patch_size = (8, 8) cs_args = { "train_args":{ "L1_reg": 1e-06, "learning_rate": 0.05, "L2_reg": 1e-05, "nepochs": 1, "cost_type": "crossentropy", "save_exp_data": False, "batch_size": 200, "normalize_weights": False }, "test_args":{ "save_exp_data":False, "batch_size": 200 } } no_of_patches = 64 print "starting pretrain" prmlp = PatchBasedMLP(pre_input, n_in=patch_size[0] * patch_size[1], n_hiddens=[2048, 2048], n_out=11, patch_size=patch_size) incremental_data_experiment(prmlp, train_datasets, test_datasets, no_of_patches=no_of_patches, patch_size=patch_size, **cs_args)
} } print "Starting the cross-validation" x = T.matrix('x') n_hiddens = [2048] train_set_patches, train_set_pre, train_set_labels = ds.Xtrain_patches, ds.Xtrain_presences, ds.Ytrain test_set_patches, test_set_pre, test_set_labels = ds.Xtest_patches, ds.Xtest_presences, ds.Ytest prmlp = PatchBasedMLP(x, n_in=patch_size[0] * patch_size[1], n_hiddens=n_hiddens, n_out=11, no_of_patches=no_of_patches, activation=NeuralActivations.Rectifier, use_adagrad=False, quiet=True) costs, pretrain_probs = prmlp.train(train_set_patches, train_set_pre, **cs_args["train_args"]) save_probs((pretrain_probs, train_set_labels), train_file) print "Testing on the training dataset." fin_test_score, post_test_train_probs = prmlp.test(train_set_patches, train_set_pre, **cs_args["test_args"]) save_probs((post_test_train_probs, train_set_labels), train_test_file)
"batch_size": 200 } } print "Starting the cross-validation" x = T.matrix('x') n_hiddens = [2048] train_set_patches, train_set_pre = ds.Xtrain_patches, ds.Xtrain_presences test_set_patches, test_set_pre = ds.Xtest_patches, ds.Xtest_presences prmlp = PatchBasedMLP(x, n_in=patch_size[0] * patch_size[1], n_hiddens=n_hiddens, n_out=11, no_of_patches=no_of_patches, activation=NeuralActivations.Rectifier, use_adagrad=False, quiet=True) prmlp.train(train_set_patches, train_set_pre, **cs_args["train_args"]) prmlp.save_data() print "Testing on the training dataset." prmlp.test(train_set_patches, train_set_pre, **cs_args["test_args"]) print "Testing on the test dataset." prmlp.test(test_set_patches, test_set_pre, **cs_args["test_args"])
"save_exp_data": False, "batch_size": 200 } } print "Starting the cross-validation" x = T.matrix('x') n_hiddens = [2048] train_set_patches, train_set_pre, train_set_labels = ds.Xtrain_patches, ds.Xtrain_presences, ds.Ytrain test_set_patches, test_set_pre, test_set_labels = ds.Xtest_patches, ds.Xtest_presences, ds.Ytest prmlp = PatchBasedMLP(x, n_in=patch_size[0] * patch_size[1], n_hiddens=n_hiddens, n_out=11, no_of_patches=no_of_patches, activation=NeuralActivations.Rectifier, use_adagrad=False, quiet=True) costs, pretrain_probs = prmlp.train(train_set_patches, train_set_pre, **cs_args["train_args"]) save_probs((pretrain_probs, train_set_labels), train_file) print "Testing on the training dataset." fin_test_score, post_test_train_probs = prmlp.test(train_set_patches, train_set_pre, **cs_args["test_args"]) save_probs((post_test_train_probs, train_set_labels), train_test_file) print "Testing on the test dataset." fin_test_score, post_test_probs = prmlp.test(test_set_patches, test_set_pre, **cs_args["test_args"]) save_probs((post_test_probs, test_set_labels), test_file)
data_path_40k = "/RQusagers/gulcehre/dataset/pentomino/pieces/pento64x64_40k_seed_39112222.npy" data_path = "/RQusagers/gulcehre/dataset/pentomino/experiment_data/pento64x64_80k_seed_39112222.npy" patch_size = (8, 8) ds.setup_pretraining_dataset(data_path=data_path_40k, patch_size=patch_size, normalize_inputs=False) x = T.matrix('x') n_hiddens = [1024, 768] no_of_patches = 64 no_of_classes = 11 prmlp = PatchBasedMLP(x, n_in=patch_size[0] * patch_size[1], n_hiddens=n_hiddens, n_out=11, no_of_patches=no_of_patches, activation=NeuralActivations.Rectifier, use_adagrad=False) params = [prmlp.params[0], prmlp.params[2], prmlp.params[4]] post_mlp = PostMLP(x, n_in=no_of_patches * no_of_classes, n_hiddens=n_hiddens, n_out=1, use_adagrad=False) pre_training(patch_mlp=prmlp, post_mlp=post_mlp, ds=ds)
"batch_size": 200, "normalize_weights": False }, "test_args": { "save_exp_data": False, "batch_size": 200 } } print "Starting the cross-validation" x = T.matrix('x') n_hiddens = [2048] train_set_patches, train_set_pre = ds.Xtrain_patches, ds.Xtrain_presences test_set_patches, test_set_pre = ds.Xtest_patches, ds.Xtest_presences prmlp = PatchBasedMLP(x, n_in=patch_size[0] * patch_size[1], n_hiddens=n_hiddens, n_out=11, no_of_patches=no_of_patches, activation=NeuralActivations.Rectifier, use_adagrad=False, quiet=True) prmlp.train(train_set_patches, train_set_pre, **cs_args["train_args"]) prmlp.save_data() print "Testing on the training dataset." prmlp.test(train_set_patches, train_set_pre, **cs_args["test_args"]) print "Testing on the test dataset." prmlp.test(test_set_patches, test_set_pre, **cs_args["test_args"])
if __name__ == "__main__": print "Task has just started." print "Loading the dataset" ds = Dataset() patch_size = (8, 8) ds_path = \ "/RQusagers/gulcehre/dataset/pentomino/experiment_data/pento64x64_80k_seed_39112222.npy" data_new =\ "/RQusagers/gulcehre/dataset/pentomino/rnd_pieces/pento64x64_5k_seed_43112222_64patches_rnd.npy" data_new_40k =\ "/RQexec/gulcehre/datasets/pentomino/pento_64x64_8x8patches/pento64x64_40k_64patches_seed_975168712_64patches.npy" ds.setup_pretraining_dataset(data_path=data_new_40k, patch_size=patch_size, normalize_inputs=False) pre_input = T.matrix('pre_input') n_hiddens = [2048] prmlp = PatchBasedMLP(pre_input, n_in=8 * 8, n_hiddens=n_hiddens, n_out=11, no_of_patches=64, activation=NeuralActivations.Rectifier, use_adagrad=False) csvm = CSVM() pre_training(prmlp, csvm, ds)