Technical University of Munich, 2019. — 184 p. The thesis presents a state-of-the-art approach for numerical modeling of reservoir sedimentation under climate change scenarios. The approach uses cascade modeling along with an automatic calibration of models using accurate sediment load boundary conditions that were reconstructed with wavelet artificial neural networks.
Ruhr University Bochum, 2018. — 231 p. With the increasing availability of digitized resources of historical documents, interest in effective natural language processing (NLP) for these documents is on the rise. However, the abundance of variant spellings makes them challenging to work with both for human users and for NLP tools. Normalization to contemporary spelling is often...
Ludwig Maximilian University of Munich, 2017. — 132 p. Sentiment Analysis (SA) is the study of opinions and emotions that are conveyed by text. This field of study has commercial applications for example in market research (e.g., “What do customers like and dislike about a product?”) and consumer behavior (e.g., “Which book will a customer buy next when he wrote a positive...
University of Wuppertal, 2018. — 153 p. In this thesis we developed waveform-relaxation methods suitable for application in spiking neural network simulators. The main achievements of this thesis are the identification of suitable methods, their efficient implementation in the existing structures of the parallel simulator NEST and their numerical and theoretical analysis.
Dissertation. — University of Manchester, 2001. — 166 p. In this study we attempt to predict the daily excess returns of FTSE 500 and S&P 500 indices over the respective Treasury Bill rate returns. Initially, we prove that the excess returns time series do not fluctuate randomly. Furthermore we apply two different types of prediction models: Autoregressive (AR) and feed forward...
University of Hamburg, 2017. — 164 p. Perceiving the actions of other people is one of the most important social skills of human beings. We are able to reliably discern a variety of socially relevant information from people’s body motion such as intentions, identity, gender, and affective states. This ability is supported by highly developed visual skills and the integration of...
Free University of Berlin, 2015. — 163 p. Everyday social interactions require a constant integration of external socio- emotional cues in order to adequately react and adapt to conspecifics, which is fundamental for engaging in fruitful collaborations and joyful interactions. The present dissertation thesis aims to elucidate the underlying neural integration mechanism by...
Dissertation, University of Toronto, 2009, -84 p. Building intelligent systems that are capable of extracting high-level representations from high-dimensional sensory data lies at the core of solving many AI related tasks, including object recognition, speech perception, and language understanding. Theoretical and biological arguments strongly suggest that building such systems...
Diploma (Master), Manchester Metropolitan University, 1996. - 123 p. In this project a new modular neural network is proposed. The basic building blocks of the architecture are small multilayer feedforward networks, trained using the Backpropagation algorithm. The structure of the modular system is similar to architectures known from logical neural networks. The new network is...
Technical University of Berlin, 2018. — 110 p. The goal of this thesis is to develop deep neural networks that are capableof learning representations for atomistic systems. Beyond that we aim to pro-vide techniques to extract insights about the obtained representation as well asthe underlying data. We will reuse the deep learning architecture in a varietyof applications, thus,...
Dissertation, University of Toronto, 2013, -101 p. Recurrent Neural Networks (RNNs) are powerful sequence models that were believed to be difficult to train, and as a result they were rarely used in machine learning applications. This thesis presents methods that overcome the difficulty of training RNNs, and applications of RNNs to challenging problems. We first describe a new...
Technical University of Braunschweig, 2015. — 268 p. Accurate predictions of storm-tide are of vital importance for many coastal areas. Along North Sea coasts, reliable storm-tide predictions are of crucial importance as a large portion of the coastal zones is not only below mean sea level but also characterized by frequent storms. Currently, the nature of mutual nonlinear...
Technical University of Munich, 2018. — 210 p. For neural networks we propose stochastic, non-parametric activation functions that are fully learnable and individual to each neuron. Overfitting is prevented by placing a Gaussian process prior over these functions. The model can handle uncertainties in its inputs and self-estimate the confidence of its predictions. Using...
Technical University of Munich, 2016. — 315 p. CRESST is a direct Dark Matter search experiment aiming at the detection of WIMP-like Dark Matter scattering off nuclei in scintillating CaWO4 crystals operated as cryogenic detectors. A method based on artificial neural networks to reject events containing a pulse with distorted shape is developed and applied. Data from 5 modules...
Comments