Spatio-temporal Events Research Articles

CDSTAEP: Cross-Domain Spatial-Temporal Association Learning for Abnormal Events Prediction

Predicting network abnormal events and behavior can enhance security situation awareness and the ability to infer attack intentions. Most of the existing abnormal event prediction methods usually rely on the temporal relationship features between events and the spatial relationship features between hosts. However, the existing spatio-temporal anomaly event prediction methods do not fully consider the spatial relationship between events and the cross-domain environment of the behavior, resulting in poor performance in practical applications. In addition, the existing methods are mostly based on Euclidean space and hyperbolic space in terms of feature space relationship representation and do not fully consider the complexity of the relationship structure of anomalous events. In this paper, we propose a cross-domain spatio-temporal abnormal events prediction method, referred to as CDSTAEP. This method divides the local event sequence based on the temporal behavior sequence of entities and realizes the graphical representation of the multi-domain event correlation relationship. In the mixed-curvature space, we realize the representation learning of the correlation relationship of complex events and combine the event mixed-curvature vector representation and attention-based long short-term memory (LSTM-ATT) to capture the spatial and temporal correlation characteristics of cross-domain events, and finally realize the prediction. In this paper the proposed CDSTAEP is verified with the live network data set collected by a national key research and development plan. The results demonstrate that CDSTAEP can retain more spatial relationship features between events, the area under roc curve (AUC) score is better than the result of single-space representation and is 4.53% and 6.699% higher than the baseline models such as LSTM and LSTM-ATT.

Identifying spatio-temporal climatic characteristics and events of the South-Asian aerosol pollution transport to the Tibetan Plateau

South Asian aerosol pollutants can be transported to the Tibetan Plateau (TP) and affect the local atmospheric environment; however, few studies have produced statistical results for the long term. In this study, 41 years of pollution transport events from South Asia to the TP were screened based on Modern-Era Retrospective Analysis for Research and Applications, Version 2 (MERRA-2) reanalysis data using objective methods, and the main pollution transport mechanisms were revealed. We used the empirical orthogonal functions (EOF) and the rotated empirical orthogonal functions (REOF) methods to obtain the spatiotemporal variation characteristics of the aerosol optical depth (AOD) over the TP and the large loading regions of AOD variation from 1980 to 2020 and combined them with the backward trajectory to screen for severe pollution events in the TP influenced by transport from South Asia. The three regions (eastern, central and western regions) with large AOD loading variations in the southern TP are shown by the REOF. Based on the objectively defined AOD indices for the three regions, 2058 days of strong pollution were identified, of which 1702 days (82.7%) were affected by pollution transport from South Asia. Of these 1702 days, the central region had the highest number of days at 661 (38.8%), followed by the western and eastern regions with 544 (32%) and 497 (29.2%) days, respectively. The results obtained from the transport events in the three regions suggest that South Asian pollutants can be transported across the Himalayas to the TP by valley winds, north-south valley passages, and cyclonic uplift. These findings deepen the understanding of the aerosol pollution of the TP affected by pollution transport from South Asia.

Detecting spatiotemporal propagation patterns of traffic congestion from fine-grained vehicle trajectory data

Traffic congestion on a road segment typically begins as a small-scale spatiotemporal event that can then propagate throughout a road network and produce large-scale disruptions to a transportation system. In current techniques for the analysis of network flow, data is often aggregated to relatively large (e.g. 5 min) discrete time steps that obscure the small-scale spatiotemporal interactions that drive larger-scale dynamics. We propose a new method that handles fine-grained data to better capture those dynamics. Propagation patterns of traffic congestion are represented as spatiotemporally connected events. Each event is captured as a time series at the temporal resolution of the available trajectory data and at the spatial resolution of the network edge. The spatiotemporal propagation patterns of traffic congestion are captured using Dynamic Time Warping and represented as a set of directed acyclic graphs of spatiotemporal events. Results from this method are compared to an existing method using fine-grained data derived from an agent-based model of traffic simulation. Our method outperforms the existing method. Our method also successfully detects congestion propagation patterns that were reported by media news using sparse real-world data derived from taxis.

Lewy Body-like Pathology and Loss of Dopaminergic Neurons in Midbrain Organoids Derived from Familial Parkinson's Disease Patient.

