- author = "Colin G. Johnson",
- title = "Solving the {Rubik}'s cube with stepwise deep learning",
- journal = "Expert Systems: The Journal of Knowledge Engineering",
- year = "2021",
- volume = "38",
- number = "3",
- month = may,
- keywords = "genetic algorithms, genetic programming, artificial intelligence, evolutionary computation, Learned Guidance Functions, LGF, fitness functions, human-like AI, loss functions, ANN, evolution strategies",
- ISSN = "0266-4720",
-
DOI = "
10.1111/exsy.12665",
- timestamp = "Tue, 01 Jun 2021 01:00:00 +0200",
-
biburl = "
https://dblp.org/rec/journals/es/Johnson21.bib",
- bibsource = "dblp computer science bibliography, https://dblp.org",
- size = "13 pages",
- abstract = "explores a novel technique for learning the fitness function for search algorithms such as evolutionary strategies and hill climbing. The aim of the new technique is to learn a fitness function (called a Learned Guidance Function) from a set of sample solutions to the problem. These functions are learned using a supervised learning approach based on deep neural network learning, that is, neural networks with a number of hidden layers. This is applied to a test problem: unscrambling the Rubik's Cube using evolutionary and hillclimbing algorithms. Comparisons are made with a previous LGF approach based on random forests, with a baseline approach based on traditional error-based fitness, and with other approaches in the literature. This demonstrates how a fitness function can be learned from existing solutions, rather than being provided by the user, increasing the autonomy of AI search processes.",
-
notes = "Keras on TensorFlow, Adam optimizer, Keras deep
learning network, dense 9x6 relu, dropout. 50 epochs.
Deep LGF + ES
p6 'fitness landscape created by a real LGF will still have local minima.'
p8 'percentage of successes in the Deep LGF + ES approach is considerably higher than the Random Forest LGF + Hillclimbing'
p11 'learning does not generalize from the anticlockwise to the clockwise.'
speculation: program code manipulated (GP). interpretable
School of Computer Science at the University of Nottingham",
