256 research outputs found
PREPROCESSING SEASONAL TIME SERIES FOR IMPROVING NEURAL NETWORK PREDICTIONS
No full textThe presence of stochastic or deterministic trends in economic time series can be a major obstacle for producing satisfactory predictions with neural networks. In this paper, we demonstrate the effects of nonstationarity on neural network predictions using the time series of the mortgage loans purchased in the Netherlands. We present different preprocessing techniques for removing nonstationarity, and evaluate their properties by producing multi-step predictions using a linear stochastic forecasting model and a neural network. The results indicate that detecting nonstationarity and selecting an appropriate preprocessing technique is highly beneficial for improving the prediction quality
Nonstationarity and Data Preprocessing for Neural Network Predictions of an Economic Time Series
No full textThe presence of stochastic or deterministic trends in economic time series can be a major obstacle for producing satisfactory predictions with neural networks. In this paper, we demonstrate the effects of nonstationarity on neural network predictions using the time series of the mortgage loans purchased in the Netherlands. We present different preprocessing techniques for removing nonstationarity, and evaluate their properties by producing multi-step predictions using a linear stochastic forecasting model and a neural network. The results indicate that detecting nonstationarity and selecting an appropriate preprocessing technique is highly beneficial for improving the prediction quality
AN APPLICATION OF DATA MINING METHODS TO AIRLINE OVERBOOKING OPTIMIZATION
No full textThis paper deals with the use of advanced statistics and data mining techniques to
extract knowledge from large databases containing passenger and booking information
(mainly the so called Passenger Name Records, PNR) of a major German airline company.
Such knowledge is used to predict passenger behavior, which in turn is used to optimize capacity planning and improve overbooking management. The preliminary results, obtained
on a sample of selected flights, show that it is possible to successfully use PNR data and appropriate models to improve the overbooking optimization process. Critical success factors
are: (a) data collection and preparation; (b) the method used for exploratory analysis and
data reduction; (c) the forecasting methods: complex methods performed better, but simple
methods might be preferred due to lower computational requirements and overall cost
ElasticHash: Semantic Image Similarity Search in Elasticsearch
No full textModels and datasets for publication: Korfhage, N., Mühling, M., Freisleben, B. (2021). ElasticHash: Semantic Image Similarity Search by Deep Hashing with Elasticsearch. In: Tsapatsoulis, N., Panayides, A., Theocharides, T., Lanitis, A., Pattichis, C., Vento, M. (eds) Computer Analysis of Images and Patterns. CAIP 2021. Lecture Notes in Computer Science(), vol 13053. Springer, Cham. https://doi.org/10.1007/978-3-030-89131-2_2 ElasticHash uses Elasticsearch for high-quality, efficient, and large-scale semantic image similarity search. It is based on a deep hashing model to learn hash codes for fine-grained image similarity search in natural images and a two-stage method for efficiently searching binary hash codes using Elasticsearch (ES). In the first stage, a coarse search based on short hash codes is performed using multi-index hashing and ES terms lookup of neighboring hash codes. In the second stage, the list of results is re-ranked by computing the Hamming distance on long hash codes. The files published here are needed to set up the system for large-scale image similarity search and to reproduce the experiments. More details can be found in the Git-Repository: https://github.com/umr-ds/ElasticHas
Identifying and Counting Avian Blood Cells in Whole Slide Images via Deep Learning
No full textThis dataset contains research data as presented in: Vogelbacher, M.; Strehmann, F.; Bellafkir, H.; Mühling, M.; Korfhage, N.; Schneider, D.; Rösner, S.; Schabo, D. G.; Farwig, N.; Freisleben, B. Identifying and Counting Avian Blood Cells in Whole Slide Images via Deep Learning. Submitted for publication. 2024.
In this article, we present a novel approach to automatically quantify avian red and white blood cells in whole slide images. Our approach is based on two deep neural network models. The first model determines image regions that are suitable for counting blood cells, and the second model is an instance segmentation model that detects the cells in the determined image regions. The region selection model achieves up to 97.3% in terms of the F1 score, and the instance segmentation model achieves up to 90.7% in terms of mean average precision. Our approach helps ornithologists to acquire hematological data from avian blood smears more precisely and efficiently.
The data published here include the raw annotated data as well as the trained models for the automated counting of blood cells in avian blood smears. Our code is publicly available at https://github.com/umr-ds/avibloodcount.Hessian State Ministry for Higher Education, Research and the Arts (HMWK) (LOEWE Natur 4.0 and hessian.AI Connectom AI4Birds, AI4BirdsDemo
Security Analysis of System Behaviour - From 'Security by Design' to 'Security at Runtime'
No full textThe Internet today provides the environment for novel applications and processes which may evolve way beyond pre-planned scope and purpose. Security analysis is growing in complexity with the increase in functionality, connectivity, and dynamics of current electronic business processes. Technical processes within critical infrastructures also have to cope with these developments. To tackle the complexity of the security analysis, the application of models is becoming standard practice. However, model-based support for security analysis is not only needed in pre-operational phases but also during process execution, in order to provide situational security awareness at runtime. This cumulative thesis provides three major contributions to modelling methodology. Firstly, this thesis provides an approach for model-based Analysis and verification of security and safety properties in order to Support fault prevention and fault removal in system design or redesign. Further- more, some construction principles for the design of well-behaved scalable systems are given. The second topic is the analysis of the exposition of vulnerabilities in the software components of networked systems to exploitation by internal or external threats. This kind of fault forecasting allows the security assessment of alternative system configurations and security policies. Validation and deployment of security policies that minimise the attack surface can now improve fault tolerance and mitigate the impact of successful attacks. Thirdly, the approach is extended to runtime applicability. An observing system monitors an event stream from the observed system with the aim to detect faults deviations from the specified behaviour or security compliance violations at runtime. Furthermore, knowledge about the expected behaviour given by an operational model is used to predict faults in the near future. Building on this, a holistic security management strategy is proposed. The architecture of the observing system is described and the applicability of model-based security analysis at runtime is demonstrated utilising processes from several industrial scenarios. The results of this cumulative thesis are provided by 19 selected peer-reviewed papers
- …
