Spatiotemporal Predictive Learning (SPL) is an essential research topic involving many practical and real-world applications, e.g., motion detection, video generation, precipitation forecasting, and traffic flow prediction. The problems and challenges of this field come from numerous data characteristics in both time and space domains, and they vary depending on the specific task. For instance, spatial analysis refers to the study of spatial features, such as spatial location, latitude, elevation, longitude, the shape of objects, and other patterns. From the time domain perspective, the temporal analysis generally illustrates the time steps and time intervals of data points in the sequence, also known as interval recording or time sampling. Typically, there are two types of time sampling in temporal analysis: regular time sampling (i.e., the time interval is assumed to be fixed) and the irregular time sampling (i.e., the time interval is considered arbitrary) related closely to the continuous-time prediction task when data are in continuous space. Therefore, an efficient spatiotemporal predictive method has to model spatial features properly at the given time sampling types. In this thesis, by taking advantage of Machine Learning (ML) and Deep Learning (DL) methods, which have achieved promising performance in many complicated computational tasks, we propose three DL-based models used for Spatiotemporal Sequence Prediction (SSP) with several types of time sampling. First, we design the Trajectory Gated Recurrent Unit Attention (TrajGRU-Attention) with novel attention mechanisms, namely Motion-based Attention (MA), to improve the performance of the standard Convolutional Recurrent Neural Networks (ConvRNNs) in the SSP tasks. In particular, the TrajGRU-Attention model can alleviate the impact of the vanishing gradient, which leads to the blurry effect in the long-term predictions and handle both regularly sampled and irregularly sampled time series. Consequently, this model can work effectively with different scenarios of spatiotemporal sequential data, especially in the case of time series with missing time steps. Second, by taking the idea of Neural Ordinary Differential Equations (NODEs), we propose Trajectory Gated Recurrent Unit integrating Ordinary Differential Equation techniques (TrajGRU-ODE) as a continuous time-series model. With Ordinary Differential Equation (ODE) techniques and the TrajGRU neural network, this model can perform continuous-time spatiotemporal prediction tasks and generate resulting output with high accuracy. Compared to TrajGRU-Attention, TrajGRU-ODE benefits from the development of efficient and accurate ODE solvers. Ultimately, we attempt to combine those two models to create TrajGRU-Attention-ODE. NODEs are still in their early stage of research, and recent ODE-based models were designed for many relatively simple tasks. In this thesis, we will train the models with several video datasets to verify the ability of the proposed models in practical applications. To evaluate the performance of the proposed models, we select four available spatiotemporal datasets based on the complexity level, including the MovingMNIST, MovingMNIST++, and two real-life datasets: the weather radar HKO-7 and KTH Action. With each dataset, we train, validate, and test with distinct types of time sampling to justify the prediction ability of our models. In summary, the experimental results on the four datasets indicate the proposed models can generate predictions properly with high accuracy and sharpness. Significantly, the proposed models outperform state-of-the-art ODE-based approaches under SSP tasks with different circumstances of interval recording.

Komplexe Systeme wie das Klima der Erde bestehen aus vielen Komponenten, die durch eine komplizierte Kopplungsstruktur miteinander verbunden sind. Für die Analyse solcher Systeme erscheint es daher naheliegend, Methoden aus der Netzwerktheorie, der Theorie dynamischer Systeme und dem maschinellen Lernen zusammenzubringen. Durch die Kombination verschiedener Konzepte aus diesen Bereichen werden in dieser Arbeit drei neuartige Ansätze zur Untersuchung komplexer Systeme betrachtet. Im ersten Teil wird eine Methode zur Konstruktion komplexer Netzwerke vorgestellt, die in der Lage ist, Windpfade des südamerikanischen Monsunsystems zu identifizieren. Diese Analyse weist u.a. auf den Einfluss der Rossby-Wellenzüge auf das Monsunsystem hin. Dies wird weiter untersucht, indem gezeigt wird, dass der Niederschlag mit den Rossby-Wellen phasenkohärent ist. So zeigt der erste Teil dieser Arbeit, wie komplexe Netzwerke verwendet werden können, um räumlich-zeitliche Variabilitätsmuster zu identifizieren, die dann mit Methoden der nichtlinearen Dynamik weiter analysiert werden können. Die meisten komplexen Systeme weisen eine große Anzahl von möglichen asymptotischen Zuständen auf. Um solche Zustände zu beschreiben, wird im zweiten Teil die Monte Carlo Basin Bifurcation Analyse (MCBB), eine neuartige numerische Methode, vorgestellt. Angesiedelt zwischen der klassischen Analyse mit Ordnungsparametern und einer gründlicheren, detaillierteren Bifurkationsanalyse, kombiniert MCBB Zufallsstichproben mit Clustering, um die verschiedenen Zustände und ihre Einzugsgebiete zu identifizieren. Bei von Vorhersagen von komplexen Systemen ist es nicht immer einfach, wie Vorwissen in datengetriebenen Methoden integriert werden kann. Eine Möglichkeit hierzu ist die Verwendung von Neuronalen Partiellen Differentialgleichungen. Hier wird im letzten Teil der Arbeit gezeigt, wie hochdimensionale räumlich-zeitlich