AI Paper Atlas

Research map / Learning theory and algorithms / Kernels and regression theory

Evaluated CMI Bounds for Meta Learning: Tightness and Expressiveness

Fredrik Hellström, Giuseppe Durisi

NeurIPS 2022 · Learning theory and algorithms › Kernels and regression theory · approach: Information-theoretic generalization bounds

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use the two-step e-CMI derivation instead of earlier mutual-information meta-learning boundsTwo-step square-root bound · Average meta-learning generalization boundbound tightnessImproved
Replace the two-step derivation with a one-step derivation over joint task and sample selectionOne-step square-root bound · Average meta-learning generalization bounddecay rate in n and mImproved
Use disintegrated CMI terms instead of mutual information with the full training datae-CMI bounds · Deterministic learners with continuous outputsbound finitenessImproved
Use binary KL divergence bounds instead of square-root boundse-CMI bounds · Low training loss regimesample-size dependenceImproved
Assume zero training loss (interpolation) in the binary KL boundBinary KL bound · Interpolating regimerate of decayImproved
Use unobserved-task training loss as auxiliary loss instead of observed-task test loss (ablation)Two-step square-root bound · Representation learning with Natarajan and VC classesdecoupling of complexity termsImproved
Apply the one-step bound to Natarajan and VC dimension classes in the interpolating settinge-CMI bound · Representation learning, zero training lossgeneralization rateImproved
Convert average bounds to high-probability bounds via probabilistic KL boundsTwo-step and one-step bounds · Representation learningrecovery of classical ratesImproved
Specialize to empirical risk minimizers with a task-diversity assumptionOne-step bound · Excess risk on a fixed target taskexcess risk scalingImproved
Use the two-step bound versus the one-step bound for the combined complexity scalingHigh-probability e-CMI bounds · Representation learningscaling in n and mMixed

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy