Human Assisted Learning by Evolutionary Multi-Objective Optimization

Authors: Dan-Xuan Liu, Xin Mu, Chao Qian

AAAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments on the tasks of medical diagnosis and content moderation show the superiority of HAL-EMO (with either NSGA-II, GSEMO or BSEMO) over previous algorithms, and that using BSEMO leads to the best performance of HAL-EMO.
Researcher Affiliation Academia Dan-Xuan Liu1, Xin Mu2, Chao Qian1 1State Key Laboratory for Novel Software Technology, Nanjing University, Nanjing 210023, China 2Peng Cheng Laboratory, Shenzhen 518000, China {liudx, qianc}@lamda.nju.edu.cn, mux@pcl.ac.cn
Pseudocode Yes Algorithm 1: HAL-EMO Framework
Open Source Code No The paper does not provide any explicit statement or link indicating that the source code for the described methodology is publicly available.
Open Datasets Yes The three real-world data sets Messidor (Decenci ere et al. 2014), Stare-H and Stare D (Hoover, Kouznetsova, and Goldbaum 2000) are used for medical diagnosis, all containing about 400 eye images. The data Hatespeech (Davidson et al. 2017) for content moderation contains about 25000 tweets... Besides Messidor, another data set Aptos containing 705 retinal images is used;
Dataset Splits Yes We use 80% of the instances for training and the rest for testing... We use 60% of the instances for training and the rest for testing.
Hardware Specification No The paper does not explicitly mention any specific hardware details such as GPU models, CPU models, or memory specifications used for running the experiments.
Software Dependencies No The paper does not provide specific version numbers for any software libraries, frameworks, or programming languages used in the experiments.
Experiment Setup Yes As HAL-EMO is an anytime algorithm, we set the number of objective evaluations to 40kn, to make a tradeoff between the performance and runtime... The regularization parameter λ is set to 1 for Messidor and Stare-D, 0.5 for Stare-H and 0.01 for Hatespeech... The population size is set to 100; the initial population consists of the all-0s vector 0 and 99 randomly generated solutions; onepoint crossover is performed in each iteration with probability 0.9.