Adversarial Task Assignment

Authors: Chen Hajaj, Yevgeniy Vorobeychik

IJCAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We now experimentally demonstrate the effectiveness of our proposed approaches. Workers proficiencies are sampled using two distributions: a uniform distribution over the [0.5, 1] interval and an exponential distribution with µ = 0.25 where proficiencies are truncated to be in this interval for the latter. We compare our adversarial assignment algorithms to three natural baselines: Split-k and two versions of Monte Carlo (involving random assignment of tasks to workers).
Researcher Affiliation Academia Chen Hajaj and Yevgeniy Vorobeychik Electrical Engineering and Computer Science Vanderbilt University, Nashville, Tennses {chen.hajaj, yevgeniy.vorobeychik}@vanderbilt.edu
Pseudocode Yes Algorithm 1 Homogeneous assignment; Algorithm 2 Heterogeneous assignment
Open Source Code No The paper does not provide an explicit statement or link to the source code for the described methodology.
Open Datasets No Workers proficiencies are sampled using two distributions: a uniform distribution over the [0.5, 1] interval and an exponential distribution with µ = 0.25 where proficiencies are truncated to be in this interval for the latter. The paper generates data based on distributions rather than using a publicly available dataset with concrete access information.
Dataset Splits No The paper mentions '5,000 sample runs' and '3,000 runs' for experiments, but it does not specify train/validation/test dataset splits or explicit use of a validation set.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory) used for running its experiments.
Software Dependencies Yes We used CPLEX version 12.51 to solve the integer linear program above.
Experiment Setup Yes Workers proficiencies are sampled using two distributions: a uniform distribution over the [0.5, 1] interval and an exponential distribution with µ = 0.25 where proficiencies are truncated to be in this interval for the latter." and "We use similar baseline methods to the ones used in studying homogeneous task assignment." and "In these experimets we use a natural weighted majority decision rule with θw = pw (i.e., workers proficiencies), and set K = 2500.