Results for "environment representation"

Agents fail to coordinate optimally.

Inferring and aligning with human preferences.

Decisions dependent on others’ actions.

Learning from data by constructing “pseudo-labels” (e.g., next-token prediction, masked modeling) without manual annotation.

A mismatch between training and deployment data distributions that can degrade model performance.

Methods to protect model/data during inference (e.g., trusted execution environments) from operators/attackers.

All possible configurations an agent may encounter.

Set of all actions available to the agent.

Simple agent responding directly to inputs.

Maintaining alignment under new conditions.

Train/test environment mismatch.

Hardware components that execute physical actions.

Devices measuring physical quantities (vision, lidar, force, IMU, etc.).

Software pipeline converting raw sensor data into structured representations.

Continuous loop adjusting actions based on state feedback.

Control without feedback after execution begins.

Modeling interactions with environment.

Robots made of flexible materials.

Randomizing simulation parameters to improve real-world transfer.

Artificial sensor data generated in simulation.

Directly optimizing control policies.

Modifying reward to accelerate learning.

Learning policies from expert demonstrations.

Inferring reward function from observed behavior.

Estimating robot position within a map.

Imagined future trajectories.

Human-like understanding of physical behavior.

Human controlling robot remotely.

Ensuring robots do not harm humans.

Intelligence emerges from interaction with the physical world.