Adaptive First-Order Methods Revisited: Convex Minimization without Lipschitz Requirements

Authors: Kimon Antonakopoulos, Panayotis Mertikopoulos

NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In Fig. 1, we report the performance of the (non-adaptive) entropic gradient descent and proportional response algorithms studied by Birnbaum et al. [10], and we compare it to the performance of ADAMIR, which consistently outperforms both methods, in terms of both last-iterate and ergodic value convergence rates. We provide a more detailed analysis in the paper s supplement. In the supplement, we also perform a numerical validation of the method in the context of a Fisher market model.
Researcher Affiliation Collaboration Kimon Antonakopoulos Panayotis Mertikopoulos Univ. Grenoble Alpes, CNRS, Inria, Grenoble INP, LIG 38000 Grenoble, France & Criteo AI Lab kimon.antonakopoulos@inria.fr panayotis.mertikopoulos@imag.fr
Pseudocode No The paper describes the ADAMIR method using mathematical equations and definitions but does not present it in a structured pseudocode or algorithm block.
Open Source Code No The paper does not provide an explicit statement about releasing source code or a link to a code repository for the methodology described.
Open Datasets No The paper conducts a case study on 'Fisher markets' where 'marginal utilities drawn i.i.d. at each epoch' are used, implying a simulated environment. It does not provide access information for a publicly available dataset.
Dataset Splits No The paper describes experiments on a simulated Fisher market but does not specify any explicit training, validation, or test dataset splits.
Hardware Specification No The paper does not provide any specific details about the hardware used for running its experiments.
Software Dependencies No The paper does not specify any software dependencies with version numbers.
Experiment Setup Yes In Fig. 1, we report the performance of the (non-adaptive) entropic gradient descent and proportional response algorithms... in a stochastic Fisher market with marginal utilities drawn i.i.d. at each epoch. The marked lines are the observed means from S = 50 realizations, whereas the shaded areas represent a 95% confidence interval.