Tag
Introduces CardioState-JEPA, a cardiac foundation model that learns a shared representation across ECG, PPG, and PCG signals using a delay-aware joint-embedding predictive architecture, improving downstream cardiac classification tasks.
Proposes temporal-distance JEPA (TD-JEPA) which mines directed temporal cost from offline trajectories to improve latent world model predictive control, achieving higher success rates on robotic environments.
This paper presents a tutorial on using Joint-Embedding Predictive Architecture (JEPA) for self-supervised learning in 6G networks, along with a beam management case study and open challenges.
Micro-JEPA is a lightweight Python implementation of the Joint Embedding Predictive Architecture (JEPA), enabling an agent to learn environment representations, predict future states in latent space, and plan actions to avoid obstacles.
Proposes CF-JEPA, a mask-free self-supervised learning framework for time-series that uses multi-horizon forward prediction from random crops and exploits asymmetry between online and target encoders for improved performance on classification, forecasting, and anomaly detection.