Progressive accumulation of α-Synuclein (αSyn) in Lewy bodies (LBs) and loss of dopaminergic (DA) neurons are the hallmark pathological features of Parkinson's disease (PD). Although currently available in vitro and in vivo models have provided crucial information about PD pathogenesis, the mechanistic link between the progressive accumulation of αSyn into LBs and the loss of DA neurons is still unclear. To address this, it is critical to model LB formation and DA neuron loss, the two key neuropathological aspects of PD, in a relevant in vitro system. In this study, we developed a human midbrain-like organoid (hMBO) model of PD. We demonstrated that hMBOs generated from induced pluripotent stem cells (hiPSCs), derived from a familial PD (fPD) patient carrying αSyn gene (SNCA) triplication accumulate pathological αSyn over time. These cytoplasmic inclusions spatially and morphologically resembled diverse stages of LB formation and were composed of key markers of LBs. Importantly, the progressive accumulation of pathological αSyn was paralleled by the loss of DA neurons and elevated apoptosis. The model developed in this study will complement the existing in vitro models of PD and will provide a unique platform to study the spatiotemporal events governing LB formation and their relation with neurodegeneration. Furthermore, this model will also be beneficial for in vitro screening and the development of therapeutic compounds.

Embryonic Spinal Cord Innervation in Human Trunk Organogenesis Gastruloids: Cardiac Versus Enteric Customization and Beyond.

Trunk-biased human gastruloids provide the ability to couple developmentally relevant spinal neurogenesis and organ morphogenesis via spatiotemporal self-organization events from derivatives of the three germ layers. The multi-lineage nature of gastruloids provides the full complexity of regulatory signaling cues that surpasses directed organoids and lays the foundation for an ex vivo self-evolving system. Here we detail two distinct protocols for trunk-biased gastruloids from an elongated, polarized structure with coordinated organ-specific neural patterning. Following an induction phase to caudalize iPSCs to trunk phenotype, divergent features of organogenesis and end-organ innervation yield separate models of enteric and cardiac nervous system formation. Both protocols are permissive to multi-lineage development and allow the study of neural integration events within a native, embryo-like context. We discuss the customizability of human gastruloids and the optimization of initial and extended conditions that maintain a permissive environment for multi-lineage differentiation and integration.

Case Study on Dynamic Identification of Overburden Fracture and Strong Mine Pressure Mechanism of Isolated Working Face Based on Microseismic Clustering

Strong mine pressure has a great impact upon the safety production of coal mines. Microseismic information provides a more advanced technical means for overburden fracture dynamic identification and mine pressure mechanism research, since it contains rich information on rock fracture sources. In this study, the isolated LW8102 working face in Tongxin Mine was investigated in order to propose a spatio-temporal microseismic event data analysis method based on the k-means clustering algorithm. This algorithm can handle dynamic identification of overburden fractures constrained by spatiotemporally discrete distributions of microseismic events. This provided the dynamic extension process and the fracture distribution pattern of the overburden: eight fracture extensions were formed in the overburden. In each extension, vertical fractures connected the low and high rock layers in the LW8102 and LW8103 goafs, and through fractures connected the LW8102 and LW8103 goafs in their high, middle, and low levels. Some extensions had fractures that were connected to form a closed loop structure. In the vertical fracture, there was a tendency for one or two layers of the stratum to fail first, and then extend to one or both sides. The process of through and vertical fracture propagation followed a certain temporal sequence, reflected primarily in two forms: firstly, as the vertical fracture extended to a certain layer, it provided the initial rupture space for through fracture spreading; secondly, the through fracture first broke, and then extended to the vertical fracture until it intersected with the vertical fracture or provided the initial rupture space for the expansion of the vertical fracture. By matching the overburden fracture to the mine pressure that responded to the support resistance, we analyzed the mechanism of mine pressure at the working face. Through fracture at the high level was found to be the primary cause of the occurrence of mining pressure. It was precisely placed that the formation of multiple adjacent high through fractures 110 m from the floor, triggering simultaneous instability motion of the lower multi-layer level rock; this was the main reason for the phenomenon of strong mine pressure at the working face. Meanwhile, high through fracture at 80 m from the floor was the main reason for the phenomenon of large mine pressure at the working face.

