BN (Batch Normalization)
1.深度学习中的Batch Normalization
2.Batch Normalization导读
3.keras BatchNormalization层
tf.layers.batch_normalization(
inputs,
axis=-1,
momentum=0.99,
epsilon=0.001,
center=True,
scale=True,
beta_initializer=tf.zeros_initializer(),
gamma_initializer=tf.ones_initializer(),
moving_mean_initializer=tf.zeros_initializer(),
moving_variance_initializer=tf.ones_initializer(),
beta_regularizer=None,
gamma_regularizer=None,
beta_constraint=None,
gamma_constraint=None,
training=False,
trainable=True,
name=None,
reuse=None,
renorm=False,
renorm_clipping=None,
renorm_momentum=0.99,
fused=None,
virtual_batch_size=None,
adjustment=None
)
Batch Normalization layer from http://arxiv.org/abs/1502.03167.
"Batch Normalization: Accelerating Deep Network Training by Reducing
Internal Covariate Shift"
Sergey Ioffe, Christian Szegedy
Arguments:
axis: An int or list of int, the axis or axes that should be
normalized, typically the features axis/axes. For instance, after a
Conv2D layer with data_format="channels_first", set axis=1. If a
list of axes is provided, each axis in axis will be normalized
simultaneously. Default is -1 which takes uses last axis. Note: when
using multi-axis batch norm, the beta, gamma, moving_mean, and
moving_variance variables are the same rank as the input Tensor, with
dimension size 1 in all reduced (non-axis) dimensions).
momentum: Momentum for the moving average.
epsilon: Small float added to variance to avoid dividing by zero.
center: If True, add offset of beta to normalized tensor. If False, beta
is ignored.
scale: If True, multiply by gamma. If False, gamma is
not used. When the next layer is linear (also e.g. nn.relu), this can be
disabled since the scaling can be done by the next layer.
beta_initializer: Initializer for the beta weight.
gamma_initializer: Initializer for the gamma weight.
moving_mean_initializer: Initializer for the moving mean.
moving_variance_initializer: Initializer for the moving variance.
beta_regularizer: Optional regularizer for the beta weight.
gamma_regularizer: Optional regularizer for the gamma weight.
beta_constraint: An optional projection function to be applied to the beta
weight after being updated by an Optimizer (e.g. used to implement
norm constraints or value constraints for layer weights). The function
must take as input the unprojected variable and must return the
projected variable (which must have the same shape). Constraints are
not safe to use when doing asynchronous distributed training.
gamma_constraint: An optional projection function to be applied to the
gamma weight after being updated by an Optimizer.
renorm: Whether to use Batch Renormalization
(https://arxiv.org/abs/1702.03275). This adds extra variables during
training. The inference is the same for either value of this parameter.
renorm_clipping: A dictionary that may map keys 'rmax', 'rmin', 'dmax' to
scalar Tensors used to clip the renorm correction. The correction
(r, d) is used as corrected_value = normalized_value * r + d, with
r clipped to [rmin, rmax], and d to [-dmax, dmax]. Missing rmax, rmin,
dmax are set to inf, 0, inf, respectively.
renorm_momentum: Momentum used to update the moving means and standard
deviations with renorm. Unlike momentum, this affects training
and should be neither too small (which would add noise) nor too large
(which would give stale estimates). Note that momentum is still applied
to get the means and variances for inference.
fused: if None or True, use a faster, fused implementation if possible.
If False, use the system recommended implementation.
trainable: Boolean, if True also add variables to the graph collection
GraphKeys.TRAINABLE_VARIABLES (see tf.Variable).
virtual_batch_size: An int. By default, virtual_batch_size is None,
which means batch normalization is performed across the whole batch. When
virtual_batch_size is not None, instead perform "Ghost Batch
Normalization", which creates virtual sub-batches which are each
normalized separately (with shared gamma, beta, and moving statistics).
Must divide the actual batch size during execution.
adjustment: A function taking the Tensor containing the (dynamic) shape of
the input tensor and returning a pair (scale, bias) to apply to the
normalized values (before gamma and beta), only during training. For
example, if axis==-1,
adjustment = lambda shape: ( tf.random_uniform(shape[-1:], 0.93, 1.07), tf.random_uniform(shape[-1:], -0.1, 0.1))
will scale the normalized value by up to 7% up or down, then shift the
result by up to 0.1 (with independent scaling and bias for each feature
but shared across all examples), and finally apply gamma and/or beta. If
None, no adjustment is applied. Cannot be specified if
virtual_batch_size is specified.
name: A string, the name of the layer.
4.tf.layers.batch_normalization
5.CS231n笔记4-Data Preprocessing, Weights Initialization与Batch Normalization
6.Batch Renormalization : Towards Reducing Minibatch Dependence in Batch-Normalized Models
GN (Group Normalization)
Group Normalization
SN(Switchable Normalization)
深度剖析 | 可微分学习的自适配归一化
Self-Normalization
1.Self-Normalizing Neural Networks
2.引爆机器学习圈：「自归一化神经网络」提出新型激活函数SELU
3.自归一化神经网络
4.SNNs github
5.Add example to compare RELU with SELU
However, looking at Kaggle challenges that are not related to vision or sequential tasks, gradient boosting, random forests, or support vector machines (SVMs) are winning most of the competitions. Deep Learning is notably absent, and for the few cases where FNNs won, they are shallow. For example, the HIGGS challenge, the Merck Molecular Activity challenge, and the Tox21 Data challenge were all won by FNNs with at most four hidden layers. Surprisingly, it is hard to find success stories with FNNs that have many hidden layers, though they would allow for different levels of abstract representations of the input.
6.ReLU、LReLU、PReLU、CReLU、ELU、SELU
7.如何评价 Self-Normalizing Neural Networks 这篇论文?
补充
1.神经网络梯度与归一化问题总结+highway network、ResNet的思考
2.ICML 2018 | Petuum提出新型正则化方法：非重叠促进型变量选择