3 ms·
For this simple example just choosing the largest possible fully-connected network with ReLU and L2 regularization to prevent overfit quickly converges to a nic
by moconnor 10y ago
For this simple example just choosing the largest possible fully-connected network with ReLU and L2 regularization to prevent overfit quickly converges to a nice spiral (test loss of 0.001 for me):
http://playground.tensorflow.org/#activation=relu®ularization=L2&batchSize=30&dataset=spiral®Dataset=reg-plane&learningRate=0.03®ularizationRate=0.003&noise=0&networkShape=8,8,8,8,8,8&seed=0.38911&showTestData=false&discretize=false&percTrainData=50&x=true&y=true&xTimesY=true&xSquared=true&ySquared=true&cosX=false&sinX=true&cosY=false&sinY=true&collectStats=false&problem=classification http://playground.tensorflow.org/#activation=relu®ulariza...
- amelius 10y agoNeat. How would the number of neurons N scale with the size of the spiral? (Size=number of turns) Will N level off, meaning that it will really understand the structure of the spiral?
- findthewords 10y agoI wouldn't call it quick... spiral in 150 iterations, with sigmoid magic: http://playground.tensorflow.org/#activation=sigmoid®ularization=L2&batchSize=10&dataset=spiral®Dataset=reg-plane&learningRate=0.1®ularizationRate=0.01&noise=0&networkShape=8&seed=0.63555&showTestData=false&discretize=false&percTrainData=50&x=true&y=true&xTimesY=true&xSquared=true&ySquared=true&cosX=false&sinX=true&cosY=false&sinY=true&collectStats=false&problem=classification http://playground.tensorflow.org/#activation=sigmoid®ular... I find the pulsating unsightly.