Results for "shortcut learning"

Adjusting learning rate over training to improve convergence.

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

Selecting the most informative samples to label (e.g., uncertainty sampling) to reduce labeling cost.

Learning from data generated by a different policy.

Learning a function from input-output pairs (labeled data), optimizing performance on predicting outputs for unseen inputs.

A model is PAC-learnable if it can, with high probability, learn an approximately correct hypothesis from finite samples.

Learning where data arrives sequentially and the model updates continuously, often under changing distributions.

Learning policies from expert demonstrations.

A branch of ML using multi-layer neural networks to learn hierarchical representations, often excelling in vision, speech, and language.

Learning structure from unlabeled data, such as discovering groups, compressing representations, or modeling data distributions.

A learning paradigm where an agent interacts with an environment and learns to choose actions to maximize cumulative reward.

Methods that learn training procedures or initializations so models can adapt quickly to new tasks with little data.

Automatically learning useful internal features (latent variables) that capture salient structure for downstream tasks.

Controls the size of parameter updates; too high diverges, too low trains slowly or gets stuck.

A theoretical framework analyzing what classes of functions can be learned, how efficiently, and with what guarantees.

Learning only from current policy’s data.

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

Training with a small labeled dataset plus a larger unlabeled dataset, leveraging assumptions like smoothness/cluster structure.

Achieving task performance by providing a small number of examples inside the prompt without weight updates.

Training across many devices/silos without centralizing raw data; aggregates updates, not data.

Gradually increasing learning rate at training start to avoid divergence.

Built-in assumptions guiding learning efficiency and generalization.

Modifying reward to accelerate learning.

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

Iterative method that updates parameters in the direction of negative gradient to minimize loss.

Reinforcement learning from human feedback: uses preference data to train a reward model and optimize the policy.

Methods like Adam adjusting learning rates dynamically.

Inferring reward function from observed behavior.

Learning by minimizing prediction error.

Humans assist or override autonomous behavior.