J. Zhang, W. Zhou, X. Chen, W. Yao and L. Cao, "Multisource Selective Transfer Framework in Multiobjective Optimization Problems," in IEEE Transactions on Evolutionary Computation, vol. 24, no. 3, pp. 424-438, June 2020, doi: 10.1109/TEVC.2019.2926107.
的论文学习笔记,只供学习使用,不作商业用途,侵权删除。并且本人学术功底有限如果有思路不正确的地方欢迎批评指正!对于实际工程中的复杂系统设计问题,以卫星系统设计为例,有很多过去的经验,例如在启动新设计之前,已将不同布局的卫星布局解决方案设计存储在数据库中。 利用以往的经验不仅可以提高优化效果,而且可以提高收敛速度。 对于复杂的系统设计而言,这非常重要,因为它通常涉及计算成本高昂的多学科分析[1]。但是,如何利用这些知识来加快新设计的速度,会使卫星设计者感到困惑。 转移学习关注从一个域到另一个域的数据迁移。 各种研究已成功应用于经典机器学习任务,如分类任务[2]-[4],如情绪分析[5],数字识别[6]。近年来,进化算法界的研究人员试图将迁移学习应用到优化任务[7]-[10]中。
三个需要关注的问题
Q1 (可迁移性) :如何识别相似的问题的可迁移性以降低不合适的负迁移。
Q2 (迁移组件) :解决方案,结构,参数等。
Q3 (迁移算法) :最后步骤是如何重用从源问题中提取的信息。 大多数研究关注于一对一的域自适应算法,例如MNIST数据集[13]和WIFI数据集[14]等基准测试中的传输成分分析(TCA)[11],TrAdaBoost [12]。 但是,很少有研究集中在多源迁移学习问题上。
在迁移学习研究领域,大部分工作集中在Q2和Q3,尤其是Q3。 但是,Q1也非常重要,因为从不适当的来源迁移会导致负迁移。 在本文中,我们特别关注多源问题的可迁移性,并研究如何衡量不同优化问题的相似性
[15]证明任务间的相关程度对于多任务学习的有效性十分重要,[16]发展了一个基于自编码器的多任务优化算法,其任务选择主要取决于 1)帕累托前沿解的交叉程度2)任务适应度景观的相似性 ,这些因素保证了相关性。 但是,在实际场景中,我们一开始无法获得目标问题的上述信息。
为了评估不同优化问题的相关性,应首先确定不同优化实例的表示形式。Santana et al. [17] 通过一种估计分布算法-估计贝叶斯网络算法(EBNA)将优化问题视为一个图形。 在他的工作中,任务用图形表示。 可以通过网络分析方法获得任务的相关性,但是表示方法在很大程度上取决于人为的图形特征,可能会丢失任务的相关信息。Yu et al. [18] 发明了一种新的表示强化学习任务的表示方法,该学习方法具有通过均匀分布采样的不同参数。他使用原型策略通过几次迭代获得的奖励向量(称为浅试验)来表示任务,从而使过去的强化学习任务得以重用。 但是,演化过程中的浅层试验会产生太多噪声,无法准确表示优化实例。
在每一个世代,利用估计分布算法
(EDA)表示种群分布,我们可以利用综合分布来表示实例,在这里称为质心分布。质心表示种群在一个世代中的中心。通过计算实例(任务)分布之间的Wasserstein distance(WD)可以获得任务之间的相似性。有了相似性,可以使用三种不同的基于遗传算法的资源选择策略来获得迁移的知识以加速进化过程中的搜索过程,为了高效使用这种策略,采用四种选择机制
以下是本文的贡献:
[1] W. Yao, X. Chen, Q. Ouyang, and M. V. Tooren, “A reliability-based multidisciplinary design optimization procedure based on combined probability and evidence theory,” Struct. Multidiscipl. Optim., vol. 48, no. 2, pp. 339–354, 2013.