Identification of Transcription Factor Networks during Mouse Hindlimb Development

Mammalian hindlimb development involves a variety of cells and the regulation of spatiotemporal molecular events, but regulatory networks of transcription factors contributing to hindlimb morphogenesis are not well understood. Here, we identified transcription factor networks during mouse hindlimb morphology establishment through transcriptome analysis. We used four stages of embryonic hindlimb transcription profiles acquired from the Gene Expression Omnibus database (GSE30138), including E10.5, E11.5, E12.5 and E13.5, to construct a gene network using Weighted Gene Co-expression Network Analysis (WGCNA), and defined seven stage-associated modules. After filtering 7625 hub genes, we further prioritized 555 transcription factors with AnimalTFDB3.0. Gene ontology enrichment showed that transcription factors of different modules were enriched in muscle tissue development, connective tissue development, embryonic organ development, skeletal system morphogenesis, pattern specification process and urogenital system development separately. Six regulatory networks were constructed with key transcription factors, which contribute to the development of different tissues. Knockdown of four transcription factors from regulatory networks, including Sox9, Twist1, Snai2 and Klf4, showed that the expression of limb-development-related genes was also inhibited, which indicated the crucial role of transcription factor networks in hindlimb development.

Combined Scattering, Interferometric, and Fluorescence Oblique Illumination for Live Cell Nanoscale Imaging

To determine the molecular and/or mechanical basis of cell migration using live cell imaging tools, it is necessary to correlate multiple 3D spatiotemporal events simultaneously. Fluorescence nanoscopy and label-free nanoscale imaging can complement each other by providing both molecular specificity and structural dynamics of sub-cellular structures. A combined imaging system would permit obtaining quantitative 3D spatial temporal details of individual cellular components. In this paper, we empirically determined optimal azimuthal scanning angles of rotating beams to achieve simultaneous and label-free nanoscale and fluorescence imaging. Label-free nanoscale imaging here refers to interferometric, bright-field (BF) and dark-field (DF) rotating coherence scattering (ROCS) microscopy, while fluorescence refers to high-inclined laminated oblique (HiLO) and total internal reflection fluorescence (TIRF) imaging. The combined capabilities of interferometric, scattering, and fluorescence imaging enable (1) the identification of molecular targets (substrate or organelle), (2) quantification of 3D cell morphodynamics, and (3) tracking of intracellular organelles in 3D. This combined imaging tool was then used to characterize migrating platelets and endothelial cells, both critical to the process of infection and wound healing. The combined imaging results of more than ∼1000 platelets suggest that serum albumin (bovine) is necessary for platelets to migrate and scavenge fibrin/fibrinogen. Furthermore, we identified new asynchronous membrane fluctuations between the leading and rear edges of a migrating platelet. We further demonstrated that interferometric imaging permits the quantification of mitochondrial dynamics on human lung microvascular endothelial cells. Our data suggests that axial displacement of mitochondria is minimal when it is closer to the nucleus or the leading edge of a cell membrane. Taken together, this combined nanoscopy platform helps to quantify multiple spatial temporal events of a migrating cell that will undoubtedly open ways to new quantitative correlative nanoscale live cell imaging.

Molecular regulation of conditioning film formation and quorum quenching in sulfate reducing bacteria.

Sensing surface topography, an upsurge of signaling biomolecules, and upholding cellular homeostasis are the rate-limiting spatio-temporal events in microbial attachment and biofilm formation. Initially, a set of highly specialized proteins, viz. conditioning protein, directs the irreversible attachment of the microbes. Later signaling molecules, viz. autoinducer, take over the cellular communication phenomenon, resulting in a mature microbial biofilm. The mandatory release of conditioning proteins and autoinducers corroborated the existence of two independent mechanisms operating sequentially for biofilm development. However, both these mechanisms are significantly affected by the availability of the cofactor, e.g., Copper (Cu). Generally, the Cu concentration beyond threshold levels is detrimental to the anaerobes except for a few species of sulfate-reducing bacteria (SRB). Remarkably SRB has developed intricate ways to resist and thrive in the presence of Cu by activating numerous genes responsible for modifying the presence of more toxic Cu(I) to Cu(II) within the periplasm, followed by their export through the outer membrane. Therefore, the determinants of Cu toxicity, sequestration, and transportation are reconnoitered for their contribution towards microbial adaptations and biofilm formation. The mechanistic details revealing Cu as a quorum quencher (QQ) are provided in addition to the three pathways involved in the dissolution of cellular communications. This review articulates the Machine Learning based data curing and data processing for designing novel anti-biofilm peptides and for an in-depth understanding of QQ mechanisms. A pioneering data set has been mined and presented on the functional properties of the QQ homolog in Oleidesulfovibrio alaskensis G20 and residues regulating the multicopper oxidase properties in SRB.

