XCS with Adaptive Action Mapping

Nakata, Masaya; Lanzi, Pier Luca; Takadama, Keiki

doi:10.1007/978-3-642-34859-4_14

Cited by 14 publications

(9 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This is an unexpected result against our previous experiment on the benchmark classification task [14,15]. This paper is motivated to understand why XCSAM sometimes fails to outperform XCS.…”

Section: Special Issue On Research On Analyzing Simulating and Regumentioning

confidence: 63%

“…The maximum iteration is 500,000. lem with fewer iterations than XCS; the classification accuracy of XCS reaches to 100% after 180,000 iterations; while XCSAM reaches optimally after the 100,000 iterations. This is an expected fact that our previous works showed XCSAM solves a simple classification task with fewer iterations than that XCS takes [14,15]. However, when the state-action space is further increasing (i.e., 4×3 CMP), the classification accuracy of XCSAM unexpectedly degrades to about 45%; while XCS outperforms XCSAM with the higher classification accuracy about 62%.…”

Section: Experimental Designmentioning

confidence: 73%

“…7, pp. 273-285, 2017 273 duced an extension of XCS called the XCSAM classifier system [14]. Different from XCS, XCSAM employs a learning strategy of best action map which only covers the highest-return action at each state (see Fig.1a).…”

Section: Special Issue On Research On Analyzing Simulating and Regumentioning

confidence: 99%

“…Different from XCS, XCSAM employs a learning strategy of best action map which only covers the highest-return action at each state (see Fig.1a). We showed XCSAM outperforms XCS in a simple benchmark classification task with binary class (the multiplexer problem [20]), while successfully producing the maximally accurate and maximally general rules as well as XCS [14,15].…”

Section: Special Issue On Research On Analyzing Simulating and Regumentioning

confidence: 99%

“…XCSAM generates maxP (st,a) ideally converges to the highest reward at st, the best action can be identified by comparing the given reward with maxP (st,a) [14]. Fig.…”

Section: Xcsam Classifier Systemmentioning

confidence: 99%

See 4 more Smart Citations

Revisit of Rule-Deletion Strategy for XCSAM Classifier System on Classification

Nakata

Hamagami

2017

Transactions of ISCIE

Self Cite

View full text Add to dashboard Cite

The XCSAM classifier system is an approach of evolutionary rule-based machine learning, which evolves rules advocating the highest-return actions at state, resulting in best classification. This paper starts with claiming a limitation that XCSAM still fails to evolutionary generate adequate rules advocating the highest-return actions. Then, under our hypothesis that this limitation is caused from the rule-deletion mechanism of XCSAM, we revisit its rule-deletion strategy in order to promote the evolution of adequate rules. Different from the existing deletion strategy which deletes two rules for each rule-evolution, our deletion strategy is modified to delete more than two rules as necessary. Experimental results on a benchmark classification task validate our modification powerfully works to evolve adequate rules, improving the performance of XCSAM. We further show our modification robustly enables XCSAM to perform well on real world classification task.

show abstract

Section: Special Issue On Research On Analyzing Simulating and Regumentioning

confidence: 63%

Section: Experimental Designmentioning

confidence: 73%

Section: Special Issue On Research On Analyzing Simulating and Regumentioning

confidence: 99%

Section: Special Issue On Research On Analyzing Simulating and Regumentioning

confidence: 99%

“…XCSAM generates maxP (st,a) ideally converges to the highest reward at st, the best action can be identified by comparing the given reward with maxP (st,a) [14]. Fig.…”

Section: Xcsam Classifier Systemmentioning

confidence: 99%

See 3 more Smart Citations