Merge pull request #1506 from luotao1/mnist

mnist api v2

demo/mnist/api_train_v2.py

@@ -1,6 +1,59 @@
 import paddle.v2 as paddle
 
 
+def softmax_regression(img):
+    predict = paddle.layer.fc(input=img,
+                              size=10,
+                              act=paddle.activation.Softmax())
+    return predict
+
+
+def multilayer_perceptron(img):
+    # The first fully-connected layer
+    hidden1 = paddle.layer.fc(input=img, size=128, act=paddle.activation.Relu())
+    # The second fully-connected layer and the according activation function
+    hidden2 = paddle.layer.fc(input=hidden1,
+                              size=64,
+                              act=paddle.activation.Relu())
+    # The thrid fully-connected layer, note that the hidden size should be 10,
+    # which is the number of unique digits
+    predict = paddle.layer.fc(input=hidden2,
+                              size=10,
+                              act=paddle.activation.Softmax())
+    return predict
+
+
+def convolutional_neural_network(img):
+    # first conv layer
+    conv_pool_1 = paddle.networks.simple_img_conv_pool(
+        input=img,
+        filter_size=5,
+        num_filters=20,
+        num_channel=1,
+        pool_size=2,
+        pool_stride=2,
+        act=paddle.activation.Tanh())
+    # second conv layer
+    conv_pool_2 = paddle.networks.simple_img_conv_pool(
+        input=conv_pool_1,
+        filter_size=5,
+        num_filters=50,
+        num_channel=20,
+        pool_size=2,
+        pool_stride=2,
+        act=paddle.activation.Tanh())
+    # The first fully-connected layer
+    fc1 = paddle.layer.fc(input=conv_pool_2,
+                          size=128,
+                          act=paddle.activation.Tanh())
+    # The softmax layer, note that the hidden size should be 10,
+    # which is the number of unique digits
+    predict = paddle.layer.fc(input=fc1,
+                              size=10,
+                              act=paddle.activation.Softmax())
+    return predict
+
+
 def main():
     paddle.init(use_gpu=False, trainer_count=1)
 
@@ -9,46 +62,58 @@ def main():
         name='pixel', type=paddle.data_type.dense_vector(784))
     label = paddle.layer.data(
         name='label', type=paddle.data_type.integer_value(10))
-    hidden1 = paddle.layer.fc(input=images, size=200)
-    hidden2 = paddle.layer.fc(input=hidden1, size=200)
-    inference = paddle.layer.fc(input=hidden2,
-                                size=10,
-                                act=paddle.activation.Softmax())
-    cost = paddle.layer.classification_cost(input=inference, label=label)
+
+    # Here we can build the prediction network in different ways. Please
+    # choose one by uncomment corresponding line.
+    predict = softmax_regression(images)
+    #predict = multilayer_perceptron(images)
+    #predict = convolutional_neural_network(images)
+
+    cost = paddle.layer.classification_cost(input=predict, label=label)
 
     parameters = paddle.parameters.create(cost)
 
-    adam_optimizer = paddle.optimizer.Adam(learning_rate=0.01)
+    optimizer = paddle.optimizer.Momentum(
+        learning_rate=0.1 / 128.0,
+        momentum=0.9,
+        regularization=paddle.optimizer.L2Regularization(rate=0.0005 * 128))
 
     trainer = paddle.trainer.SGD(cost=cost,
                                  parameters=parameters,
-                                 update_equation=adam_optimizer)
+                                 update_equation=optimizer)
+
+    lists = []
 
     def event_handler(event):
         if isinstance(event, paddle.event.EndIteration):
-            if event.batch_id % 1000 == 0:
+            if event.batch_id % 100 == 0:
+                print "Pass %d, Batch %d, Cost %f, %s" % (
+                    event.pass_id, event.batch_id, event.cost, event.metrics)
+        if isinstance(event, paddle.event.EndPass):
             result = trainer.test(reader=paddle.reader.batched(
-                    paddle.dataset.mnist.test(), batch_size=256))
-
-                print "Pass %d, Batch %d, Cost %.2f, %s\n" \
-                      "Testing cost %.2f metrics %s" % (
-                          event.pass_id, event.batch_id, event.cost,
-                          event.metrics,
-                          result.cost, result.metrics)
-        else:
-            pass
+                paddle.dataset.mnist.test(), batch_size=128))
+            print "Test with Pass %d, Cost %f, %s\n" % (
+                event.pass_id, result.cost, result.metrics)
+            lists.append((event.pass_id, result.cost,
+                          result.metrics['classification_error_evaluator']))
 
     trainer.train(
         reader=paddle.reader.batched(
             paddle.reader.shuffle(
                 paddle.dataset.mnist.train(), buf_size=8192),
-            batch_size=32),
-        event_handler=event_handler)
+            batch_size=128),
+        event_handler=event_handler,
+        num_passes=100)
+
+    # find the best pass
+    best = sorted(lists, key=lambda list: float(list[1]))[0]
+    print 'Best pass is %s, testing Avgcost is %s' % (best[0], best[1])
+    print 'The classification accuracy is %.2f%%' % (100 - float(best[2]) * 100)
 
     # output is a softmax layer. It returns probabilities.
     # Shape should be (100, 10)
     probs = paddle.infer(
-        output=inference,
+        output=predict,
         parameters=parameters,
         reader=paddle.reader.batched(
             paddle.reader.firstn(

python/paddle/v2/event.py

@@ -53,8 +53,9 @@ class EndPass(WithMetric):
     Event On One Pass Training Complete.
     """
 
-    def __init__(self, pass_id, evaluator):
+    def __init__(self, pass_id, cost, evaluator):
         self.pass_id = pass_id
+        self.cost = cost
         WithMetric.__init__(self, evaluator)