[2] C. B. Do and A. Y. Ng, “Transfer learning for text classification,” in Proc. 18th Adv. Neural Inf. Process. Syst., Vancouver, BC, Canada, Dec. 2005, pp. 299–306. [Online]. Available: http://papers.nips.cc/paper/2843-transfer-learning-for-text-classification
[3] A. Quattoni, M. Collins, and T. Darrell, “Transfer learning for image classification with sparse prototype representations,” in Proc. IEEE Comput. Soc. Conf. Comput. Vis. Pattern Recognit. (CVPR), Anchorage, AK, USA, Jun. 2008, pp. 1–8. [Online]. Available: https://doi.org/10.1109/CVPR.2008.4587637
[4] S. J. Pan and Q. Yang, “A survey on transfer learning,” IEEE Trans. Knowl. Data Eng., vol. 22, no. 10, pp. 1345–1359, Oct. 2010. [Online]. Available: https://doi.org/10.1109/TKDE.2009.191
[5] Y. Yoshida, T. Hirao, T. Iwata, M. Nagata, and Y. Matsumoto, “Transfer learning for multiple-domain sentiment analysis— Identifying domain dependent/independent word polarity,” in Proc. 25th AAAI Conf. Artif. Intell., San Francisco, CA, USA, Aug. 2011, pp. 1286–1291. [Online]. Available: http://www.aaai.org/ocs/index.php/AAAI/AAAI11/paper/view/3597
[6] H. Hosseinzadeh and F. Razzazi, “LMDT: A weakly-supervised largemargin- domain-transfer for handwritten digit recognition,” Eng. Appl. Artif. Intell., vol. 52, pp. 119–125, Jun. 2016. [Online]. Available: https://doi.org/10.1016/j.engappai.2016.02.014
[7] L. Feng, Y.-S. Ong, S. Jiang, and A. Gupta, “Autoencoding evolutionary search with learning across heterogeneous problems,” IEEE Trans. Evol. Comput., vol. 21, no. 5, pp. 760–772, Oct. 2017. [Online]. Available: https://doi.org/10.1109/TEVC.2017.2682274
[8] M. Jiang, Z. Huang, L. Qiu, W. Huang, and G. G. Yen, “Transfer learning-based dynamic multiobjective optimization algorithms,” IEEE Trans. Evol. Comput., vol. 22, no. 4, pp. 501–514, Aug. 2018. [Online]. Available: https://doi.org/10.1109/TEVC.2017.2771451
[9] M. Jiang, L. Qiu, Z. Huang, and G. G. Yen, “Dynamic multi-objective estimation of distribution algorithm based on domain adaptation and nonparametric estimation,” Inf. Sci., vol. 435, pp. 203–223, Apr. 2018. [Online]. Available: https://doi.org/10.1016/j.ins.2017.12.058
[10] R. Santana, A. Mendiburu, and J. A. Lozano, “Structural transfer using EDAs: An application to multi-marker tagging SNP selection,” in Proc. IEEE Congr. Evol. Comput. (CEC), Brisbane, QLD, Australia, Jun. 2012, pp. 1–8. [Online]. Available: https://doi.org/10.1109/CEC.2012.6252963
[11] S. J. Pan, I. W. Tsang, J. T. Kwok, and Q. Yang, “Domain adaptation via transfer component analysis,” IEEE Trans. Neural Netw., vol. 22, no. 2, pp. 199–210, Feb. 2011. [Online]. Available: https://doi.org/10.1109/TNN.2010.2091281
[12] W. Dai, Q. Yang, G.-R. Xue, and Y. Yu, “Boosting for transfer learning,” in Proc. 24th Int. Conf. Mach. Learn. (ICML), Corvallis, OR, USA, Jun. 2007, pp. 193–200. [Online]. Available: https://doi.org/10.1145/1273496.1273521
[13] Y. Lecun, L. Bottou, Y. Bengio, and P. Haffner, “Gradient-based learning applied to document recognition,” Proc. IEEE, vol. 86, no. 11, pp. 2278–2324, Nov. 1998.
[14] J. G. Rohra, B. Perumal, S. J. Narayanan, P. Thakur, and R. B. Bhatt, “User localization in an indoor environment using fuzzy hybrid of particle swarm optimization & gravitational search algorithm with neural networks,” in Proc. 6th Int. Conf. Soft Comput. Problem Solving, 2017, pp. 286–295.
[15] T. Liu, Q. Yang, and D. Tao, “Understanding how feature structure transfers in transfer learning,” in Proc. 26th Int. Joint Conf. Artif. Intell. (IJCAI), Melbourne, VIC, Australia, Aug. 2017, pp. 2365–2371. [Online]. Available: https://doi.org/10.24963/ijcai.2017/329
[16] L. Feng et al., “Evolutionary multitasking via explicit autoencoding,” IEEE Trans. Cybern., vol. 49, no. 9, pp. 3457–3470, Sep. 2019.
[17] R. Santana, C. Bielza, and P. Larranaga, “Network measures for re-using problem information in EDAs,” Dept. Artif. Intell., Faculty Informat., Tech. Univ. Madrid, Madrid, Spain, Rep. UPM-FI/DIA/2010–3, 2010.
