Efficient Online Learning for Dynamic k-Clustering

Authors: Dimitris Fotakis, Georgios Piliouras, Stratis Skoulakis

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

Reproducibility Variable Result LLM Response
Research Type Experimental In Section 6, we present the results of an experimental evaluation, indicating that for client locations generated in a variety of natural and practically relevant ways, the realized regret of the proposed algorithms is way smaller than Θ (min(k, r)).
Researcher Affiliation Academia 1Departement of Electrical and Computer Engineering, National Technical University of Athens, Athens, Greece 2Pillar of Engineering Systems and Design, Singapore University of Technology and Design, Singapore.
Pseudocode Yes Algorithm 1 A time-efficient algorithm for solving the dual program of Lemma 2; Algorithm 2 A no-regret algorithm for Fractional Dynamic k-Clustering; Algorithm 3 Deterministic Rounding Scheme; Algorithm 4 A Θ(k)-regret deterministic online learning algorithm for Dynamic k-Clustering; Algorithm 5 A Θ(r)-regret randomized online learning algorithm
Open Source Code No The paper does not provide any explicit statement about releasing source code or a link to a code repository.
Open Datasets No The paper describes client data as 'randomly generated according to a time-varying uniform distribution' and '20 clients arrive according to several static or time-varying two-dimensional probability distributions', implying synthetic data generated by the authors, without providing access information to a pre-existing public dataset.
Dataset Splits No The paper does not provide explicit training, validation, or test dataset splits. It discusses evaluating 'time-average connection cost' over rounds.
Hardware Specification No The paper does not provide any specific details about the hardware used for the experiments, such as CPU/GPU models or memory specifications.
Software Dependencies No The paper does not provide specific software names with version numbers used in the experiments. It mentions 'Python' in relation to examples, but no specific libraries or versions.
Experiment Setup Yes In the following simulations we select ϵ = 0.1 and track the ratio between the time-average cost of Algorithm 4 and of Algorithm 2 which acts as an upper bound on the regret.