diff --git a/.gitignore b/.gitignore
index 1883a3f..1c3cd4c 100644
--- a/.gitignore
+++ b/.gitignore
@@ -34,6 +34,11 @@ var/
 pip-log.txt
 pip-delete-this-directory.txt
 
+# PyCharms
+*.xml
+*.iml
+.idea/
+
 # Unit test / coverage reports
 htmlcov/
 .tox/
diff --git a/keras_contrib/layers/advanced_activations.py b/keras_contrib/layers/advanced_activations.py
index 6b39244..f83f044 100644
--- a/keras_contrib/layers/advanced_activations.py
+++ b/keras_contrib/layers/advanced_activations.py
@@ -1,4 +1,84 @@
 from .. import initializations
 from keras.engine import Layer
+from keras.utils.generic_utils import get_custom_objects
 from .. import backend as K
 import numpy as np
+
+
+class PELU(Layer):
+    """Parametric Exponential Linear Unit.
+    It follows:
+    `f(x) = alphas * (exp(x / betas) - 1) for x < 0`,
+    `f(x) = (alphas / betas) * x for x >= 0`,
+    where `alphas` & `betas` are learned arrays with the same shape as x.
+    # Input shape
+        Arbitrary. Use the keyword argument `input_shape`
+        (tuple of integers, does not include the samples axis)
+        when using this layer as the first layer in a model.
+    # Output shape
+        Same shape as the input.
+    # Arguments
+        alphas_init: initialization function for the alpha variable weights.
+        betas_init: initialization function for the beta variable weights.
+        weights: initial weights, as a list of a single Numpy array.
+        shared_axes: the axes along which to share learnable
+            parameters for the activation function.
+            For example, if the incoming feature maps
+            are from a 2D convolution
+            with output shape `(batch, height, width, channels)`,
+            and you wish to share parameters across space
+            so that each filter only has one set of parameters,
+            set `shared_axes=[1, 2]`.
+    # References
+        - [PARAMETRIC EXPONENTIAL LINEAR UNIT FOR DEEP CONVOLUTIONAL NEURAL NETWORKS](https://arxiv.org/abs/1605.09332v3)
+    """
+
+    def __init__(self, alphas_init='one', betas_init='one', weights=None, shared_axes=None, **kwargs):
+        self.supports_masking = True
+        self.alphas_init = initializations.get(alphas_init)
+        self.betas_init = initializations.get(betas_init)
+        self.initial_weights = weights
+        if not isinstance(shared_axes, (list, tuple)):
+            self.shared_axes = [shared_axes]
+        else:
+            self.shared_axes = list(shared_axes)
+        super(PELU, self).__init__(**kwargs)
+
+    def build(self, input_shape):
+        param_shape = list(input_shape[1:])
+        self.param_broadcast = [False] * len(param_shape)
+        if self.shared_axes[0] is not None:
+            for i in self.shared_axes:
+                param_shape[i - 1] = 1
+                self.param_broadcast[i - 1] = True
+
+        # Initialised as ones to emulate the default ELU
+        self.alphas = self.alphas_init(param_shape,
+                                       name='{}_alphas'.format(self.name))
+        self.betas = self.betas_init(param_shape,
+                                     name='{}_betas'.format(self.name))
+
+        self.trainable_weights = [self.alphas, self.betas]
+
+        if self.initial_weights is not None:
+            self.set_weights(self.initial_weights)
+            del self.initial_weights
+
+    def call(self, x, mask=None):
+        if K.backend() == 'theano':
+            pos = x * (K.pattern_broadcast(self.alphas, self.param_broadcast) /
+                               K.pattern_broadcast(self.betas, self.param_broadcast))
+            neg = (K.pattern_broadcast(self.alphas, self.param_broadcast) *
+                   (K.exp(x / K.pattern_broadcast(self.betas, self.param_broadcast)) - 1))
+        else:
+            pos = K.relu(x) * (self.alphas / self.betas)
+            neg = self.alphas * (K.exp(x / self.betas) - 1)
+        return K.switch(x < 0., neg, pos)
+
+    def get_config(self):
+        config = {'alphas_init': self.alphas_init.__name__,
+                  'betas_init': self.betas_init.__name__}
+        base_config = super(PELU, self).get_config()
+        return dict(list(base_config.items()) + list(config.items()))
+
+get_custom_objects().update({"PELU": PELU})
diff --git a/tests/keras_contrib/layers/test_advanced_activations.py b/tests/keras_contrib/layers/test_advanced_activations.py
index d510fc9..0dbfb69 100644
--- a/tests/keras_contrib/layers/test_advanced_activations.py
+++ b/tests/keras_contrib/layers/test_advanced_activations.py
@@ -1,5 +1,18 @@
 import pytest
 from keras.utils.test_utils import layer_test, keras_test
+from keras_contrib.layers import advanced_activations
+
+
+@keras_test
+def test_pelu():
+    layer_test(advanced_activations.PELU, kwargs={},
+               input_shape=(2, 3, 4))
+
+
+@keras_test
+def test_pelu_share():
+    layer_test(advanced_activations.PELU, kwargs={'shared_axes': 1},
+               input_shape=(2, 3, 4))
 
 
 if __name__ == '__main__':