Investigating functional consistency of mobility-related urban zones via motion-driven embedding vectors and local POI-type distributions.

Urban morphology and human mobility are two sides of the complex mixture of elements that implicitly define urban functionality. By leveraging the emerging availability of crowdsourced data, we aim for novel insights on how they relate to each other, which remains a substantial scientific challenge. Specifically, our study focuses on extracting spatial-temporal information from taxi trips in an attempt on grouping urban space based on human mobility, and subsequently assess its potential relationship with urban functional characteristics in terms of local points-of-interest (POI) distribution. Proposing a vector representation of urban areas, constructed via unsupervised machine learning on trip data's temporal and geographic factors, the underlying idea is to define areas as "related" if they often act as destinations of similar departing regions at similar points in time, regardless of any other explicit information. Hidden relations are mapped within the generated vector space, whereby areas are represented as points and stronger/weaker relatedness is conveyed through relative distances. The mobility-related outcome is then compared with the POI-type distribution across the urban environment, to assess the functional consistency of mobility-based clusters of urban areas. Results indicate a meaningful relationship between spatial-temporal motion patterns and urban distributions of a diverse selection of POI-type categorizations, paving the way to ideally identify homogenous urban functional zones only based on the movement of people. Our data-driven approach is intended to complement traditional urban development studies on providing a novel perspective to urban activity modeling, standing out as a reference for mining information out of mobility and POI data types in the context of urban management and planning.

TraceBERT-A Feasibility Study on Reconstructing Spatial-Temporal Gaps from Incomplete Motion Trajectories via BERT Training Process on Discrete Location Sequences.

Trajectory data represent an essential source of information on travel behaviors and human mobility patterns, assuming a central role in a wide range of services related to transportation planning, personalized recommendation strategies, and resource management plans. The main issue when dealing with trajectory recordings, however, is characterized by temporary losses in the data collection, causing possible spatial-temporal gaps and missing trajectory segments. This is especially critical in those use cases based on non-repetitive individual motion traces, when the user's missing information cannot be directly reconstructed due to the absence of historical individual repetitive routes. Inserted in the context of location-based trajectory modeling, we tackle the problem by proposing a technical parallelism with the natural language processing domain. Specifically, we introduce the use of the Bidirectional Encoder Representations from Transformers (BERT), a state-of-the-art language representation model, into the trajectory processing research field. By training deep bidirectional representations from unlabeled location sequences, jointly conditioned on both left and right context, we derive an explicit predicted estimation of the missing locations along the trace. The proposed framework, named TraceBERT, was tested on a real-world large-scale trajectory dataset of short-term tourists, exploring an effective attempt of adapting advanced language modeling approaches into mobility-based applications and demonstrating a prominent potential on trajectory reconstruction over traditional statistical approaches.

A comprehensive transferability evaluation of U-Net and ResU-Net for landslide detection from Sentinel-2 data (case study areas from Taiwan, China, and Japan).

Earthquakes and heavy rainfalls are the two leading causes of landslides around the world. Since they often occur across large areas, landslide detection requires rapid and reliable automatic detection approaches. Currently, deep learning (DL) approaches, especially different convolutional neural network and fully convolutional network (FCN) algorithms, are reliably achieving cutting-edge accuracies in automatic landslide detection. However, these successful applications of various DL approaches have thus far been based on very high resolution satellite images (e.g., GeoEye and WorldView), making it easier to achieve such high detection performances. In this study, we use freely available Sentinel-2 data and ALOS digital elevation model to investigate the application of two well-known FCN algorithms, namely the U-Net and residual U-Net (or so-called ResU-Net), for landslide detection. To our knowledge, this is the first application of FCN for landslide detection only from freely available data. We adapt the algorithms to the specific aim of landslide detection, then train and test with data from three different case study areas located in Western Taitung County (Taiwan), Shuzheng Valley (China), and Eastern Iburi (Japan). We characterize three different window size sample patches to train the algorithms. Our results also contain a comprehensive transferability assessment achieved through different training and testing scenarios in the three case studies. The highest f1-score value of 73.32% was obtained by ResU-Net, trained with a dataset from Japan, and tested on China's holdout testing area using the sample patch size of 64 × 64 pixels.

Trace2trace-A Feasibility Study on Neural Machine Translation Applied to Human Motion Trajectories.

Neural machine translation is a prominent field in the computational linguistics domain. By leveraging the recent developments of deep learning, it gave birth to powerful algorithms for translating text from one language to another. This study aims to assess the feasibility of transferring the neural machine translation approach into a completely different context, namely human mobility and trajectory analysis. Building a conceptual parallelism between sentences (sequences of words) and motion traces (sequences of locations), we aspire to translate individual trajectories generated by a certain category of users into the corresponding mobility traces potentially generated by a different category of users. The experiment is inserted in the background of tourist mobility analysis, with the goal of translating the motion behavior of tourists belonging to a specific nationality into the motion behavior of tourists belonging to a different nationality. The model adopted is based on the seq2seq approach and consists of an encoder-decoder architecture based on long short-term memory (LSTM) neural networks and neural embeddings. The encoder turns an input location sequence into a corresponding hidden vector; the decoder reverses the process, turning the vector into an output location sequence. The proposed framework, tested on a real-world large-scale dataset, explores an effective attempt of motion transformation between different entities, arising as a potentially powerful source of mobility information disclosure, especially in the context of crowd management and smart city services.