncertainty.
The findings of this research reveal a clear relationship between gas extraction, pore pressure depletion, and seismic activity, with significant clustering of earthquakes observed in regions of pronounced reservoir compaction. The enhanced predictive models, which integrate geomechanical and statistical techniques, offer an improvement in seismic hazard assessments. Furthermore, the decision-making framework demonstrates its utility in supporting safer and more sustainable production strategies. In conclusion, this dissertation advances the state of knowledge on induced seismicity in the Groningen gas field by presenting a robust, interdisciplinary framework for seismic monitoring, prediction, and mitigation. The integration of statistical models, geomechanical data, and decision-making tools provides a comprehensive approach to managing seismic risk. While the findings are directly relevant to the Groningen gas field, the methodologies developed herein have broader applicability to other regions and contexts where resource extraction induces seismic activity. Future research should focus on incorporating additional geological variables, real-time data assimilation for dynamic hazard assessment, and extending the decision-making framework to other subsurface activities
Missing value replacement in strings and applications
Get PDFMissing values arise routinely in real-world sequential (string) datasets due to: (1) imprecise data measurements; (2) flexible sequence modeling, such as binding profiles of molecular sequences; or (3) the existence of confidential information in a dataset which has been deleted deliberately for privacy protection. In order to analyze such datasets, it is often important to replace each missing value, with one or more valid letters, in an efficient and effective way. Here we formalize this task as a combinatorial optimization problem: the set of constraints includes the context of the missing value (i.e., its vicinity) as well as a finite set of user-defined forbidden patterns, modeling, for instance, implausible or confidential patterns; and the objective function seeks to minimize the number of new letters we introduce. Algorithmically, our problem translates to finding shortest paths in special graphs that contain forbidden edges representing the forbidden patterns. Our work makes the following contributions: (1) we design a linear-time algorithm to solve this problem for strings over constant-sized alphabets; (2) we show how our algorithm can be effortlessly applied to fully sanitize a private string in the presence of a set of fixed-length forbidden patterns [Bernardini et al. 2021a]; (3) we propose a methodology for sanitizing and clustering a collection of private strings that utilizes our algorithm and an effective and efficiently computable distance measure; and (4) we present extensive experimental results showing that our methodology can efficiently sanitize a collection of private strings while preserving clustering quality, outperforming the state of the art and baselines. To arrive at our theoretical results, we employ techniques from formal languages and combinatorial pattern matching
Strategies for adiabatic state preparation of quantum many-body systems
Get PDFQuantum computers represent a relatively new and promising development in computing technology. One of the applications of quantum computers is modelling quantum many-body systems, which describe interactions between particles on atomic and subatomic scales. Such systems are highly complex and challenging to model with classical computers due to the vast number of possible states and entanglement of the particles. In this dissertation, we examine the extent to which quantum computers can accelerate the modelling of quantum many-body systems compared to classical computers. Specifically, this dissertation presents research on adiabatic state preparation: a quantum algorithmic technique that uses the adiabatic principle from quantum mechanics to approximate eigenstates. We describe three new techniques that fall within this category and, in certain cases, offer advantages over standard methods.
First, we consider cases of ground state preparation for fermionic many-body systems, where standard direct interpolation between the initial and final hamiltonian is hindered by level crossings due to discrete symmetries. As an alternative to direct interpolation, we propose adiabatic paths in a higher-dimensional space, which break the relevant symmetries.
Next, we present an adiabatic echo verification protocol which mitigates both coherent and incoherent errors, arising from non-adiabatic transitions and hardware noise, respectively. We show that the estimator bias of the observable is reduced when compared to standard adiabatic preparation, achieving up to a quadratic improvement.
Finally, we propose a general, fully gate-based and nonvariational quantum algorithm for counterdiabatic driving. We provide a rigorous quantum gate complexity upper bound in terms of the minimum gap around this eigenstate. We find that, in the worst case, the algorithm can be run with at most quantum gates such that a target state fidelity of at least is achieved. In certain cases, the gap dependence can be improved to quadratic
An analysis of constraint-relaxation in PDE-based inverse problems
No full textMany inverse problems are naturally formulated as a PDE-constrained optimization problem. These non-linear, large-scale, constrained optimization problems know many challenges, of which the inherent non-linearity of the problem is an important one. In this paper, we focus on a relaxed formulation of the PDE-constrained optimization problem and provide analysis for its properties including convexity under certain assumptions. Starting from an infinite-dimensional formulation of the inverse problem with discrete data, we propose a general framework for the analysis and discretisation of such problems. The relaxed formulation of the PDE-constrained optimization problem is shown to reduce to a weighted non-linear least-squares problem. The weight matrix turns out to be the Gram matrix of solutions of the PDE, and in some cases be estimated directly from the measurements. The latter observation points to a potential way to unify recently proposed data-driven reduced-order models for inverse problems with PDE-constrained optimization. We provide a number of representative case studies and numerical examples to illustrate our findings
Modeling Alzheimer’s disease: Bayesian copula graphical model from demographic, cognitive, and neuroimaging data
No full textBackground. The early detection of Alzheimer’s disease (AD) requires an understanding of the relationships between a wide range of features. Conditional independencies and partial correlations are suitable measures for these relationships, because they can identify the effects of confounding and mediating variables.
