Find relevant scholarly articles now

Shared mobility systems have been recently tested and piloted in many cities across the globe, with great potentials to be deeply woven into the fabric of future urban planning. Those systems, not only represent a more sustainable paradigm that can effectively cut unnecessary emissions, but could also bring significant societal benefits by offering a much more affordable on-demand mobility option to the general public. As a relatively new mobility trend, they expand at impressive speeds, pouring more fleet and infrastructure into the urban spaces than ever before. Coupled with the unpredictability of real-world environments, this brings numerous challenges in the deployment and operation of those systems, impacting their usability and practicality. In this thesis, we aim to better understand the structure, process and interaction of the shared mobility system with such dynamics, from a spatio-temporal learning perspective. Particularly, this thesis addresses the following question: How deep spatio-temporal learning can be tailored to improve the prediction, optimisation and actuation of shared mobility systems in the presence of real-world dynamics? The key insight is that such dynamics, although complex, shouldn’t be treated independently in an isolated way, but rather be considered and embraced at full stack, from data modelling to inference and learning across the spatio-temporal domain. Specifically, we tackle this in a number of research threads. Firstly, we propose D3P, a novel data-driven prediction framework, which is able to directly model the spatio-temporal dynamics from the data with time-varying graphs, and uses bespoke dynamic Graph Convolutional Neural Networks (GCNs) to accurately forecast the future demand of the shared mobility systems in both short and long terms. We further proposeMANS, a new optimisation approach for mobility infrastructure deployment based on multi-agent neural search, which can effectively discover the optimal infrastructure deployment strategies for shared mobility systems across space and time, so that the services provided are ubiquitous to the users while sustainable in profitability. Finally, we propose ac-PPO, a novel user-incentive based fleet management approach, which uses a deep reinforcement learning paradigm to guide the rebalancing of fleet with cooperative users, and incorporates the dynamics by a new action cascading technique. All the proposed approaches have been comprehensively evaluated through large-scale experiments with extensive real-world datasets, and the results have shown superior performance compared with the state-of-the-art, demonstrating their potential impact on a broad spectrum of application scenarios.

While the city has been central to the Indian novel in English since the 1980s, the profusion of urban novels, essays and literary reportages published since the 2000s has triggered a formal and thematic renewal of the literary discourse on the Indian city. At the crossroads of literature and urban studies, this thesis locates this literary phenomenon in the context of India’s embrace of global capitalism in the 1990s, which has resulted in the accelerated expansion and transformation of Indian cities, inspired by the model of the global city. Based on a corpus of fictional and non-fictional texts on Delhi, Mumbai and Kolkata, I study the development of a critical urban imaginary which expresses the contradictory experience of this urban metamorphosis through the interplay between two major aesthetic modes, and challenges both orientalist and nationalist discourses on the Indian city. These texts oscillate between an epic mode, which defamiliarize urban modernisation and amplifies the collision between antagonistic global social forces in the city, and an ordinary mode, which explores this historical process at the scale of the locality through the lens of everyday life, obliquely shedding light on structural violence but also on tactics devised by urban outcasts to reclaim urban space. These two modes are considered as the two faces of a critical literary approach of the city, which rests on the strong historical consciousness of the writers. Their works unveil the multiple layers of a fragmented urban history which contemporary urban planning endeavours to erase.

The actions of many interacting entities within socio-economic systems proclaim the configurations such as the spatial structure in urban environments. Hence understanding these underlying interactions are important in making the location decisions for growth in urban systems. In this thesis, a Bayesian spatial interaction model, henceforth BSIM, is developed to provide probabilistic predictions about revenues generated by a particular business location, based on its features and the potential customers’ characteristics in a given region. BSIM explicitly accounts for the competition among the facilities through a probability determined by evaluating a store-specific Gaussian distribution at a particular customer location. I propose a scalable variational inference framework that exhibits comparable performance in terms of parameter identification and uncertainty quantification while being significantly faster than competing Markov Chain Monte Carlo inference schemes. The advantages of the introduced BSIM are explored in addressing the competitive facility location problem that typically arises when businesses plan to enter a new market or expand their presence in an environment with existing competitors. A mathematical modelling framework is formulated to simultaneously identify the location and design of new stores in order to maximise the revenue predicted from BSIM in a geographical region. Solving the underlying optimisation problem requires the provision of an exhaustive set of potential sites, which is difficult in practice. Instead, a search algorithm is proposed based on the quadtree method to overcome this challenge by hierarchically exploring geographic regions of varying spatial resolution. This thesis introduces multiple large-scale real-world datasets compiled with open and proprietary data. Finally, demonstrate the proposed framework by producing optimal facility locations and corresponding designs for two case studies in the supermarket and pub sectors in Greater London, providing valuable insights for planning and decision-making under uncertainty.

Surveillance of diarrhoea and enteric infection among infants is key to improving infant health in LMICs. However, the performance of current surveillance methods in estimating rates of diarrhoea and enteric infection have been called into question by various academics. Diarrhoea rates are often estimated through subjective measures such as community surveys. These measures may introduce variation into the estimated rates. Rates of enteric infection are often estimated using diarrhoea as a proxy marker. However, it is possible that diarrhoea does not classify enteric infection well. In this thesis, I examine variation in estimates of diarrhoea rates associated with the method used; and the extent of misclassification when using diarrhoea as a proxy marker of enteric infection. To examine variation in estimates of diarrhoea rates associated with the method used, I conducted two studies: 1) a systematic review and meta-analysis of the literature; and 2) a randomised controlled trial (RCT) in a Tanzanian urban informal settlement. My findings in both confirmed that the methods used in estimating diarrhoea rates are associated with variations in estimates. The RCT in Tanzania also showed that text message surveys are a suitable substitute for in-person surveys. To examine the reliability of diarrhoea to classify enteric infection, I conducted a ‘diagnostic’ study in a Bangladeshi Refugee Camp. The evidence showed that diarrhoea misclassified enteric infection, offering no better a classification of enteric infection than flipping a coin. I also found evidence of clear seasonality in infection. I have demonstrated that the methods used in diarrhoea surveillance can influence estimated rates, and that diarrhoea widely misclassifies enteric infection when used as a proxy marker. For this reason, I hazard against the use of the current methods in epidemiological surveys when trying to estimate rates of diarrhoea and classify cases of enteric infection. Alternative, less subjective, methods, such as stool sampling for enteric infection, should be used.