Browsing by Subject "Neural networks"

Now showing 1 - 3 of 3

Effects of using deep learning to predict the geographic origin of barley genebank accessions on genome–environment association studies
(2025) Chang, Che-Wei; Schmid, Karl
Genome–environment association (GEA) is an approach for identifying adaptive loci by combining genetic variation with environmental parameters, offering potential for improving crop resilience. However, its application to genebank accessions is limited by missing geographic origin data. To address this limitation, we explored the use of neural networks to predict the geographic origins of barley accessions and integrate imputed environmental data into GEA. Neural networks demonstrated high accuracy in cross-validation but occasionally produced ecologically implausible predictions as models solely considered geographical proximity. For example, some predicted origins were located within non-arable regions, such as the Mediterranean Sea. Using barley flowering time genes as benchmarks, GEA integrating imputed environmental data ( N=11,032) displayed partially concordant yet complementary detection of genomic regions near flowering time genes compared to regular GEA ( N=1,626), highlighting the potential of GEA with imputed data to complement regular GEA in uncovering novel adaptive loci. Also, contrary to our initial hypothesis anticipating a significant improvement in GEA performance by increasing sample size, our simulations yield unexpected insights. Our study suggests potential limitations in the sensitivity of GEA approaches to the considerable expansion in sample size achieved through predicting missing geographical data. Overall, our study provides insights into leveraging incomplete geographical origin data by integrating deep learning with GEA. Our findings indicate the need for further development of GEA approaches to optimize the use of imputed environmental data, such as incorporating regional GEA patterns instead of solely focusing on global associations between allele frequencies and environmental gradients across large-scale landscapes.
Entwicklung von datengetriebenen Auswerteverfahren zur Analyse und Schätzungder Reaktorleistung von Biogasanlagen
(2020) Beltramo, Tanja; Hitzmann, Bernd
The production of biogas is very complex process, which runs in some stages involving different microorganisms. Microbiological diversity of the process depends mainly on the composition of substrate and ambient conditions, such as process temperature. The fact is, the development and composition of the microbiological communities of the process are difficult to predict. Thus, the control and evaluation of such complex biological processes are very time consuming and expensive. In Germany the evaluation of the biogas plants can be performed according to the VDI-Norm 4630, which describes the methods for the evaluation of fermentation of organic materials including characterization of the substrate, sampling, collection of material data and fermentation tests. For that specially equipment and skilled personnel are required. Moreover, the evaluation procedure is very time consuming. That is why a new state-of-the-art alternative for the evaluation purposes is necessary to simplify and to speed up the assessment of the biogas production processes. The aim of this doctoral thesis is the development of a fast and reliable method for the evaluation of the biogas production processes. Therefore the mathematical modelling should identify significant process variables able to evaluate the whole process. For the optimization of mathematical models metaheuristic tools were used. In this doctoral thesis two different data sets were used – experimental data and simulated data. The experimental data were collected in projects “Biogas-Biocoenosis” (FKZ 22010711, Dr. Michael Klocke, Leibnitz-Institute für Agrartechnik und Bioökonomie e.V., Potsdam) and “Biogas-Enzyme” (FKZ 22027707, Dr. Monika Heiermann, Leibnitz-Institute für Agrartechnik und Bioökonomie e.V., Potsdam). The simulated data set was generated using the Anaerobic Digestion Model No.1 (ADM1). The chemical process variables were used as the independent process variable set, while the biogas production output represented the dependent process variable. Prediction of the biogas production was done using linear and nonlinear mathematic models. Here, Partial-Least-Square-Regression (PLSR), Locally-Weighted-Regression (LWR) and Artificial Neural Networks (ANN) were implemented. In order to identify the most significant undependable process variables optimization algorithms were used, Ant Colony Optimization (ACO) and Genetic Algorithm (GA). Prediction capacity was evaluated using two model evaluation variables, Root Mean Square Error (RMSE) and Coefficient of Determination (R2). Figure 1 in Supplementary represents the flow chart of the developed methodology applied for ADM1 generated data set. In Figure 2 (Supplementary) there is a flow chart of the developed methodology applied for the experimentally collected data. The developed approaches could be successfully used for the prediction of the desired process variable, biogas production rate. The variable selection done with the help of metaheuristic optimization algorithms improved the prediction results and reduced number of the independent process variables. Hydraulic retention time, dry matter, neutral detergent fibre, acid detergent fibre and n-butyric acid were identified as the most significant ones. The best prediction was obtained using ANN models. Here, the error of prediction was low and the coefficient of determination high. The successful implementation of the developed methodology proved mathematical models to be an effective alternative method capable to evaluate and to optimize complicated biological processes. Furthermore, it would be mandatory further experimental evaluation of the developed strategy, using the model-based process information.
Schätzung betrieblicher Kostenfunktionen mit künstlichen neuronalen Netzen
(2015) Simen, Jan-Philipp; Troßmann, Ernst
In this thesis a concept for estimating cost relationships with artificial neural networks is developed. The resulting open-source software application Cenobi (http://sourceforge.net/projects/cenobi/) is able to assess the impact of cost drivers on activity cost, plot non-linear cost functions, do forecasting and budgeting, calculate incremental cost, do unit costing, job costing etc., calculate cost driver rates and analyse cost variances. An object-oriented implementation of neural networks optimized by genetic algorithms provides the basis for these calculations.