Free Reads
Sign in to view your remaining parses.
Tag Filter
Self-Supervised Learning
AdaWorld: Learning Adaptable World Models with Latent Actions
Published:3/25/2025
Autoregressive World ModelsLatent Action ExtractionSelf-Supervised LearningHighly Adaptable ModelsLimited Interaction Environments
The paper presents AdaWorld, a novel approach to learning adaptable world models by extracting latent actions through selfsupervised learning from videos, enabling efficient transfer and learning in diverse environments. Experiments show AdaWorld excels in simulation quality and
02
Disentangled Self-Supervision in Sequential Recommenders
Published:8/20/2020
Sequential Recommender SystemsSelf-Supervised LearningSequence-to-Sequence TrainingIntention DisentanglementFuture Behavior Sequence Reconstruction
The paper introduces a latent selfsupervised and disentangled sequencetosequence training strategy to address myopic predictions and lack of diversity in traditional sequential recommenders, showing significant performance improvements on real and synthetic datasets.
02