NeST: Neighborhood-aware semantic alignment and temporal modulation for LLM based time series forecasting

Name: NeST: Neighborhood-aware semantic alignment and temporal modulation for LLM based time series forecasting
Uploaded: 2026-09-02T17:45:30.383975+00:00
Duration: 554 s
Description: The discussion of 'NeST' explores a method for LLM-based time series forecasting. The paper addresses aligning continuous numerical data with discrete language models by using Time Series Compatible Text Prototypes (TCTPs). The hosts conclude that this approach is highly efficient, achieving strong performance with minimal training data.

The discussion of 'NeST' explores a method for LLM-based time series forecasting. The paper addresses aligning continuous numerical data with discrete language models by using Time Series Compatible Text Prototypes (TCTPs). The hosts conclude that this approach is highly efficient, achieving strong performance with minimal training data.

Open the full episode