Content deleted Content added
Citation bot (talk | contribs) Altered volume. | Use this bot. Report bugs. | Suggested by Headbomb | #UCB_toolbar |
Rescuing 1 sources and tagging 0 as dead.) #IABot (v2.0.9.5) (LeapTorchGear - 17897 |
||
Line 26:
== Multi-objective search ==
While most approaches solely focus on finding architecture with maximal predictive performance, for most practical applications other objectives are relevant, such as memory consumption, model size or inference time (i.e., the time required to obtain a prediction). Because of that, researchers created a [[Multi-objective optimization|multi-objective]] search.<ref name="Elsken 2018">{{cite arXiv|last1=Elsken|first1=Thomas|last2=Metzen|first2=Jan Hendrik|last3=Hutter|first3=Frank|date=2018-04-24|title=Efficient Multi-objective Neural Architecture Search via Lamarckian Evolution|eprint=1804.09081|class=stat.ML}}</ref><ref name="Zhou 2018">{{cite web|url=https://www.sysml.cc/doc/2018/94.pdf|title=Neural Architect: A Multi-objective Neural Architecture Search with Performance Prediction|last1=Zhou|first1=Yanqi|last2=Diamos|first2=Gregory|date=|website=|publisher=Baidu|access-date=2019-09-27|archive-date=2019-09-27|archive-url=https://web.archive.org/web/20190927090457/https://www.sysml.cc/doc/2018/94.pdf|url-status=dead}}</ref>
LEMONADE<ref name="Elsken 2018" /> is an evolutionary algorithm that adopted [[Lamarckism]] to efficiently optimize multiple objectives. In every generation, child networks are generated to improve the [[Pareto efficiency#Pareto frontier|Pareto frontier]] with respect to the current population of ANNs.
| |||