KairosHope: A Next-Generation Time-Series Foundation Model for Specialized Classification via Dual-Memory Architecture

In three linesKairosHope is a time-series foundation model replacing quadratic attention with dual-memory architecture (Titans modules + Continuum Memory System). Pre-trained on Monash archive via MTSM and contrastive learning, it fuses latent representations with statistical features. Superior results on UCR for HAR and sensor data.

Summary generated by Claude — human-verified