Pith. sign in

REVIEW 3 cited by

Amortized Implicit Differentiation for Stochastic Bilevel Optimization

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2111.14580 v3 pith:6EX5ARLC submitted 2021-11-29 math.OC cs.LG

classification math.OCcs.LG
keywords algorithmsoptimizationbilevelamortizeddifferentiationexperimentsframeworkgradient
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

We study a class of algorithms for solving bilevel optimization problems in both stochastic and deterministic settings when the inner-level objective is strongly convex. Specifically, we consider algorithms based on inexact implicit differentiation and we exploit a warm-start strategy to amortize the estimation of the exact gradient. We then introduce a unified theoretical framework inspired by the study of singularly perturbed systems (Habets, 1974) to analyze such amortized algorithms. By using this framework, our analysis shows these algorithms to match the computational complexity of oracle methods that have access to an unbiased estimate of the gradient, thus outperforming many existing results for bilevel optimization. We illustrate these findings on synthetic experiments and demonstrate the efficiency of these algorithms on hyper-parameter optimization experiments involving several thousands of variables.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. A Nearly Optimal Single Loop Algorithm for Stochastic Bilevel Optimization under Unbounded Smoothness

    cs.LG 2024-12 conditional novelty 7.0 of 10

    SLIP is the first single-loop stochastic bilevel optimizer with eO(1/epsilon^4) oracle complexity under unbounded upper-level smoothness, both in expectation and with high probability.

  2. Exploring the Generalization Capabilities of AID-based Bi-level Optimization

    cs.LG 2024-11 conditional novelty 7.0 of 10

    AID-based bi-level optimization is uniformly stable with sample-dependent bounds comparable to single-level nonconvex SGD, and diminishing step sizes yield smaller generalization gaps than constant step sizes.

  3. Linear Convergence Analysis of Single-loop Algorithm for Bilevel Optimization via Small-gain Theorem

    math.OC 2024-12 conditional novelty 6.0 of 10

    Using a small-gain argument from control theory, the authors prove the standard single-loop bilevel optimization algorithm converges linearly in the strongly convex setting.

Pith tools