Re-purposing Heterogeneous Generative Ensembles with Evolutionary Computation

被引:11
作者
Toutouh, Jamal [1 ]
Hemberg, Erik [1 ]
O'Reily, Una-May [1 ]
机构
[1] MIT, CSAIL, Cambridge, MA 02139 USA
来源
GECCO'20: PROCEEDINGS OF THE 2020 GENETIC AND EVOLUTIONARY COMPUTATION CONFERENCE | 2020年
基金
欧盟地平线“2020”;
关键词
Generative adversarial networks; ensembles; genetic algorithms; diversity; NEURAL-NETWORKS; REGRESSION;
D O I
10.1145/3377930.3390229
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Generative Adversarial Networks (GANs) are popular tools for generative modeling. The dynamics of their adversarial learning give rise to convergence pathologies during training such as mode and discriminator collapse. In machine learning, ensembles of predictors demonstrate better results than a single predictor for many tasks. In this study, we apply two evolutionary algorithms (EAs) to create ensembles to re-purpose generative models, i.e., given a set of heterogeneous generators that were optimized for one objective (e.g., minimize Frechet Inception Distance), create ensembles of them for optimizing a different objective (e.g., maximize the diversity of the generated samples). The first method is restricted by the exact size of the ensemble and the second method only restricts the upper bound of the ensemble size. Experimental analysis on the MNIST image benchmark demonstrates that both EA ensembles creation methods can re-purpose the models, without reducing their original functionality. The EA-based demonstrate significantly better performance compared to other heuristic-based methods. When comparing both evolutionary, the one with only an upper size bound on the ensemble size is the best.
引用
收藏
页码:425 / 434
页数:10
相关论文
共 64 条
  • [1] Abbass HA, 2003, IEEE C EVOL COMPUTAT, P2074
  • [2] A novel SVM-kNN-PSO ensemble method for intrusion detection system
    Aburomman, Abdulla Amin
    Reaz, Mamun Bin Ibne
    [J]. APPLIED SOFT COMPUTING, 2016, 38 : 360 - 372
  • [3] A survey on evolutionary machine learning
    Al-Sahaf, Harith
    Bi, Ying
    Chen, Qi
    Lensen, Andrew
    Mei, Yi
    Sun, Yanan
    Tran, Binh
    Xue, Bing
    Zhang, Mengjie
    [J]. JOURNAL OF THE ROYAL SOCIETY OF NEW ZEALAND, 2019, 49 (02) : 205 - 228
  • [4] Alexandropoulos StamatiosAggelos N., 2019, MULTIOBJECTIVE EVOLU, P35
  • [5] [Anonymous], 2017, P 31 INT C NEURAL IN, DOI DOI 10.5555/3295222.3295408
  • [6] [Anonymous], 2013, THESIS
  • [7] [Anonymous], 2017, ARXIV170609549
  • [8] [Anonymous], 2016, ARXIV161200991
  • [9] [Anonymous], 2017, ARXIV PREPRINT ARXIV
  • [10] [Anonymous], ADV NEURAL INFORM PR