Results for "transfer problem"

Alternative formulation providing bounds.

Reusing knowledge from a source task/domain to improve learning on a target task/domain, typically via pretrained models.

Ensuring AI allows shutdown.

Ensuring AI systems pursue intended human goals.

Protecting data during network transfer and while stored; essential for ML pipelines handling sensitive data.

Training a smaller “student” model to mimic a larger “teacher,” often improving efficiency while retaining performance.

Assigning category labels to images.

Updating a pretrained model’s weights on task-specific data to improve performance or adapt style/behavior.

Activation max(0, x); improves gradient flow and training speed in deep nets.

Gradients grow too large, causing divergence; mitigated by clipping, normalization, careful init.

Gradients shrink through layers, slowing learning in early layers; mitigated by ReLU, residuals, normalization.

Converts constrained problem to unconstrained form.

Methods for breaking goals into steps; can be classical (A*, STRIPS) or LLM-driven with tool calls.

Breaking tasks into sub-steps.

Optimizes future actions using a model of dynamics.

Computing joint angles for desired end-effector pose.

Training one model on multiple tasks simultaneously to improve generalization through shared structure.

Ordering training samples from easier to harder to improve convergence or generalization.

Techniques that fine-tune small additional components rather than all weights to reduce compute and storage.

PEFT method injecting trainable low-rank matrices into layers, enabling efficient fine-tuning.

Increasing performance via more data.

Task instruction without examples.

Startup latency for services.

Hardware components that execute physical actions.

Continuous loop adjusting actions based on state feedback.

Control without feedback after execution begins.

Mathematical framework for controlling dynamic systems.

The physical system being controlled.

Randomizing simulation parameters to improve real-world transfer.

Combining simulation and real-world data.