Search

Motility of unicellular algal cells, especially its ability to respond to environmental cues, is crucial in industrial and ecological contexts and has been studied extensively with the model organism Chlamydomonas. However, little is known about the relationship between motility and the cell-cycle, despite the apparent link of mitosis dependent morphological changes which involve the flagella and the dependency on light/dark conditions of the cellular reactions. In this study, the cell swimming speed, the rotational diffusivity and its swimming bias against gravity were measured with high-speed video microscopy with Chlamydomonas reinhardtii cultures highly synchronised to a diurnal cycle. A simulation of gravitactic cell trajectories was developed to assist the statistical analysis of the cell trajectories from the images, which subject to a projection effect and has not been addressed previously. Its morphological changes, including cell size, shape, mass density, and presence of flagella were also evaluated. Un unforeseen change of cell motility at a critical mitosis time point was discovered, and our analysis suggests a connection to the alternating cell energetic regimes rather than the cell morphology. As indicated by results obtained from imaging based experimental measurements and by computational methods, the motility variation has direct consequences on the collective motion of algal cells in a Hagen-Poiseuille flow, a relevant component flow for air-lift photobioreactors. The cell number density profiles were calculated by an individual-based simulation and a continuum system inclusive of the buoyancy effect imposed by the aggregated cells on their surrounding fluid. Qualitative experimental-theoretical agreement suggests that the models can be employed for optimisation problems of photobioreactor flow designs inclusive of the non-negligible cell-cycle effects, which has been commonly overlooked in previous studies.

Graphs are ubiquitous as they naturally capture interactions between entities, such as user interaction in online social media, paper citations in bibliographic networks, and user-product preferences in sales networks. Recently, graph representation learning has gained significant popularity in both academia and industry thanks to its state-of-the-art performance in a variety of downstream machine learning (ML) tasks, such as friend recommendations and anomaly detection. Specifically, node representation learning (embedding) aims to find a dense vector of rich latent features per entity that can be used in ML tasks. However, these dense representations with fixed dimensions come with computational and storage challenges for real-world graphs with many millions or billions of nodes, and the "black-box" nature of the latent features impedes interpretability. On the other hand, graph summarization aims to find a concise and interpretable representation of the original graph that describes its key information, but it is often lossy and trades off space and performance in ML tasks. In this thesis, we bridge the two lines of research, node embedding and graph summarization: we introduce scalable methods for generating summaries of latent or non-latent (original) node features that achieve the state-of-the-art performance on ML tasks while requiring significantly reduced storage and supporting interpretability. Specifically, we introduce a new problem, latent network summarization, which summarizes the graph structural features in static networks as latent node embeddings for storage and query efficiency, and extend this idea to incorporate temporal proximity in temporal summaries of continuous-time dynamic networks. We also perform an extensive systematic study of temporal summaries and show that they capture the graph structure and temporal dependency at least as well as recently-proposed dynamic embedding approaches, while having significantly less complexity (i.e., no transitional or latent variables). Unlike methods that are based on complex models as "black boxes", our temporal summaries are easy-to-understand, which motivates their usage for practitioners in predictive applications. Finally, we summarize the non-latent graph features by modeling feature importance as the high-level knowledge through traditional and deep learning models that can be used for graph analysis and transfer learning. Throughout the thesis, we demonstrate the effectiveness, scalability and space efficiency of our methods on industrial applications such as entity linkage, user stitching, professional role inference, and temporal link prediction, and present insights that can inform further methodological development and applications.

This dissertation presents three new quantum key distribution protocols over a dual quantum-channel to enhance the convergence speed of the key distribution process. The proposed protocols effectively use a dual-quantum channel between the communicating entities and, when combined with the proposed selection of the basis, improve the efficiency of the key distribution process as well as deter Eve. With the use of an initialization vector, one of the proposed protocols does not necessitate information exchange about the bases between Alice and Bob, and further eliminates the requirement of the availability of a two-way conventional channel. Simultaneously, it allows the detection of the presence of Eve in real time and proposes several effective defense mechanisms to frustrate Eve's attack.