检索结果-内蒙古大学图书馆

High-Frequency Causality in the VIX Index and its derivatives: Empirical Evidence

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Farokhnia, Kia Osterrieder, Joerg Institute of Data Analysis and Process Design School of Engineering Zurich University of Applied Sciences Switzerland

In February 2018, the VIX index has seen its largest ever increase and has lead to significant losses for some major volatility-related products. Despite many efforts, the precise underlying reasons are yet to be discovered. We study the role of linear causality in the VIX index and its derivatives during January and February 2018. Due to the shortcomings of statistical inferences for stochastic volatility models, the dynamics of the volatility expectation index VIX remain controversial. Leveraging intraday data, we discover novel empirical results describing their interaction. We find bidirectional causality between the VIX spot and the implied volatility of Standard & Poor's 500 options, suggesting a volatility feedback effect. The spot index tends to be lagging its own futures, while the vector autoregression's error correction mechanism reveals a significant mean-reverting equilibrium relationship. The evidence is consistent with recent theories indicating that implied volatility has stronger feedback than realized volatility. The paper reveals a retroactive information flow and highlights novel insights for the market microstructure of VIX derivatives and their related S&P 500 options. © 2022, CC0.

关键词： Error correction

Approximative Modelling of Manufacturing Systems

学校读者我要写书评

暂无评论

IFAC Proceedings Volumes 2003年第3期36卷 143-148页

作者： Christoph Heitz Institute for Data Analysis and Process Design Zurich University of Applied Sciences Winterthur IPLnet: Swiss National Competence Network for Integrated Production and Logistics

A new model for an approximative description of a production line is presented. Instead of a bottom-up modelling, an approach is presented which allows the modelling of large parts of the production line as a single "module" which is characterized by a characteristic function describing the dynamical behaviour under stationary conditions. Given the characteristic functions of the different modules, the production rate of the whole system can be calculated without explicit simulation of the whole line. Since the model is based on a macro-level description, it is close to a “management view” of the system.

关键词： Stochastic modelling Methodology Industrial production systems Automobile industry

Artificial neural networks to impute rounded zeros in compositional data

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Templ, Matthias Institute of Data Analysis and Process Design Zurich University of Applied Sciences Rosenstrasse 3 WinterthurCH-8401 Switzerland

Methods of deep learning have become increasingly popular in recent years, but they have not arrived in compositional data analysis. Imputation methods for compositional data are typically applied on additive, centered or isometric log-ratio representations of the data. Generally, methods for compositional data analysis can only be applied to observed positive entries in a data matrix. Therefore one tries to impute missing values or measurements that were below a detection limit. In this paper, a new method for imputing rounded zeros based on artificial neural networks is shown and compared with conventional methods. We are also interested in the question whether for ANNs, a representation of the data in log-ratios for imputation purposes, is relevant. It can be shown, that ANNs are competitive or even performing better when imputing rounded zeros of data sets with moderate size. They deliver better results when data sets are big. Also, we can see that log-ratio transformations within the artificial neural network imputation procedure nevertheless help to improve the results. This proves that the theory of compositional data analysis and the fulfillment of all properties of compositional data analysis is still very important in the age of deep learning. © 2020, CC BY-NC-ND.

关键词： Neural networks

Online sequential extreme learning machines for fault detection

学校读者我要写书评

暂无评论

Online sequential extreme learning machines for fault detect...

International Conference on Prognostics and Health Management, PHM

作者： Yang Hu Olga Fink Thomas Palmé Institute for Data Analysis and Process Design (IDP) Zurich University of Applied Sciences Winterthur Switzerland General Electric (GE) Baden Switzerland

ISBN: (纸本)9781509003839

In this paper, we propose the application of a new fault detection approach with a sequential updating function under new operating conditions or natural evolving degradation processes. The proposed approach is based on Online Sequential Extreme Learning Machines (OS-ELM). OS-ELM have the advantages of a strong learning ability, very fast training, and online learning. The concept of applying OS-ELM to fault detection is demonstrated on an artificial case study. We expect that OS-ELM can contribute to improve the fault detection and also the associated initiation of maintenance activities for engineering components working in an evolving environment such as electric components, bearings, gears, alternators, shafts and pumps, in which the monitored signals are not only significantly affected by working load and surrounding environment but may also experience some modifications due to a slow aging and degradation process.

关键词： Training Fault detection Signal reconstruction Mathematical model Degradation Machine learning algorithms data models

Gradual (In)Compatibility of Fairness Criteria

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Hertweck, Corinna Räz, Tim Institute for Data Analysis and Process Design Zurich University of Applied Sciences Winterthur Switzerland Department of Informatics University of Zurich Zurich Switzerland Institute of Philosophy University of Bern Bern Switzerland

Impossibility results show that important fairness measures (independence, separation, sufficiency) cannot be satisfied at the same time under reasonable assumptions. This paper explores whether we can satisfy and/or improve these fairness measures simultaneously to a certain degree. We introduce information-theoretic formulations of the fairness measures and define degrees of fairness based on these formulations. The information-theoretic formulations suggest unexplored theoretical relations between the three fairness measures. In the experimental part, we use the information-theoretic expressions as regularizers to obtain fairness-regularized predictors for three standard datasets. Our experiments show that a) fairness regularization directly increases fairness measures, in line with existing work, and b) some fairness regularizations indirectly increase other fairness measures, as suggested by our theoretical findings. This establishes that it is possible to increase the degree to which some fairness measures are satisfied at the same time-some fairness measures are gradually compatible. Copyright © 2021, The Authors. All rights reserved.

