FaST: Efficient and Effective Long-Horizon Forecasting for Large-Scale Spatial-Temporal Graphs via Mixture-of-Experts

Dimensions overall score 8.0

GitHub Code Pulse

No public code linked for this paper yet.

• FaST enables one-week-ahead (672 steps) long-horizon forecasting for large-scale spatial-temporal graphs with thousands of nodes.technical 95%
• FaST utilizes an adaptive graph agent attention mechanism to reduce computational burden on large-scale graphs.method 95%
• FaST employs a parallel Mixture-of-Experts (MoE) module with Gated Linear Units (GLUs) for efficient and scalable processing.method 95%
• FaST achieves superior long-horizon predictive accuracy compared to state-of-the-art baselines on real-world datasets.result 90%
• FaST demonstrates remarkable computational efficiency compared to state-of-the-art baselines.result 90%
• The approach requires an accurate graph structure, which may not always be available in real-world scenarios.limitation 90%
• The complexity of setup and maintenance in municipalities with varying technological adoption rates could pose challenges.limitation 85%
• The market opportunity for FaST is significant due to the global push towards smart cities and infrastructure optimization.market 85%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card