[18] Y. Yu, S.-Y. Chen, Q. Da, and Z.-H. Zhou, “Reusable reinforcement learning via shallow trails,” IEEE Trans. Neural Netw. Learn. Syst., vol. 29, no. 6, pp. 2204–2215, Jun. 2018. [Online]. Available: https://doi.org/10.1109/TNNLS.2018.2803729
[19] K. Deb, A. Pratap, S. Agarwal, and T. Meyarivan, “A fast and elitist multiobjective genetic algorithm: NSGA-II,” IEEE Trans. Evol. Comput., vol. 6, no. 2, pp. 182–197, Apr. 2002.
[20] A. Gupta, Y.-S. Ong, and L. Feng, “Insights on transfer optimization: Because experience is the best teacher,” IEEE Trans. Emerg. Topics Comput. Intell., vol. 2, no. 1, pp. 51–64, Feb. 2018. [Online]. Available: https://doi.org/10.1109/TETCI.2017.2769104
[21] M. Hauschild and M. Pelikan, “An introduction and survey of estimation of distribution algorithms,” Swarm Evol. Comput., vol. 1, no. 3, pp. 111–128, 2011. [Online]. Available: https://doi.org/10.1016/j.swevo.2011.08.003
[22] S. Baluja, “Population-based incremental learning. A method for integrating genetic search based function optimization and competitive learning,” Tech. Rep., 1994. [Online]. Available: https://dl.acm.org/ citation.cfm?id=865123
[23] H. Mühlenbein, “The equation for response to selection and its use for prediction,” Evol. Comput., vol. 5, no. 3, pp. 303–346, 1997. [Online]. Available: https://doi.org/10.1162/evco.1997.5.3.303
[24] G. R. Harik, F. G. Lobo, and D. E. Goldberg, “The compact genetic algorithm,” IEEE Trans. Evol. Comput., vol. 3, no. 4, pp. 287–297, Nov. 1999.
[25] J. S. D. Bonet, C. L. Isbell, and P. Viola, “MIMIC: Finding optima by estimating probability densities,” in Proc. Int. Conf. Neural Inf. Process. Syst., Denver, CO, USA, 1996, pp. 424–430.
[26] M. Pelikan and H. Muehlenbein, The Bivariate Marginal Distribution Algorithm. London, U.K.: Springer, 1999.
[27] M. Pelikan, D. E. Goldberg, and E. Cantú-Paz, “BOA: The Bayesian optimization algorithm,” in Proc. Conf. Genet. Evol. Comput., Orlando, FL, USA, 1999, pp. 525–532.
[28] S. Shakya, J. Mccall, and D. Brown, “Using a Markov network model in a univariate EDA: An empirical cost-benefit analysis,” in Proc. Genet. Evol. Comput. Conf. (GECCO), Washington, DC, USA, Jun. 2005, pp. 727–734.
[29] Y. Gao, L. Peng, F. Li, M. Liu, and X. Hu, “EDA-based multi-objective optimization using preference order ranking and multivariate Gaussian copula,” in Proc. 10th Int. Symp. Neural Netw. Adv. Neural Netw. II (ISNN), Dalian, China, Jul. 2013, pp. 341–350. [Online]. Available: https://doi.org/10.1007/978-3-642-39068-5_42
[30] H. Karshenas, R. Santana, C. Bielza, and P. Larrañaga, “Multiobjective estimation of distribution algorithm based on joint modeling of objectives and variables,” IEEE Trans. Evol. Comput., vol. 18, no. 4, pp. 519–542, Aug. 2014. [Online]. Available: https://doi.org/10.1109/TEVC.2013.2281524
[31] V. A. Shim, K. C. Tan, and C. Y. Cheong, “A hybrid estimation of distribution algorithm with decomposition for solving the multiobjective multiple traveling salesman problem,” IEEE Trans. Syst., Man, Cybern. C, Appl. Rev., vol. 42, no. 5, pp. 682–691, Sep. 2012. [Online]. Available: https://doi.org/10.1109/TSMCC.2012.2188285
[32] V. A. Shim, K. C. Tan, J. Y. Chia, and A. A. Mamun, “Multi-objective optimization with estimation of distribution algorithm in a noisy environment,” Evol. Comput., vol. 21, no. 1, pp. 149–177, Mar. 2013. [Online]. Available: https://doi.org/10.1162/EVCO_a_00066
[33] J. Luo, Y. Qi, J. Xie, and X. Zhang, “A hybrid multi-objective PSO–EDA algorithm for reservoir flood control operation,” Appl. Soft Comput., vol. 34, pp. 526–538, Sep. 2015. [Online]. Available: https://doi.org/10.1016/j.asoc.2015.05.036
[34] P. Lipinski, “Solving the firefighter problem with two elements using a multi-modal estimation of distribution algorithm,” in Proc. IEEE Congr. Evol. Comput. (CEC), Donostia-San Sebastián, Spain, Jun. 2017, pp. 2161–2168. [Online]. Available: https://doi.org/10.1109/CEC.2017.7969566
[35] Q. Yang, W.-N. Chen, Y. Li, C. L. P. Chen, X.-M. Xu, and J. Zhang, “Multimodal estimation of distribution algorithms,” IEEE Trans. Cybern., vol. 47, no. 3, pp. 636–650, Mar. 2017. [Online]. Available: https://doi.org/10.1109/TCYB.2016.2523000
[36] M. Pelikan, M. W. Hauschild, and P. L. Lanzi, “Transfer learning, soft distance-based bias, and the hierarchical BOA,” in Proc. 12th Int. Conf. Parallel Problem Solving Nat. I (PPSN XII), Taormina, Italy, Sep. 2012, pp. 173–183. [Online]. Available: https://doi.org/10.1007/978-3-642- 32937-1_18
[37] S. Jiang and S. Yang, “A steady-state and generational evolutionary algorithm for dynamic multiobjective optimization,” IEEE Trans. Evol. Comput., vol. 21, no. 1, pp. 65–82, Feb. 2017.
