混合 CNN 上的 InvalidArgumentError
InvalidArgumentError on a mixed CNN
我对 Tensorflow/Keras 和深度学习还很陌生，所以提前致歉。
我正在创建一个基本的混合卷积神经网络来对图像和元数据进行分类。我使用 Keras Functional API 创建了以下内容：
# Define inputs
meta_inputs = tf.keras.Input(shape=(2065,))
img_inputs = tf.keras.Input(shape=(80,120,3,))
# Model 1
meta_layer1 = tf.keras.layers.Dense(64, activation='relu')(meta_inputs)
meta_output_layer = tf.keras.layers.Dense(1, activation='sigmoid')(meta_layer1)
# Model 2
img_conv_layer1 = tf.keras.layers.Conv2D(32, kernel_size=(3,3), padding='same', activation='relu')(img_inputs)
img_pooling_layer1 = tf.keras.layers.MaxPooling2D()(img_conv_layer1)
img_conv_layer2 = tf.keras.layers.Conv2D(64, kernel_size=(3,3), padding='same', activation='relu')(img_pooling_layer1)
img_pooling_layer2 = tf.keras.layers.MaxPooling2D()(img_conv_layer2)
img_flatten_layer = tf.keras.layers.Flatten()(img_pooling_layer2)
img_dense_layer = tf.keras.layers.Dense(1024, activation='relu')(img_flatten_layer)
img_output_layer = tf.keras.layers.Dense(1, activation='sigmoid')(img_dense_layer)
# Merge models
merged = tf.keras.layers.add([meta_output_layer, img_output_layer])
# Define functional model
model = tf.keras.Model(inputs=[meta_inputs, img_inputs], outputs=merged)
# Compile model
auc = tf.keras.metrics.AUC(name = 'auc')
model.compile('adam', loss='binary_crossentropy', metrics=[auc])
然后我继续拟合模型：
epochs = 15
history = model.fit([meta_train, img_train], y_train, epochs=epochs, batch_size=500, validation_data=([meta_test, img_test], y_test))
这会产生错误，坦率地说，我不确定如何处理它：
InvalidArgumentError Traceback (most recent call last)
<ipython-input-11-5ec0cf9ac1d1> in <module>
1 epochs = 15
----> 2 history = model.fit([meta_train, img_train], y_train, epochs=epochs, batch_size=500, validation_data=([meta_test, img_test], y_test))
~/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/training.py in fit(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, validation_batch_size, validation_freq, max_queue_size, workers, use_multiprocessing)
1098 _r=1):
1099 callbacks.on_train_batch_begin(step)
-> 1100 tmp_logs = self.train_function(iterator)
1101 if data_handler.should_sync:
1102 context.async_wait()
~/anaconda3/lib/python3.8/site-packages/tensorflow/python/eager/def_function.py in __call__(self, *args, **kwds)
826 tracing_count = self.experimental_get_tracing_count()
827 with trace.Trace(self._name) as tm:
--> 828 result = self._call(*args, **kwds)
829 compiler = "xla" if self._experimental_compile else "nonXla"
830 new_tracing_count = self.experimental_get_tracing_count()
~/anaconda3/lib/python3.8/site-packages/tensorflow/python/eager/def_function.py in _call(self, *args, **kwds)
886 # Lifting succeeded, so variables are initialized and we can run the
887 # stateless function.
--> 888 return self._stateless_fn(*args, **kwds)
889 else:
890 _, _, _, filtered_flat_args = \
~/anaconda3/lib/python3.8/site-packages/tensorflow/python/eager/function.py in __call__(self, *args, **kwargs)
2940 (graph_function,
2941 filtered_flat_args) = self._maybe_define_function(args, kwargs)
-> 2942 return graph_function._call_flat(
2943 filtered_flat_args, captured_inputs=graph_function.captured_inputs) # pylint: disable=protected-access
2944
~/anaconda3/lib/python3.8/site-packages/tensorflow/python/eager/function.py in _call_flat(self, args, captured_inputs, cancellation_manager)
1916 and executing_eagerly):
1917 # No tape is watching; skip to running the function.
