Discrete Prototypical Memories for Federated Time Series Foundation Models

1. 1Identify the limitations of existing LLMs in handling time series data
2. 2Develop discrete prototypical memories to align time series data with the latent space of LLMs
3. 3Implement federated learning mechanisms to preserve data privacy
4. 4Evaluate the performance of the proposed approach on time series data

Data scientists and AI engineers working on time series data and federated learning can benefit from this research as it enhances the performance of foundation models

💡 Discrete prototypical memories can mitigate semantic misalignment between time series data and LLMs

📈 Improve time series forecasting with discrete prototypical memories and federated learning