Submodular Meta-Learning

Authors: Arman Adibi, Aryan Mokhtari, Hamed Hassani

NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We provide two experimental setups to evaluate the performance of our proposed algorithms and compare with other baselines. Each setup involves a different set of tasks which are represented as submodular maximization problems subject to the k-cardinality constraint.
Researcher Affiliation Academia Arman Adibi ESE Department University of Pennsylvania Philadelphia, PA 19104 aadibi@seas.upenn.edu; Aryan Mokhtari ECE Department University of Texas at Austin Austin, TX 78712 mokhtari@austin.utexas.edu; Hamed Hassani ESE Department University of Pennsylvania Philadelphia, PA 19104 hassani@seas.upenn.edu
Pseudocode Yes Algorithm 1; Algorithm 2; Algorithm 3 Meta-Greedy; Algorithm 4 Randomized meta-Greedy
Open Source Code No The paper does not contain any explicit statement or link indicating that the source code for the described methodology is publicly available.
Open Datasets Yes We will formalize and solve a facility location problem on the Uber dataset [49]. Our experiments were run on the portion of data corresponding to Uber pick-ups in Manhattan in the period of September 2014. This portion consists of 106 data points each represented as a triplet (latitude, longitude, Date Time)... We use the Movielens dataset [50] which consists of 106 ratings (from 1 to 5) by 6041 users for 4000 movies.
Dataset Splits No The paper describes partitioning users into training and test phases, e.g., 'We partitioned the 200 users into 100 users for the training phase and 100 other users for the test phase.' However, it does not explicitly mention a separate validation set or provide specific split percentages/counts for training, validation, and test sets that would allow full reproduction of data partitioning.
Hardware Specification No The paper does not provide any specific details about the hardware used to run the experiments (e.g., CPU, GPU models, or memory specifications).
Software Dependencies No The paper does not list any specific software dependencies with version numbers.
Experiment Setup No The paper provides details on varying k and l values for the experiments (e.g., 'k = 20, and vary l from 5 to 18', 'k changes from 5 to 30, and l is 80% of k'), but it does not specify common experimental setup details like hyperparameter values (e.g., learning rate, batch size, optimizer settings) or model initialization.