Chicxulub-like Gale impact into an ocean/land interface on Mars: An explanation for the formation of Mount Sharp

A geologic investigation of Gale Crater and surrounding terrain was conducted combining orbital data with reported in situ information collected through the Mars Science Laboratory (MSL Curiosity) rover cameras to better explain the anomalous features in and nearby Gale Crater with the primary focus being the formational history of Aeolis Mons, informally named Mount Sharp. Features identified through the orbital- and field-based data collectively can be best explained by an impact into the dichotomy boundary, with the northern half of the target environment being an ocean and water-enriched ignimbrite and the southern half being crustal basement/protolith of Terra Cimmeria (part ocean and part land). Such a spatial association, together with the released impact energy of the resulting ∼154 km-diameter Gale Crater, is similar to the Chicxulub impact crater on Earth in which case the target material being both ocean water and continental crust (i.e., continental shelf underlain by granite-composing continental crust) and the planetary environmental and life conditions being highly altered. This paper presents a working hypothesis, backed by orbital and ground data available thus far, for spatio-temporal events that may have transpired during impact and post-impact to the present. This proposed sequence of events and other working hypotheses will be tested by the ongoing MSL Curiosity investigation as the rover traverses up Mount Sharp.

FSOLAP: A fuzzy logic-based spatial OLAP framework for effective predictive analytics

Nowadays, with the rise in sensor technology, the amount of spatial and temporal data increases day by day. Fast, effective, and accurate analysis and prediction of collected data have become more essential than ever. Spatial Online Analytical Processing (SOLAP) emerged to perform data mining on spatial and temporal data that naturally contains the hierarchical structure used in many complex applications. In addition, uncertainty and fuzziness are inherently essential elements of data in many complex data applications, particularly in spatial–temporal database applications. In this study, FSOLAP is proposed as a new fuzzy SOLAP-based framework to compose the benefits of fuzzy logic and SOLAP concepts and is extended with inference capability to the framework to support predictive analytics. The predictive accuracy and resource utilization performance of FSOLAP are compared using real data with some well-known machine learning techniques such as Support Vector Machine, Random Forest, and Fuzzy Random Forest. The extensive experimental results show that the FSOLAP framework for the predictive analytics of various spatiotemporal events in big meteorological databases is considerably more accurate and scalable than using conventional machine learning techniques.

The glycocalyx affects the mechanotransductive perception of the topographical microenvironment

The cell/microenvironment interface is the starting point of integrin-mediated mechanotransduction, but many details of mechanotransductive signal integration remain elusive due to the complexity of the involved (extra)cellular structures, such as the glycocalyx. We used nano-bio-interfaces reproducing the complex nanotopographical features of the extracellular matrix to analyse the glycocalyx impact on PC12 cell mechanosensing at the nanoscale (e.g., by force spectroscopy with functionalised probes). Our data demonstrates that the glycocalyx configuration affects spatio-temporal nanotopography-sensitive mechanotransductive events at the cell/microenvironment interface. Opposing effects of major glycocalyx removal were observed, when comparing flat and specific nanotopographical conditions. The excessive retrograde actin flow speed and force loading are strongly reduced on certain nanotopographies upon strong reduction of the native glycocalyx, while on the flat substrate we observe the opposite trend. Our results highlight the importance of the glycocalyx configuration in a molecular clutch force loading-dependent cellular mechanism for mechanosensing of microenvironmental nanotopographical features.Graphical

LaSER: Language-specific event recommendation

While societal events often impact people worldwide, a significant fraction of events has a local focus that primarily affects specific language communities. Examples include national elections, the development of the Coronavirus pandemic in different countries, and local film festivals such as the César Awards in France and the Moscow International Film Festival in Russia. However, existing entity recommendation approaches do not sufficiently address the language context of recommendation. This article introduces the novel task of language-specific event recommendation, which aims to recommend events relevant to the user query in the language-specific context. This task can support essential information retrieval activities, including web navigation and exploratory search, considering the language context of user information needs. We propose LaSER, a novel approach toward language-specific event recommendation. LaSER blends the language-specific latent representations (embeddings) of entities and events and spatio-temporal event features in a learning to rank model. This model is trained on publicly available Wikipedia Clickstream data. The results of our user study demonstrate that LaSER outperforms state-of-the-art recommendation baselines by up to 33 percentage points in MAP@5 concerning the language-specific relevance of recommended events.

