QFCQT: A Chaotically Gated Quantformer Framework for Volatile Time-Series Forecasting

Name: QFCQT: A Chaotically Gated Quantformer Framework for Volatile Time-Series Forecasting
Uploaded: 2026-08-10T22:28:28.104003+00:00
Duration: 758 s
Description: The hosts discuss a paper proposing QFCQT, a transformer for volatile time-series forecasting that uses a chaotic Lee Oscillator activation, gated with GELU, to handle abrupt changes. They highlight strong MSE improvements on volatile datasets like ETTh2, and conclude that activation design deserves more attention in forecasting research.

The hosts discuss a paper proposing QFCQT, a transformer for volatile time-series forecasting that uses a chaotic Lee Oscillator activation, gated with GELU, to handle abrupt changes. They highlight strong MSE improvements on volatile datasets like ETTh2, and conclude that activation design deserves more attention in forecasting research.

Open the full episode