Wearable devices including accelerometers are increasingly being used to collect high-frequency human activity data in situ. There is tremendous potential to use such data to inform medical decision making and public health policies. However, modeling such data is challenging as they are high-dimensional, heterogeneous, and subject to informative missingness, e.g., zero readings when the device is removed by the participant. We propose a flexible and extensible continuous-time hidden Markov model to extract meaningful activity patterns from human accelerometer data. To facilitate estimation with massive data we derive an efficient learning algorithm that exploits the hierarchical structure of the parameters indexing the proposed model. We also propose a bootstrap procedure for interval estimation. The proposed methods are illustrated using data from the 2003 - 2004 and 2005 - 2006 National Health and Nutrition Examination Survey.
Quantum many-body systems exhibit a rich and diverse range of exotic behaviours, owing to their underlying non-classical structure. These systems present a deep structure beyond those that can be captured by measures of correlation and entanglement alone. Using tools from complexity science, we characterise such structure. We investigate the structural complexities that can be found within the patterns that manifest from the observational data of these systems. In particular, using two prototypical quantum many-body systems as test cases - the one-dimensional quantum Ising and Bose-Hubbard models - we explore how different information-theoretic measures of complexity are able to identify different features of such patterns. This work furthers the understanding of fully-quantum notions of structure and complexity in quantum systems and dynamics.
We perform a systematic analysis on the large-scale taxi trip data to uncover urban mobility and city dynamics in multimodal urban transportation environments. As a case study, we use the taxi origin-destination trip data and some additional data sources in Washington DC area. We first study basic characteristics of taxi trips, then focus on five important aspects. Three of them concern urban mobility, which are respectively mobility and cost including effect of traffic congestion, trip safety, and multimodal connectivity; the other two pertain to city dynamics, which are respectively transportation resilience and the relation between trip patterns and land use. For these aspects, we use appropriate statistical methods and geographic techniques to mine patterns and characteristics from taxi trip data for better understanding qualitative and quantitative impacts of the inputs from key stakeholders on available measures of effectiveness on urban mobility and city dynamics, where key stakeholders include road users, system operators, and city. Finally, we briefly summarize our findings and discuss some critical roles and implications of the uncovered patterns and characteristics from the relation between taxi system and key stakeholders. The results can support road users by providing evidence-based information of trip cost, mobility, safety, multimodal connectivity and transportation resilience, can assist taxi drivers and operators to deliver transportation services in a higher quality of mobility, safety and operational efficiency, and can also help city planners and policy makers to transform multimodal transportation and to manage urban resources in a more effective and better way.
The state-of-the-art performance of deep learning algorithms has led to a considerable increase in the utilization of machine learning in security-sensitive and critical applications. However, it has recently been shown that a small and carefully crafted perturbation in the input space can completely fool a deep model. In this study, we explore the extent to which face recognition systems are vulnerable to geometrically-perturbed adversarial faces. We propose a fast landmark manipulation method for generating adversarial faces, which is approximately 200 times faster than the previous geometric attacks and obtains 99.86% success rate on the state-of-the-art face recognition models. To further force the generated samples to be natural, we introduce a second attack constrained on the semantic structure of the face which has the half speed of the first attack with the success rate of 99.96%. Both attacks are extremely robust against the state-of-the-art defense methods with the success rate of equal or greater than 53.59%.
We propose a new method for event extraction (EE) task based on an imitation learning framework, specifically, inverse reinforcement learning (IRL) via generative adversarial network (GAN). The GAN estimates proper rewards according to the difference between the actions committed by the expert (or ground truth) and the agent among complicated states in the environment. EE task benefits from these dynamic rewards because instances and labels yield to various extents of difficulty and the gains are expected to be diverse -- e.g., an ambiguous but correctly detected trigger or argument should receive high gains -- while the traditional RL models usually neglect such differences and pay equal attention on all instances. Moreover, our experiments also demonstrate that the proposed framework outperforms state-of-the-art methods, without explicit feature engineering.