-> 1918 return self._build_call_outputs(self._inference_function.call(
1919 ctx, args, cancellation_manager=cancellation_manager))
1920 forward_backward = self._select_forward_and_backward_functions(
~/anaconda3/lib/python3.8/site-packages/tensorflow/python/eager/function.py in call(self, ctx, args, cancellation_manager)
553 with _InterpolateFunctionError(self):
554 if cancellation_manager is None:
--> 555 outputs = execute.execute(
556 str(self.signature.name),
557 num_outputs=self._num_outputs,
~/anaconda3/lib/python3.8/site-packages/tensorflow/python/eager/execute.py in quick_execute(op_name, num_outputs, inputs, attrs, ctx, name)
57 try:
58 ctx.ensure_initialized()
---> 59 tensors = pywrap_tfe.TFE_Py_Execute(ctx._handle, device_name, op_name,
60 inputs, attrs, num_outputs)
61 except core._NotOkStatusException as e:
InvalidArgumentError: assertion failed: [predictions must be <= 1] [Condition x <= y did not hold element-wise:] [x (model/add/add:0) = ] [[1.47704351][1.48876262][1.50816929]...] [y (Cast_4/x:0) = ] [1]
[[{{node assert_less_equal/Assert/AssertGuard/else/_11/assert_less_equal/Assert/AssertGuard/Assert}}]] [Op:__inference_train_function_1331]
Function call stack:
train_function
一般来说，作为 Tensorflow、Keras 和深度学习的新手，我什至不确定从哪里开始诊断问题。我不知道 [predictions must be <= 1] [Condition x <= y did not hold element-wise:] 指的是什么，例如： y_train 是一个 1 和 0 的数组，应该坚持那个断言。我什至尝试将其重塑为 (N,1)-D 数组，但没有效果。
任何人都可以指出我正确的方向吗？
AUC 指标需要 [0,1] 内的概率。
在您的模型中，这不会发生，因为您在 merged 层中进行了求和。例如，您可以使用平均值而不是求和来求解：
merged = tf.keras.layers.Average()([meta_output_layer, img_output_layer])
我对 Tensorflow/Keras 和深度学习还很陌生，所以提前致歉。
我正在创建一个基本的混合卷积神经网络来对图像和元数据进行分类。我使用 Keras Functional API 创建了以下内容：
# Define inputs
meta_inputs = tf.keras.Input(shape=(2065,))
img_inputs = tf.keras.Input(shape=(80,120,3,))
# Model 1
meta_layer1 = tf.keras.layers.Dense(64, activation='relu')(meta_inputs)
meta_output_layer = tf.keras.layers.Dense(1, activation='sigmoid')(meta_layer1)
# Model 2
img_conv_layer1 = tf.keras.layers.Conv2D(32, kernel_size=(3,3), padding='same', activation='relu')(img_inputs)
img_pooling_layer1 = tf.keras.layers.MaxPooling2D()(img_conv_layer1)
img_conv_layer2 = tf.keras.layers.Conv2D(64, kernel_size=(3,3), padding='same', activation='relu')(img_pooling_layer1)
img_pooling_layer2 = tf.keras.layers.MaxPooling2D()(img_conv_layer2)
img_flatten_layer = tf.keras.layers.Flatten()(img_pooling_layer2)
img_dense_layer = tf.keras.layers.Dense(1024, activation='relu')(img_flatten_layer)
img_output_layer = tf.keras.layers.Dense(1, activation='sigmoid')(img_dense_layer)
# Merge models
merged = tf.keras.layers.add([meta_output_layer, img_output_layer])
# Define functional model
model = tf.keras.Model(inputs=[meta_inputs, img_inputs], outputs=merged)
# Compile model
auc = tf.keras.metrics.AUC(name = 'auc')
model.compile('adam', loss='binary_crossentropy', metrics=[auc])
然后我继续拟合模型：
epochs = 15
history = model.fit([meta_train, img_train], y_train, epochs=epochs, batch_size=500, validation_data=([meta_test, img_test], y_test))
这会产生错误，坦率地说，我不确定如何处理它：
InvalidArgumentError Traceback (most recent call last)
<ipython-input-11-5ec0cf9ac1d1> in <module>
1 epochs = 15
----> 2 history = model.fit([meta_train, img_train], y_train, epochs=epochs, batch_size=500, validation_data=([meta_test, img_test], y_test))