chaotische Systeme mit einem solchen Ansatz modelliert und vorhergesagt werden können. / Complex systems such as the Earth's climate are comprised of many constituents that are interlinked through an intricate coupling structure. For the analysis of such systems it therefore seems natural to bring together methods from network theory, dynamical systems theory and machine learning. By combining different concepts from these fields three novel approaches for the study of complex systems are considered throughout this thesis. In the first part, a novel complex network construction method is introduced that is able to identify the most important wind paths of the South American Monsoon system. Aside from the importance of cross-equatorial flows, this analysis points to the impact Rossby Wave trains have both on the precipitation and low-level circulation. This connection is then further explored by showing that the precipitation is phase coherent to the Rossby Wave. As such, the first part of this thesis demonstrates how complex networks can be used to identify spatiotemporal variability patterns within large amounts of data, that are then further analysed with methods from nonlinear dynamics. Most complex systems exhibit a large number of possible asymptotic states. To investigate and track such states, Monte Carlo Basin Bifurcation analysis (MCBB), a novel numerical method is introduced in the second part. Situated between the classical analysis with macroscopic order parameters and a more thorough, detailed bifurcation analysis, MCBB combines random sampling with clustering methods to identify and characterise the different asymptotic states and their basins of attraction. Forecasts of complex system are the next logical step. When doing so, it is not always straightforward how prior knowledge in data-driven methods. One possibility to do is by using Neural Partial Differential Equations. Here, it is demonstrated how high-dimensional spatiotemporally chaotic systems can be modelled and predicted with such an approach in the last part of the thesis.

Today, a large amount of time series data is being produced from a variety of different devices such as smart speakers, cell phones and vehicles. This data can be used to make inferences and predictions. Neural network based methods are among one of the most popular ways to model time series data. The field of neural networks is constantly expanding and new methods and model variants are frequently introduced. In 2018, a new family of neural networks was introduced. Namely, Neural Ordinary Differential Equations (Neural ODEs). Neural ODEs have shown great potential in modelling the dynamics of temporal data. Here we present an investigation into using Neural Ordinary Differential Equations for anomaly detection. We tested two model variants, LSTM-ODE and latent-ODE. The former model utilises a neural ODE to model the continuous-time hidden state in between observations of an LSTM model, the latter is a variational autoencoder that uses the LSTM-ODE as encoding and a Neural ODE as decoding. Both models are suited for modelling sparsely and irregularly sampled time series data. Here, we test their ability to detect anomalies on various sparsity and irregularity ofthe data. The models are compared to a Gaussian mixture model, a vanilla LSTM model and an LSTM variational autoencoder. Experimental results using the Human Activity Recognition dataset showed that the Neural ODEbased models obtained a better ability to detect anomalies compared to their LSTM based counterparts. However, the computational training cost of the Neural ODE models were considerably higher than for the models that onlyutilise the LSTM architecture. The Neural ODE based methods were also more memory consuming than their LSTM counterparts. / Idag produceras en stor mängd tidsseriedata från en mängd olika enheter som smarta högtalare, mobiltelefoner och fordon. Denna datan kan användas för att dra slutsatser och förutsägelser. Neurala nätverksbaserade metoder är bland de mest populära sätten att modellera tidsseriedata. Mycket forskning inom området neurala nätverk pågår och nya metoder och modellvarianter introduceras ofta. Under 2018 introducerades en ny familj av neurala nätverk. Nämligen, Neurala Ordinära Differentialekvationer (NeuralaODE:er). Neurala ODE:er har visat en stor potential i att modellera dynamiken hos temporal data. Vi presenterar här en undersökning i att använda neuralaordinära differentialekvationer för anomalidetektion. Vi testade två olika modellvarianter, en som kallas LSTM-ODE och en annan som kallas latent-ODE.Den förstnämnda använder Neurala ODE:er för att modellera det kontinuerliga dolda tillståndet mellan observationer av en LSTM-modell, den andra är en variational autoencoder som använder LSTM-ODE som kodning och en Neural ODE som avkodning. Båda dessa modeller är lämpliga för att modellera glest och oregelbundet samplade tidsserier. Därför testas deras förmåga att upptäcka anomalier på olika gleshet och oregelbundenhet av datan. Modellerna jämförs med en gaussisk blandningsmodell, en vanlig LSTM modell och en LSTM variational autoencoder. Experimentella resultat vid användning av datasetet Human Activity Recognition (HAR) visade att de Neurala ODE-baserade modellerna erhöll en bättre förmåga att upptäcka avvikelser jämfört med deras LSTM-baserade motsvarighet. Träningstiden förde Neurala ODE-baserade modellerna var dock betydligt långsammare än träningstiden för deras LSTM-baserade motsvarighet. Neurala ODE-baserade metoder krävde också mer minnesanvändning än deras LSTM motsvarighet.