Detecting characteristics of extreme precipitation events using regional and satellite-based precipitation gridded datasets over a region in Central Europe

Perception of the spatio-temporal events of extreme precipitation and their variations is essential for diminishing the natural hazards linked with extreme events. In this research, a satellite-based precipitation dataset derived from remotely sensed soil moisture (SM2RAIN-ASCAT, obtained from ASCAT satellite soil moisture data through the Soil Moisture to Rain algorithm) was selected to evaluate the accuracy of daily precipitation and extreme events estimations against a regional gridded weather dataset by employing various performance indicators, and ETCCDI indicators (CDD, and CWD, SDII, R10mm, R20mm, R95p, R99p, Rx1day, and Rx5day). The study area includes entire Poland as well as small parts of Ukraine, Belarus, Slovakia, the Czech Republic, Russia, and Germany. According to PBIAS (~ −3.9 %) and coefficient of correlation (~0.74), SM2RAIN-ASCAT has good accuracy in the study area. Assessments reveal that, in general, over southern, mountainous part SM2RAIN-ASCAT does not have accurate estimations. According to the reference dataset, during the 2007–2019 period, on average, the length of dry days was ~22 days, while SM2RAIN-ASCAT shows ~19.6 consecutive dry days. In contrast, SM2RAIN-ASCAT overestimated (16 days/year) the consecutive wet days compared to the reference dataset (~8.7 days/year). SM2RAIN-ASCAT underestimated the number of heavy precipitation days index (R10mm) over the northern part of the region, close to the Baltic Sea), but the accuracy increased in the southern parts. SM2RAIN-ASCAT underestimated the maximum 1-day rainfall total and relative max 5-day precipitation amount indices. The total precipitation divided by the amount of wet days index shows that SM2RAIN-ASCAT has relatively acceptable accuracy in the center and south of the study area. Our results show that SM2RAIN-ASCAT should be improved for relatively higher extreme indicators.

Event stream learning using spatio-temporal event surface

Event cameras sense changes in light intensity and record them as an asynchronous event stream. Efficiently encoding and learning spatiotemporal information of the event streams remain challenging. In this paper, we propose a novel event descriptor to encode the spatio-temporal features for event streams and a local-search based multi-spike learning algorithm for spiking neural networks to classify encoded features. The spatio-temporal event surface (STES) descriptor explicitly captures both spatial and temporal correlations among events, and thus can characterize spatiotemporal features more accurately than existing feature descriptors that focus only on temporal or spatial information. In classification with multi-spike learning, we introduce a local search and gradient clipping mechanism to ensure the efficiency and stability of learning, which avoids other multi-spike learning rules’ time-consuming global search and the gradient explosion problem. Experimental results demonstrate the superior classification performance of our proposed model, especially for event streams with rich spatiotemporal dynamics.

Representing and modeling spatio-temporal uncertainty using belief function theory in flood extent mapping

In our world of rapid change, floods are a growing threat. In this context, flood extent mapping is important for damage and accurate and timely information about flood-affected areas are needed. Due to the suitability of remote sensing in mapping flood inundation, multi-source satellite images have been used in recent years. This paper presents a new approach to representing and modeling uncertainty in spatio-temporal information of satellite images using belief function theory for flooded areas extraction. Interval approaches and the associated arithmetic operations are used for modeling the uncertainty of spatial locations and temporal durations from 2D dimensional model in order to take the direction of flow in the flood. Based on topological spatial relationships, the similarity for the spatio-temporal inundation events is performed and the similar intervals are fused. Detailed examples are given to illustrate each step. A case study of flood risk in Chad Lake from 2003 to 2019 using Sentinel-2, MODIS and Landsat-8 images shows how spatio-temporal uncertainty can be problematic and bias the results of even the simplest flood mapping, and how coupling a Dempster–Shafer approach with topological relationships nevertheless offers a useful solution.

