Urban Vibrancy Embedding and Application on Traffic Prediction
Urban Vibrancy embeddings from real-time data improve traffic prediction via VAE and LSTM.
Feb 9, 2025
Multiple Area Feature-aware Spatio-temporal Public Transit Prediction
This paper introduces the ST-MFGCRN model, a spatio-temporal multi-feature-aware graph convolutional recurrent network for improving public transit prediction by integrating multiple regional features and using sentinel attention to handle feature relevance, showing up to 12% and 7% improvement over state-of-the-art models on the Daejeon and TaxiBJ datasets, respectively.
Aug 23, 2024
Improving Real Estate Appraisal with POI Integration and Areal Embedding
This study addresses the impact of Points of Interest (POI) on property values and introduces the AMMASI model, which enhances real estate appraisal by incorporating road-network-based Areal Embedding and masked multi-head attention, significantly improving house price prediction compared to existing models.
Nov 20, 2023
*Enhancing Spatiotemporal Traffic Prediction through Urban Human Activity Analysis
Improved traffic prediction using vehicle trajectory generation based on human activity for GCN models.
Oct 25, 2023
Spatio-Temporal Road Traffic Prediction using Real-time Regional Knowledge
Proposes a method for mid-term traffic prediction using region-level data, improving road-level prediction accuracy over baselines.
Aug 23, 2021
Exploring Commercial Gentrification using Instagram Data
This paper proposes a novel method to estimate the status and progression of commercial gentrification in a target area using Instagram data by analyzing social features from images and texts, defining gentrification phases based on urban studies, and employing regression models for phase prediction, showing promising results for use by urban planners and policymakers.
Dec 7, 2020
Discovering Daily POI Exploitation Using LTE Cell Tower Access Traces in Urban Environment
This paper demonstrates that LTE cell tower access trace data can reveal the types of Points of Interest (POIs) in an area and introduces a daily POI exploitation discovery scheme to analyze how POIs are used daily, offering valuable insights for urban planning and tourism services.
Oct 7, 2020
Human Activity Recognition Using Semi-supervised Multi-modal DEC for Instagram Data
This paper proposes a semi-supervised multi-modal deep embedding clustering method for Human Activity Recognition (HAR) on Instagram, which combines supervised and unsupervised learning to leverage both labeled and unlabeled data, improving recognition accuracy and outperforming existing approaches.
May 6, 2020