Objective. To estimate conditional dependencies and partial correlations between relevant features in AD using a Bayesian approach to Gaussian copula graphical models (GCGMs). This approach has two key advantages. First, it includes binary, discrete, and continuous variables. Second, it quantifies the uncertainty of the estimates. Despite these advantages, Bayesian GCGMs have not been applied to AD research yet.
Methods. We design a GCGM to find the conditional dependencies and partial correlations among brain-region specific gray matter volume and glucose uptake, amyloid-beta levels, demographic information, and cognitive test scores. We applied our model to participants, including healthy and cognitively impaired, across different stages of AD.
Results. We found that aging reduces cognition through three indirect pathways: hippocampal volume loss, posterior cingulate cortex (PCC) volume loss, and amyloid-beta accumulation. We found a positive partial correlation between being woman and cognition, but also discovered four indirect pathways that dampen this association in women: lower hippocampal volume, lower PCC volume, more amyloid-beta accumulation, and less education. We found limited relations between brain-region specific glucose uptake and cognition, but discovered that the hippocampus and PCC volumes are related to cognition.
Conclusions. This study shows that the use of GCGMs offers valuable insights into AD pathogenesis
Uitgezocht: Hoe algoritmes bepalen wat jij ziet
No full textNiet meer kunnen stoppen met swipen. Veel kinderen hebben er last van. Vind je een filmpje niet leuk, dan ga je snel door naar de volgende. Zo kan je soms wel uren doorgaan. En dat is precies wat sociale media willen.
De 'For You Page' ziet er bij iedereen anders uit. Mensen krijgen dus niet precies dezelfde video's te zien. Alleen de filmpjes die ze zelf waarschijnlijk leuk vinden.
Hoe weten sociale media wat jij leuk vindt? Annabelle heeft het voor je Uitgezocht en deelt het geheim van haar tijdlijn
Towards contextual sensitive data detection
No full textThe emergence of open data portals necessitates more attention to protecting sensitive data before datasets get published and exchanged. While an abundance of methods for suppressing sensitive data exist, the conceptualization of sensitive data and methods to detect it, focus particularly on personal data that, if disclosed, may be harmful or violate privacy. We observe the need for refining and broadening our definitions of sensitive data, and argue that the sensitivity of data depends on its context. Based on this definition, we introduce two mechanisms for contextual sensitive data detection that consider the broader context of a dataset at hand. First, we introduce type contextualization, which first detects the semantic type of particular data values, then considers the overall context of the data values within the dataset or document. Second, we introduce domain contextualization which determines sensitivity of a given dataset in the broader context based on the retrieval of relevant rules from documents that specify data sensitivity (e.g., data topic and geographic origin). Experiments with these mechanisms, assisted by large language models (LLMs), confirm that: 1) type-contextualization significantly reduces the number of false positives for type-based sensitive data detection and reaches a recall of 94% compared to 63% with commercial tools, and 2) domain-contextualization leveraging sensitivity rule retrieval is effective for context-grounded sensitive data detection in non-standard data domains such as humanitarian datasets. Evaluation with humanitarian data experts also reveals that context-grounded LLM explanations provide useful guidance in manual data auditing processes, improving consistency. We open-source mechanisms and annotated datasets for contextual sensitive data detection at https://github.com/trl-lab/sensitive-data-detection
Waarom grijpen we zo vaak naar onze telefoon? Geef 'verslavende' algoritmen niet zomaar de schuld
No full textPredictive processing in neuroscience, computational modeling and psychology
No full textOver the past decades, predictive processing has emerged as a powerful theoretical framework that holds promise for explaining a wide range of phenomena, including perception and imagery but also sensorimotor control and consciousness. Here we focus on the question if and how predictive processing may be implemented in the mammalian and human brain, and what its scope of perceptual and cognitive functions is. We review basic and advanced computational models of predictive processing, expanding the range of computational, cognitive and sensorimotor capacities and enhancing their biological plausibility. Based on empirical evidence, major steps will need to be taken to flesh out how predictive processing may be precisely implemented in the brain, but the overall framework holds great potential as an explanatory framework in the neurosciences and psychology, strongly linking to Artificial Intelligence