Survey of machine learning methods for spatiotemporal sequence forecasting.
problem Forecasting multi-step future of spatiotemporal systems based on past observations.
method Defined STSF problem, classified into subcategories, identified challenges, reviewed existing methods.
result No unified perspective on machine learning for STSF previously existed.
Spacetimeformer learns spatiotemporal relationships from data alone.
problem Forecasting multivariate time series with distinct spatial relationships.
method Transformers with dynamic graph connections learning interactions between space, time, and value.
result Competitive results on various time series prediction benchmarks.
CloudLSTM forecasts geospatial point-cloud data streams accurately.
problem Forecasting over geospatial point-cloud data streams.
method Introduces CloudLSTM, a recurrent neural model with a Dynamic Point-cloud Convolution (DConv) operator.
result CloudLSTM outperforms competitor models in long-term predictions for point-cloud data.
STConvS2S improves weather forecasting using only convolutional layers.
problem Predicting future weather conditions more accurately.
method Proposes a deep learning architecture combining spatiotemporal convolutional layers.
result Outperforms state-of-the-art architectures for forecasting tasks.
Enformer and GEnformer use Transformers with stochastic learning to forecast multivariate and spatiotemporal data with uncertainty.
problem Uncertainty quantification in multivariate time series and spatiotemporal forecasting.
method Synthesizing Transformer's expressive power with stochastic learning to model conditional distributions directly.
result Enformer and GEnformer yield calibrated probabilistic forecasts and outperform state-of-the-art baselines.
Study develops advanced models to forecast complex LOB data.
problem Forecasting high-frequency data in a limit order book (LOB).
method Advanced multidimensional sequence-to-sequence models with compound multivariate embedding.
result Method outperforms other multivariate forecasting methods, achieving lowest forecasting error.
A deep learning model predicts traffic conditions over multiple steps.
problem Multistep traffic forecasting on road networks.
method Attention Graph Convolutional Sequence-to-Sequence model (AGC-Seq2Seq) with attention mechanism.
result AGC-Seq2Seq model outperforms other models in multistep traffic prediction.
Paper uses DMD to embed time in spatiotemporal forecasting.
problem Forecasting long-range seasonal dependencies in spatiotemporal data.
method Dynamic Mode Decomposition (DMD) for time representation.
result DMD-based embedding improves long-horizon forecasting accuracy.
Deep models forecast epidemics with uncertainty quantification.
problem Accurate probabilistic forecasting of epidemics is challenging due to nonlinear temporal dependencies and spatial interactions.
method Deep spatiotemporal engression methods with geometric ergodicity and asymptotic stationarity.
result Proposed methods outperform benchmarks in point and probabilistic forecasting.
ProGen improves spatiotemporal forecasting with SDEs and diffusion models.
problem Complex spatial and temporal dependencies in spatiotemporal data.
method ProGen uses Stochastic Differential Equations and diffusion-based generative models.
result ProGen outperforms state-of-the-art models on traffic datasets.
This paper studies uncertainty quantification in deep spatiotemporal forecasting.
problem Uncertainty quantification in deep spatiotemporal forecasting models.
method Analysis of UQ methods from Bayesian and frequentist perspectives, including statistical decision theory.
result Different UQ methods have different strengths and weaknesses, with Bayesian methods being more robust in mean prediction and frequentist methods providing more extensive coverage.
A new CNN approach for time series forecasting outperforms traditional RNNs.
problem Time series forecasting using conventional RNNs.
method Temporally folded convolutional neural networks (TFC's) for sequence forecasting.
result TFC's outperform conventional RNNs on sequential MNIST and JSB chorals datasets.
Predicts financial asset dependencies using spatiotemporal patterns.
problem Complex dependency structures in financial assets for risk mitigation.
method Proposes Asset Dependency Matrix (ADM) and Asset Dependency Neural Network (ADNN) with ConvLSTM for spatiotemporal asset dependency prediction.
result ADNN outperforms baselines in predicting asset dependencies and their applications.
This study improves weather forecasting accuracy with spatiotemporal models.
problem Complexity and resource-intensive nature of weather forecasting.
method Spatiotemporal forecasting models integrating machine learning and deep neural networks.
result Spatiotemporal models reduce computational costs and improve accuracy.
DYffusion improves diffusion models for spatiotemporal forecasting.
problem Challenges in generating stable and accurate forecasts for dynamic data.
method Leverages temporal dynamics in data, directly coupling it with diffusion steps.
result Improves computational efficiency and performs competitively on complex dynamics.
TK-GCN forecasts spatiotemporal dynamics using Koopman-enhanced graph convolutional networks.
problem Forecasting complex spatiotemporal dynamics over irregular domains.
method Two-stage framework: Koopman-enhanced Graph Convolutional Network (K-GCN) for spatial encoding and Transformer for temporal modeling.
result TK-GCN outperforms state-of-the-art methods in spatiotemporal cardiac dynamics forecasting.
