TY - CHAP
T1 - An analysis of Lamarckian learning in changing environments
AU - Curran, Dara
AU - O'Sullivan, Barry
PY - 2011
Y1 - 2011
N2 - It is widely recognised that many species adapt to complex and dynamic environments, but it is no longer accepted that an organism passes characteristics acquired during its lifetime to its offspring. However, in evolutionary computation such Lamarckian inheritance can be useful. Simulations of the benefits of Lamarckian inheritance have been reported in the literature. However, in this paper we present the first formal proof that Lamarckian inheritance can dominate more traditional individual (non-inheritable) learning. We present a parameterised model that can demonstrate conditions in which different inheritance types perform best, which we empirically validate.
AB - It is widely recognised that many species adapt to complex and dynamic environments, but it is no longer accepted that an organism passes characteristics acquired during its lifetime to its offspring. However, in evolutionary computation such Lamarckian inheritance can be useful. Simulations of the benefits of Lamarckian inheritance have been reported in the literature. However, in this paper we present the first formal proof that Lamarckian inheritance can dominate more traditional individual (non-inheritable) learning. We present a parameterised model that can demonstrate conditions in which different inheritance types perform best, which we empirically validate.
UR - https://www.scopus.com/pages/publications/79959947656
U2 - 10.1007/978-3-642-21314-4_18
DO - 10.1007/978-3-642-21314-4_18
M3 - Chapter
AN - SCOPUS:79959947656
SN - 9783642213137
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 142
EP - 149
BT - Advances in Artificial Life
T2 - 10th European Conference of Artificial Life, ECAL 2009
Y2 - 13 September 2009 through 16 September 2009
ER -