From 98139f573eb179c8f5a06ba6c8d8883376814ccf Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=C3=89tienne=20Simon?= <esimon@esimon.eu>
Date: Thu, 2 Jul 2015 12:59:15 -0400
Subject: Remove _simple

---
 .../joint_mlp_tgtcls_111_cswdtx_bigger_dropout.py  | 60 ++++++++++++++++++++++
 1 file changed, 60 insertions(+)
 create mode 100644 config/joint_mlp_tgtcls_111_cswdtx_bigger_dropout.py

(limited to 'config/joint_mlp_tgtcls_111_cswdtx_bigger_dropout.py')

diff --git a/config/joint_mlp_tgtcls_111_cswdtx_bigger_dropout.py b/config/joint_mlp_tgtcls_111_cswdtx_bigger_dropout.py
new file mode 100644
index 0000000..e0448cc
--- /dev/null
+++ b/config/joint_mlp_tgtcls_111_cswdtx_bigger_dropout.py
@@ -0,0 +1,60 @@
+import os
+import cPickle
+
+from blocks.initialization import IsotropicGaussian, Constant
+from blocks.bricks import Rectifier, Tanh, Logistic
+from blocks.filter import VariableFilter
+from blocks import roles
+
+import data
+from model.joint_mlp_tgtcls import Model, Stream
+
+
+n_begin_end_pts = 10     # how many points we consider at the beginning and end of the known trajectory
+
+with open(os.path.join(data.path, 'arrival-clusters.pkl')) as f:
+    dest_tgtcls = cPickle.load(f)
+
+# generate target classes for time prediction as a Fibonacci sequence
+time_tgtcls = [1, 2]
+for i in range(21):
+    time_tgtcls.append(time_tgtcls[-1] + time_tgtcls[-2])
+
+dim_embeddings = [
+    ('origin_call', data.origin_call_size, 15),
+    ('origin_stand', data.stands_size, 10),
+    ('week_of_year', 52, 10),
+    ('day_of_week', 7, 10),
+    ('qhour_of_day', 24 * 4, 10),
+    ('day_type', 3, 10),
+    ('taxi_id', 448, 10),
+]
+
+# Common network part
+dim_input = n_begin_end_pts * 2 * 2 + sum(x for (_, _, x) in dim_embeddings)
+dim_hidden = [5000]
+
+# Destination prediction part
+dim_hidden_dest = [1000]
+dim_output_dest = dest_tgtcls.shape[0]
+
+# Time prediction part
+dim_hidden_time = [500]
+dim_output_time = len(time_tgtcls)
+
+# Cost ratio between distance cost and time cost
+time_cost_factor = 4
+
+embed_weights_init = IsotropicGaussian(0.01)
+mlp_weights_init = IsotropicGaussian(0.1)
+mlp_biases_init = Constant(0.01)
+
+dropout = 0.5
+dropout_inputs = VariableFilter(bricks=[Rectifier], name='output')
+
+# use adadelta, so no learning_rate or momentum
+batch_size = 200
+
+valid_set = 'cuts/test_times_0'
+
+max_splits = 100
-- 
cgit v1.2.3