Paper proposes a novel approach to improve spatiotemporal precipitation forecasts.
problem Improving accuracy of spatiotemporal precipitation forecasts for flood damage mitigation.
method Introduces a rain-code fusion approach using ConvLSTM and multi-frame fusion for spatiotemporal precipitation code-to-code forecasting.
result Demonstrates enhanced accuracy in precipitation forecasts beyond 3 timesteps using the rain-code fusion.
This study forecasts climate data in Chile using EOFs and machine learning models.
problem Predicting climatic variability in Chile for resource management and planning.
method Combines EOF decomposition, wavelet analysis, and neural networks for spatiotemporal forecasting.
result Improved accuracy in forecasting climate data through a hybrid ML approach.
A new machine learning model forecasts COVID-19 incidence at county level in the USA.
problem Inaccurate disease spread forecasting due to spatiotemporal homogeneity assumptions.
method Spatiotemporal machine learning using LSTM architecture with spatial and temporal features.
result COVID-LSTM outperforms COVID-19 Forecast Hub's Ensemble model in accuracy.
We propose a generic spatiotemporal event forecasting method, which we developed for the National Institute of Justice's (NIJ) Real-Time Crime Forecasting Challenge. Our method is a spatiotemporal forecasting model combining scalable randomized Reproducing Kernel Hilbert Space (RKHS) methods for approximating Gaussian …
Enhances traffic forecasting with dynamic regression incorporating error modeling.
problem Improving accuracy of traffic forecasts using deep spatiotemporal models.
method Integrates matrix-variate autoregressive (AR) model into loss function for error series of base model.
result Improved traffic forecasting performance on SOTA models with interpretable AR coefficients.
Unified model forecasts epidemics with spatial and temporal dynamics.
problem Limited accuracy in traditional models and lack of interpretability in deep learning models.
method CSTGNN integrates Spatio-Contact SIR model with Graph Neural Networks.
result Effective spatiotemporal epidemic forecasting with interpretability.
GraphSVR forecasts urban air pollution robustly across stations and seasons.
problem Nonlinear, nonstationary, spatiotemporally dependent urban air pollution forecasting challenges.
method Combines graph convolutional learning and support vector regression.
result GraphSVR improves predictive accuracy and maintains stable performance across seasons and outlier-prone episodes.
A new model learns demand patterns from data, reducing complexity and improving accuracy.
problem Forecasting short-term demand from spatiotemporal data with complex patterns.
method Temporal-Guided Network (TGNet) using graph networks and temporal-guided embedding.
result TGNet achieves competitive performance with fewer parameters compared to state-of-the-art models.
A new network log-ARCH model improves stock market volatility forecasting.
problem Improving stock market volatility forecasting accuracy.
method Dynamic network autoregressive conditional heteroscedasticity (ARCH) model integrating lagged and adjacent node volatility information.
result The model shows significant improvements in forecasting accuracy compared to univariate log-ARCH models.
MIM networks predict non-stationary spatiotemporal dynamics using differential signals.
problem Predicting non-stationary spatiotemporal processes with high-order variations.
method Memory In Memory (MIM) networks with cascaded memory modules.
result Achieved state-of-the-art results on four spatiotemporal prediction tasks.
CauSTream forecasts streamflow by integrating causal graphs for better interpretability.
problem Streamflow forecasting lacks interpretability and generalization due to fixed causal models.
method CauSTream learns causal graphs for meteorological forcings and routing dependencies.
result CauSTream outperforms existing methods, especially at longer forecast windows.
Graph Neural Networks improve El Niño forecasts.
problem Improving seasonal forecasting models for ENSO.
method Application of spatiotemporal Graph Neural Networks.
result Preliminary results outperform state-of-the-art systems for 1 and 3-month projections.
New method improves traffic forecasting models by adapting to spatial shifts.
problem Improving traffic forecasting models' ability to handle spatial shifts over years.
method Proposes a novel Mixture of Experts (MoE) framework for spatiotemporal models.
result Significant improvement in performance for handling spatial distribution shifts.
PeakWeather provides Swiss weather station data for machine learning.
problem Accurate weather forecasting for various activities and decision-making.
method High-quality dataset of surface weather observations from 8 years of Swiss stations.
result Dataset supports a wide range of spatiotemporal tasks.
New CapsNet and NLSTM framework improves traffic forecasting accuracy.
problem Accurate traffic forecasting for complex transportation networks.
method CapsNet for spatial features, nested LSTM for temporal dependencies.
result Framework outperforms multiple baseline models in Beijing network.
TREP learns pedestrian trajectories efficiently without needing full datasets.
problem Learning fixed-length vector representations of variable-length trajectories.
method Actor-critic sequence-to-sequence autoencoder with spatial-aware objective function.
result TREP efficiently learns trajectory representations without needing full datasets.