Correction to: E-ware: a big data system for the incremental discovery of spatio-temporal events from microblogs

Correction to: E-ware: a big data system for the incremental discovery of spatio-temporal events from microblogs

MDOE: A Spatiotemporal Event Representation Considering the Magnitude and Density of Events

Event-based sensors (e.g., DVS cameras) are capable of higher dynamic range, higher temporal resolution, lower time latency, and better power efficiency compared to conventional devices (e.g., RGB cameras). However, learning from these sensors remains challenging; event-based sensors output a stream of asynchronous events, which cannot be directly used by state-of-the-art convolutional neural networks (CNNs). In this paper, we present a novel event-based representation called MDOE that considers both the magnitude and density of events. Compared to existing representations, which discard one or more types of information about event polarity, temporal information, and/or density, MDOE contains richer information about events. It has two benefits: (i) it is a conceptually-simple generic representation that is task-independent; (ii) it achieves superior performance relative to existing representations on a variety of event-based datasets.

Deep-Eware: spatio-temporal social event detection using a hybrid learning model

Event detection from social media aims at extracting specific or generic unusual happenings, such as, family reunions, earthquakes, and disease outbreaks, among others. This paper introduces a new perspective for the hybrid extraction and clustering of social events from big social data streams. We rely on a hybrid learning model, where supervised deep learning is used for feature extraction and topic classification, whereas unsupervised spatial clustering is employed to determine the event whereabouts. We present ‘Deep-Eware’, a scalable and efficient event-aware big data platform that integrates data stream and geospatial processing tools for the hybrid extraction and dissemination of spatio-temporal events. We introduce a pure incremental approach for event discovery, by developing unsupervised machine learning and NLP algorithms and by computing events’ lifetime and spatial spanning. The system integrates a semantic keyword generation tool using KeyBERT for dataset preparation. Event classification is performed using CNN and bidirectional LSTM, while hierarchical density-based spatial clustering was used for location-inference of events. We conduct experiments over Twitter datasets to measure the effectiveness and efficiency of our system. The results demonstrate that this hybrid approach for spatio-temporal event extraction has a major advantage for real-time spatio-temporal event detection and tracking from social media. This leads to the development of unparalleled smart city applications, such as event-enriched trip planning, epidemic disease evolution, and proactive emergency management services.

Federated Learning for Privacy-Aware Human Mobility Modeling.

Human mobility modeling is a complex yet essential subject of study related to modeling important spatiotemporal events, including traffic, disease spreading, and customized directions and recommendations. While spatiotemporal data can be collected easily via smartphones, current state-of-the-art deep learning methods require vast amounts of such privacy-sensitive data to generate useful models. This work investigates the creation of spatiotemporal models using a Federated Learning (FL) approach—a machine learning technique that avoids sharing personal data with centralized servers. More specifically, we examine three centralized models for next-place prediction: a simple Gated Recurrent Unit (GRU) model, as well as two state-of-the-art centralized approaches, Flashback and DeepMove. Flashback is a Recurrent Neural Network (RNN) that utilizes historical hidden states with similar context as the current spatiotemporal context to improve performance. DeepMove is an attentional RNN that aims to capture human mobility's regularity while coping with data sparsity. We then implemented models based on FL for the two best-performing centralized models. We compared the performance of all models using two large public datasets: Foursquare (9,450 million check-ins, February 2009 to October 2010) and Gowalla (3,300 million check-ins, April 2012 to January 2014). We first replicated the performance of both Flashback and DeepMove, as reported in the original studies, and compared them to the simple GRU model. Flashback and GRU proved to be the best performing centralized models, so we further explored both in FL scenarios, including several parameters such as the number of clients, rounds, and epochs. Our results indicated that the training process of the federated models was less stable, i.e., the FL versions of both Flashback and GRU tended to have higher variability in the loss curves. The higher variability led to a slower convergence and thus a poorer performance when compared to the corresponding centralized models. Model performance was also highly influenced by the number of federated clients and the sparsity of the evaluation dataset. We additionally provide insights into the technical challenges of applying FL to state-of-the-art deep learning methods for human mobility.