关键词： Information theory

Greening urban logistics: Introducing a method to evaluate optimised decoupling hub locations

学校读者我要写书评

暂无评论

Transportation Research Procedia 2023年 72卷 572-579页

作者： Maike Scherrer Albert Steiner Viola Rühlin Zurich University of Applied Sciences School of Engineering Institute of Sustainable Development Technikumstrasse 9 8401 Winterthur Switzerland Zurich University of Applied Sciences School of Engineering Institute of Data Analysis and Process Design Rosenstrasse 3 8401 Winterthur Switzerland

Sustainable urban logistics has become challenging over the course of the last decades due to logistics sprawl and increased demands of goods in urban areas. To overcome this problem, a fine-grained network of decoupling hubs, separating suppliers and customers in a spacial and timely manner, are needed. The paper at hand introduces a model of how to calculate optimal decoupling hub locations within urban areas with the goal that inhabitants walk or cycle to the next decoupling hub and do not take a motorised vehicle. The study at hand uses the northern district of Zurich as unit of analysis and shows that by using the developed model, 80% of the inhabitants need to walk less than 250 metres to their next decoupling hub, which is within the radius of 100 metres to the next public transport stop. This supports the integration of picking up or dropping off deliveries while commuting.

关键词： decoupling hubs sustainable urban logistics facility location probem

Quality assessment of Affymetrix GeneChip data using the EM algorithm and a naïve Bayes classifier

学校读者我要写书评

暂无评论

Quality assessment of Affymetrix GeneChip data using the EM ...

7th IEEE International Conference on Bioinformatics and Bioengineering, BIBE

作者： Howard, Brian E. Sick, Beate Perera, Imara Yang, Ju Im Winter-Sederoff, Heike Heber, Steffen Bioinformatics Research Center North Carolina State University Raleigh NC United States Institute of Data Analysis and Process Design Zurich University of Applied Science Winterthur Winterthur Switzerland Department of Plant Biology North Carolina State University Raleigh NC United States Department of Computer Science North Carolina State University Raleigh NC United States

ISBN: (纸本)1424415098

Recent research has demonstrated the utility of using supervised classification systems for automatic identification of low quality microarray data. However, this approach requires annotation of a large training set by a qualified expert. In this paper we demonstrate the utility of an unsupervised classification technique based on the Expectation-Maximization (EM) algorithm and naïve Bayes classification. On our test set, this system exhibits performance comparable to that of an analogous supervised learner constructed from the same training data. ©2007 IEEE.

关键词： Maximum principle

Deep Interpretable Ensembles

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Kook, Lucas Götschi, Andrea Baumann, Philipp F. M. Hothorn, Torsten Sick, Beate Epidemiology Biostatistics & Prevention Institute University of Zurich CH-8001 Switzerland Institute for Data Analysis and Process Design Zurich University of Applied Sciences CH-8400 Switzerland KOF Swiss Economic Institute ETH Zurich CH-8092 Switzerland

Ensembles improve prediction performance and allow uncertainty quantification by aggregating predictions from multiple models. In deep ensembling, the individual models are usually black box neural networks, or recently, partially interpretable semi-structured deep transformation models. However, interpretability of the ensemble members is generally lost upon aggregation. This is a crucial drawback of deep ensembles in high-stake decision fields, in which interpretable models are desired. We propose a novel transformation ensemble which aggregates probabilistic predictions with the guarantee to preserve interpretability and yield uniformly better predictions than the ensemble members on average. Transformation ensembles are tailored towards interpretable deep transformation models but are applicable to a wider range of probabilistic neural networks. In experiments on several publicly available data sets, we demonstrate that transformation ensembles perform on par with classical deep ensembles in terms of prediction performance, discrimination, and calibration. In addition, we demonstrate how transformation ensembles quantify both aleatoric and epistemic uncertainty, and produce minimax optimal predictions under certain conditions. © 2022, CC BY.

关键词： Uncertainty analysis

Deep transformation models for functional outcome prediction after acute ischemic stroke

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Herzog, Lisa Kook, Lucas Götschi, Andrea Petermann, Katrin Hänsel, Martin Hamann, Janne Dürr, Oliver Wegener, Susanne Sick, Beate Epidemiology Biostatistics & Prevention Institute University of Zurich Switzerland Institute for Data Analysis and Process Design Zurich University of Applied Sciences Switzerland Department of Neurology University Hospital Zurich Switzerland Insitute for Optical Systems Konstanz University of Applied Sciences Germany

In many medical applications, interpretable models with high prediction performance are sought. Often, those models are required to handle semi-structured data like tabular and image data. We show how to apply deep transformation models (DTMs) for distributional regression which fulfill these requirements. DTMs allow the data analyst to specify (deep) neural networks for different input modalities making them applicable to various research questions. Like statistical models, DTMs can provide interpretable effect estimates while achieving the state-of-the-art prediction performance of deep neural networks. In addition, the construction of ensembles of DTMs that retain model structure and interpretability allows quantifying epistemic and aleatoric uncertainty. In this study, we compare several DTMs, including baseline-adjusted models, trained on a semi-structured data set of 407 stroke patients with the aim to predict ordinal functional outcome three months after stroke. We follow statistical principles of model-building to achieve an adequate trade-off between interpretability and flexibility while assessing the relative importance of the involved data modalities. We evaluate the models for an ordinal and dichotomized version of the outcome as used in clinical practice. We show that both, tabular clinical and brain imaging data, are useful for functional outcome prediction, while models based on tabular data only outperform those based on imaging data only. There is no substantial evidence for improved prediction when combining both data modalities. Overall, we highlight that DTMs provide a powerful, interpretable approach to analyzing semi-structured data and that they have the potential to support clinical decision making. © 2022, CC BY.

关键词： Deep neural networks