[38] S. Jiang and S. Yang, “Evolutionary dynamic multiobjective optimization: Benchmarks and algorithm comparisons,” IEEE Trans. Cybern., vol. 47, no. 1, pp. 198–211, Jan. 2017.
[39] Q. Li, J. Zou, S. Yang, J. Zheng, and R. Gan, “A predictive strategy based on special points for evolutionary dynamic multi-objective optimization,” Soft Comput., vol. 23, no. 11, pp. 3723–3739, 2019.
[40] G. Ruan, G. Yu, J. Zheng, J. Zou, and S. Yang, “The effect of diversity maintenance on prediction in dynamic multi-objective optimization,” Appl. Soft Comput., vol. 58, pp. 631–647, Sep. 2017. [Online]. Available: https://doi.org/10.1016/j.asoc.2017.05.008
[41] J. Zou, Q. Li, S. Yang, H. Bai, and J. Zheng, “A prediction strategy based on center points and knee points for evolutionary dynamic multi-objective optimization,” Appl. Soft Comput., vol. 61, pp. 806–818, Dec. 2017. [Online]. Available: https://doi.org/10.1016/j.asoc.2017.08.004
[42] A. Zhou, Y. Jin, Q. Zhang, B. Sendhoff, and E. P. K. Tsang, “Prediction-based population re-initialization for evolutionary dynamic multi-objective optimization,” in Proc. 4th Int. Conf. Evol. Multi Criterion Optim. (EMO), Matsushima, Japan, Mar. 2007, pp. 832–846. [Online]. Available: https://doi.org/10.1007/978-3-540-70928-2_62
[43] J. Branke, “Memory enhanced evolutionary algorithms for changing optimization problems,” in Proc. Congr. Evol. Comput. (CEC), vol. 3. Washington, DC, USA, 1999, pp. 1875–1882.
[44] C. K. Goh and K. C. Tan, “A competitive-cooperative coevolutionary paradigm for dynamic multiobjective optimization,” IEEE Trans. Evol. Comput., vol. 13, no. 1, pp. 103–127, Feb. 2009.
[45] I. Hatzakis and D. Wallace, “Dynamic multi-objective optimization with evolutionary algorithms: A forward-looking approach,” in Proc. 8th Annu. Conf. Genet. Evol. Comput., Seattle, WA, USA, 2006, pp. 1201–1208.
[46] Z. Zhang and S. Qian, “Artificial immune system in dynamic environments solving time-varying non-linear constrained multi-objective problems,” Soft Comput., vol. 15, no. 7, pp. 1333–1349, 2011.
[47] L. T. Bui, Z. Michalewicz, E. Parkinson, and M. B. Abello, “Adaptation in dynamic environments: A case study in mission planning,” IEEE Trans. Evol. Comput., vol. 16, no. 2, pp. 190–209, Apr. 2012.
[48] R. Shang, L. Jiao, Y. Ren, L. Li, and L. Wang, “Quantum immune clonal coevolutionary algorithm for dynamic multiobjective optimization,” Soft Comput., vol. 18, no. 4, pp. 743–756, 2014.
[49] Y. Netzer, T. Wang, A. Coates, A. Bissacco, B. Wu, and A. Y. Ng, “Reading digits in natural images with unsupervised feature learning,” 2011. [Online]. Available: https://ai.google/research/pubs/pub37648
[50] I. Olkin and F. Pukelsheim, “The distance between two random vectors with given dispersion matrices,” Linear Algebra Appl., vol. 48, no. 82, pp. 257–263, Dec. 1982.