Proposes LC-ST-FCN for better ride-sourcing demand forecasting.
problem Local statistical differences in ride-sourcing demand across a city.
method LC-ST-FCN framework combining 3D and 2D convolutions, locally connected layers.
result Significant improvements in demand forecasting compared to baselines.
NAOMI improves imputation accuracy for long-range sequences.
problem Missing value imputation in spatiotemporal data.
method Non-autoregressive deep generative model exploiting multiresolution structure.
result Significant improvement in imputation accuracy (60% reduction in average prediction error).
IGNNK uses GNN for spatiotemporal kriging, improving scalability and transferability.
problem Efficiently recovering signals for unsampled locations in spatiotemporal data.
method Developed an Inductive Graph Neural Network Kriging (IGNNK) model to learn spatial message passing.
result IGNNK effectively learns spatial message passing and can be transferred to new graph structures.
New model predicts citywide air pollution using deep learning.
problem Predicting air quality for entire cities is challenging due to spatiotemporal factors.
method Convolutional Long Short-Term Memory (ConvLSTM) model for spatiotemporal data.
result ConvLSTM model outperforms other methods in predicting citywide air pollution.
EGDL predicts TB outbreaks with deep learning, integrating epidemiological models.
problem Predicting TB outbreaks with complex spatiotemporal dynamics.
method Modified MN-SIR model with Bayesian inference, deep neural networks.
result EGDL delivers robust and accurate TB outbreak predictions.
Proposes a method to improve urban spatiotemporal forecasting using multi-modal graph interaction.
problem Improving spatiotemporal forecasting in urban areas using graph convolution networks.
method Develops modality interaction mechanisms for multi-graph convolution networks to reduce generalization error.
result Proposed techniques improve prediction accuracy and model robustness compared to state-of-the-art baselines.
PSTN improves traffic condition forecasting with deep neural networks.
problem Challenges in accurately forecasting traffic conditions due to complex spatiotemporal correlations.
method Proposes PSTN with three modules: graph convolutional network, temporal convolutional network, and gated recurrent unit framework.
result Significantly outperforms state-of-the-art benchmarks in short-term traffic conditions forecasting.
ERDM integrates rolling forecasts with diffusion models for complex dynamics.
problem Forecasting complex dynamics with rolling forecasts and diffusion models.
method Adapting EDM components for rolling forecasts, introducing novel loss weighting, efficient initialization, and hybrid architecture.
result ERDM outperforms diffusion-based baselines in 2D Navier-Stokes simulations and ERA5 weather forecasting.
Deep learning predicts availability of mobile crowdsourced services spatially and temporally.
problem Predicting the availability of mobile crowdsourced services in space and time.
method Two-stage prediction model: clustering services into regions, then forecasting availability duration using time series.
result Effectiveness validated through multiple experiments.
Graph-partitioning-based DCRNN improves traffic forecasting for large highways.
problem Challenges in accurately forecasting traffic on large highway networks.
method Graph-partitioning method to decompose large networks into smaller, independent networks.
result Demonstrated improved traffic forecasting on a large California highway network.
Paper develops a new model for forecasting ocean currents.
problem Forecasting the Loop Current and its eddies for the Gulf of Mexico.
method Physics-informed Tensor-train ConvLSTM, incorporating prior physical knowledge.
result PITT-ConvLSTM outperforms state-of-the-art methods in volumetric velocity forecasting.
Combines ML and KB modeling for large chaotic systems.
problem Predicting large, complex, spatiotemporal systems with limited data.
method Parallel ML prediction and hybrid approach combining ML and KB.
result Excellent performance and reduced training data needed.
LSTM models predict low likelihood of another COVID-19 wave in India.
problem Inaccurate and unreliable COVID-19 infection forecasting models due to data limitations and model complexity.
method Application of LSTM, bidirectional LSTM, and encoder-decoder LSTM models for multi-step infection forecasting.
result Predictions indicate low likelihood of another wave in October and November 2021.
Model predicts COVID-19 spread with better accuracy than existing methods.
problem Limited daily samples in time for data-driven methods.
method Integrated spatiotemporal model combining epidemic differential equations and RNN.
result Model outperforms existing methods in forecasting cases.
New model fills in missing traffic data efficiently.
problem Missing data in large-scale spatiotemporal traffic data.
method Developed scalable tensor learning model LSTC-Tubal for imputation.
result LSTC-Tubal achieves high accuracy with lower computational cost.
Paper presents a method for imputing and forecasting structural response from incomplete sensor data.
problem Missing sensor data in structural health monitoring (SHM).
method Incremental Bayesian tensor learning for spatiotemporal missing data reconstruction and forecasting.
result The proposed method achieves accurate and robust imputation and prediction even with high rates of missing data.