DeepMind Control
PulseAugur coverage of DeepMind Control — every cluster mentioning DeepMind Control across labs, papers, and developer communities, ranked by signal.
1 day(s) with sentiment data
-
New MINT method improves off-policy RL by tracking trajectory meeting times
Researchers have introduced MINT (Meeting-time Indicators for Truncation), a novel method for off-policy reinforcement learning that addresses the trade-off between variance control and signal propagation in multi-step …
-
JEPA-Anything framework unifies world model creation across 7 domains
Researchers have introduced JEPA-Anything, a novel domain-agnostic framework designed to build world models. This framework extends joint-embedding predictive architectures (JEPAs) by incorporating Orthogonal Predictive…
-
StarWM introduces self-supervised attention routing for robust AI world models
Researchers have introduced StarWM, a novel self-supervised learning approach for world models in AI. StarWM employs a cross-attention module to selectively apply reconstruction to relevant environmental dynamics, there…
-
New TrojanWorld framework backdoors reinforcement learning agents via imagination steering
Researchers have developed a new framework called TrojanWorld designed to backdoor world-model agents used in reinforcement learning. This framework exploits the predictive core of these agents by steering their interna…
-
New RL algorithm unifies model-free efficiency with model-based representations
A research paper introduces Unified Latent Dynamics (ULD), a new reinforcement learning algorithm designed to combine the efficiency of model-free methods with the representational power of model-based approaches. ULD a…