diff options
| -rw-r--r-- | .gitignore | 1 | ||||
| -rw-r--r-- | config/simple_mlp_tgtcls_0.py | 25 | ||||
| -rw-r--r-- | model/simple_mlp_tgtcls.py | 74 | ||||
| -rw-r--r-- | train.py | 7 |
4 files changed, 104 insertions, 3 deletions
@@ -66,3 +66,4 @@ target/ # saved params taxi_model/* +model_data/* diff --git a/config/simple_mlp_tgtcls_0.py b/config/simple_mlp_tgtcls_0.py new file mode 100644 index 0000000..9ccfbfd --- /dev/null +++ b/config/simple_mlp_tgtcls_0.py @@ -0,0 +1,25 @@ +import cPickle + +import data + +import model.simple_mlp_tgtcls as model + +n_dow = 7 # number of division for dayofweek/dayofmonth/hourofday +n_dom = 31 +n_hour = 24 + +n_begin_end_pts = 5 # how many points we consider at the beginning and end of the known trajectory +n_end_pts = 5 + +n_valid = 1000 + +with open(data.DATA_PATH + "/arrival-clusters.pkl") as f: tgtcls = cPickle.load(f) + +dim_embed = 10 +dim_input = n_begin_end_pts * 2 * 2 + dim_embed + dim_embed +dim_hidden = [200] +dim_output = tgtcls.shape[0] + +learning_rate = 0.0001 +momentum = 0.99 +batch_size = 32 diff --git a/model/simple_mlp_tgtcls.py b/model/simple_mlp_tgtcls.py new file mode 100644 index 0000000..99a459a --- /dev/null +++ b/model/simple_mlp_tgtcls.py @@ -0,0 +1,74 @@ +import numpy + +import theano +from theano import tensor + +from blocks.bricks import MLP, Rectifier, Linear, Sigmoid, Identity, Softmax +from blocks.bricks.lookup import LookupTable + +from blocks.initialization import IsotropicGaussian, Constant + +import data +import hdist + +class Model(object): + def __init__(self, config): + # The input and the targets + x_firstk_latitude = (tensor.matrix('first_k_latitude') - data.porto_center[0]) / data.data_std[0] + x_firstk_longitude = (tensor.matrix('first_k_longitude') - data.porto_center[1]) / data.data_std[1] + + x_lastk_latitude = (tensor.matrix('last_k_latitude') - data.porto_center[0]) / data.data_std[0] + x_lastk_longitude = (tensor.matrix('last_k_longitude') - data.porto_center[1]) / data.data_std[1] + + x_client = tensor.lvector('origin_call') + x_stand = tensor.lvector('origin_stand') + + y = tensor.concatenate((tensor.vector('destination_latitude')[:, None], + tensor.vector('destination_longitude')[:, None]), axis=1) + + # Define the model + client_embed_table = LookupTable(length=data.n_train_clients+1, dim=config.dim_embed, name='client_lookup') + stand_embed_table = LookupTable(length=data.n_stands+1, dim=config.dim_embed, name='stand_lookup') + mlp = MLP(activations=[Rectifier() for _ in config.dim_hidden] + [Softmax()], + dims=[config.dim_input] + config.dim_hidden + [config.dim_output]) + classes = theano.shared(numpy.array(config.tgtcls, dtype=theano.config.floatX), name='classes') + + # Create the Theano variables + client_embed = client_embed_table.apply(x_client) + stand_embed = stand_embed_table.apply(x_stand) + inputs = tensor.concatenate([x_firstk_latitude, x_firstk_longitude, + x_lastk_latitude, x_lastk_longitude, + client_embed, stand_embed], + axis=1) + # inputs = theano.printing.Print("inputs")(inputs) + cls_probas = mlp.apply(inputs) + outputs = tensor.dot(cls_probas, classes) + + # Normalize & Center + # outputs = theano.printing.Print("normal_outputs")(outputs) + outputs = data.data_std * outputs + data.porto_center + + # outputs = theano.printing.Print("outputs")(outputs) + # y = theano.printing.Print("y")(y) + + outputs.name = 'outputs' + + # Calculate the cost + cost = hdist.erdist(outputs, y).mean() + cost.name = 'cost' + hcost = hdist.hdist(outputs, y).mean() + hcost.name = 'hcost' + + # Initialization + client_embed_table.weights_init = IsotropicGaussian(0.001) + stand_embed_table.weights_init = IsotropicGaussian(0.001) + mlp.weights_init = IsotropicGaussian(0.01) + mlp.biases_init = Constant(0.001) + + client_embed_table.initialize() + stand_embed_table.initialize() + mlp.initialize() + + self.cost = cost + self.hcost = hcost + self.outputs = outputs @@ -38,7 +38,8 @@ if __name__ == "__main__": if len(sys.argv) != 2: print >> sys.stderr, 'Usage: %s config' % sys.argv[0] sys.exit(1) - config = importlib.import_module(sys.argv[1]) + model_name = sys.argv[1] + config = importlib.import_module(model_name) def setup_train_stream(): @@ -107,8 +108,8 @@ def main(): every_n_batches=1000), Printing(every_n_batches=1000), # Checkpoint('model.pkl', every_n_batches=100), - Dump('taxi_model', every_n_batches=1000), - LoadFromDump('taxi_model'), + Dump('model_data/' + model_name, every_n_batches=1000), + LoadFromDump('model_data/' + model_name), FinishAfter(after_epoch=5) ] |