~/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/training.py in fit(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, validation_batch_size, validation_freq, max_queue_size, workers, use_multiprocessing)
1098 _r=1):
1099 callbacks.on_train_batch_begin(step)
-> 1100 tmp_logs = self.train_function(iterator)
1101 if data_handler.should_sync:
1102 context.async_wait()
~/anaconda3/lib/python3.8/site-packages/tensorflow/python/eager/def_function.py in __call__(self, *args, **kwds)
826 tracing_count = self.experimental_get_tracing_count()
827 with trace.Trace(self._name) as tm:
--> 828 result = self._call(*args, **kwds)
829 compiler = "xla" if self._experimental_compile else "nonXla"
830 new_tracing_count = self.experimental_get_tracing_count()
~/anaconda3/lib/python3.8/site-packages/tensorflow/python/eager/def_function.py in _call(self, *args, **kwds)
886 # Lifting succeeded, so variables are initialized and we can run the
887 # stateless function.
--> 888 return self._stateless_fn(*args, **kwds)
889 else:
890 _, _, _, filtered_flat_args = \
~/anaconda3/lib/python3.8/site-packages/tensorflow/python/eager/function.py in __call__(self, *args, **kwargs)
2940 (graph_function,
2941 filtered_flat_args) = self._maybe_define_function(args, kwargs)
-> 2942 return graph_function._call_flat(
2943 filtered_flat_args, captured_inputs=graph_function.captured_inputs) # pylint: disable=protected-access
2944
~/anaconda3/lib/python3.8/site-packages/tensorflow/python/eager/function.py in _call_flat(self, args, captured_inputs, cancellation_manager)
1916 and executing_eagerly):
1917 # No tape is watching; skip to running the function.
-> 1918 return self._build_call_outputs(self._inference_function.call(
1919 ctx, args, cancellation_manager=cancellation_manager))
1920 forward_backward = self._select_forward_and_backward_functions(
~/anaconda3/lib/python3.8/site-packages/tensorflow/python/eager/function.py in call(self, ctx, args, cancellation_manager)
553 with _InterpolateFunctionError(self):
554 if cancellation_manager is None:
--> 555 outputs = execute.execute(
556 str(self.signature.name),
557 num_outputs=self._num_outputs,
~/anaconda3/lib/python3.8/site-packages/tensorflow/python/eager/execute.py in quick_execute(op_name, num_outputs, inputs, attrs, ctx, name)
57 try:
58 ctx.ensure_initialized()
---> 59 tensors = pywrap_tfe.TFE_Py_Execute(ctx._handle, device_name, op_name,
60 inputs, attrs, num_outputs)
61 except core._NotOkStatusException as e:
InvalidArgumentError: assertion failed: [predictions must be <= 1] [Condition x <= y did not hold element-wise:] [x (model/add/add:0) = ] [[1.47704351][1.48876262][1.50816929]...] [y (Cast_4/x:0) = ] [1]
[[{{node assert_less_equal/Assert/AssertGuard/else/_11/assert_less_equal/Assert/AssertGuard/Assert}}]] [Op:__inference_train_function_1331]
Function call stack:
train_function
一般来说，作为 Tensorflow、Keras 和深度学习的新手，我什至不确定从哪里开始诊断问题。我不知道 [predictions must be <= 1] [Condition x <= y did not hold element-wise:] 指的是什么，例如： y_train 是一个 1 和 0 的数组，应该坚持那个断言。我什至尝试将其重塑为 (N,1)-D 数组，但没有效果。
任何人都可以指出我正确的方向吗？
AUC 指标需要 [0,1] 内的概率。
在您的模型中，这不会发生，因为您在 merged 层中进行了求和。例如，您可以使用平均值而不是求和来求解：
merged = tf.keras.layers.Average()([meta_output_layer, img_output_layer])