mindspore.ops.ApplyCenteredRMSProp

class mindspore.ops.ApplyCenteredRMSProp(*args, **kwargs)

Optimizer that implements the centered RMSProp algorithm. Please refer to the usage in source code of nn.RMSProp.

The updating formulas of ApplyCenteredRMSProp algorithm are as follows,

$$\begin{array}{r}\begin{array}{l}\\ g_t=\rho g_{t-1}+(1-\rho )\mathrm{\nabla }{Q}_i(w)\\ s_t=\rho s_{t-1}+(1-\rho )(\mathrm{\nabla }{Q}_i(w){)}^2\\ m_t=\beta m_{t-1}+\frac{\eta }{\sqrt{s_t-g_t^2+ϵ}}\mathrm{\nabla }{Q}_i(w)\\ w=w-m_t\end{array}\end{array}$$

where $w$ represents var, which will be updated. $g_t$ represents mean_gradient, $g_{t-1}$ is the last momentent of $g_t$. $s_t$ represents mean_square, $s_{t-1}$ is the last momentent of $s_t$, $m_t$ represents moment, $m_{t-1}$ is the last momentent of $m_t$. $rho$ represents decay. $beta$ is the momentum term, represents momentum. $epsilon$ is a smoothing term to avoid division by zero, represents epsilon. $eta$ represents learning_rate. $nabla{Q}_i(w)$ represents grad.

Parameters

use_locking (bool) – Whether to enable a lock to protect the variable and accumlation tensors from being updated. Default: False.

Inputs:

• var (Tensor) - Weights to be update.

• mean_gradient (Tensor) - Mean gradients, must have the same type as var.

• mean_square (Tensor) - Mean square gradients, must have the same type as var.

• moment (Tensor) - Delta of var, must have the same type as var.

• grad (Tensor) - Gradient, must have the same type as var.

• learning_rate (Union[Number, Tensor]) - Learning rate. Must be a float number or a scalar tensor with float16 or float32 data type.

• decay (float) - Decay rate.

• momentum (float) - Momentum.

• epsilon (float) - Ridge term.

Outputs:

Tensor, parameters to be update.

Raises

• TypeError – If use_locking is not a bool.

• TypeError – If var, mean_gradient, mean_square, moment or grad is not a Tensor.

• TypeError – If learing_rate is neither a Number nor a Tensor.

• TypeError – If dtype of learing_rate is neither float16 nor float32.

• TypeError – If dtype of decay, momentum or epsilon is not float.

Supported Platforms:

Ascend GPU CPU

Examples

>>> centered_rms_prop = ops.ApplyCenteredRMSProp()
>>> input_x = Tensor(np.arange(-2, 2).astype(np.float32).reshape(2, 2), mindspore.float32)
>>> mean_grad = Tensor(np.arange(4).astype(np.float32).reshape(2, 2), mindspore.float32)
>>> mean_square = Tensor(np.arange(-3, 1).astype(np.float32).reshape(2, 2), mindspore.float32)
>>> moment = Tensor(np.arange(4).astype(np.float32).reshape(2, 2), mindspore.float32)
>>> grad = Tensor(np.arange(4).astype(np.float32).reshape(2, 2), mindspore.float32)
>>> learning_rate = Tensor(0.9, mindspore.float32)
>>> decay = 0.0
>>> momentum = 1e-10
>>> epsilon = 0.05
>>> output = centered_rms_prop(input_x, mean_grad, mean_square, moment, grad,
...                            learning_rate, decay, momentum, epsilon)
>>> output
Tensor(shape=[2, 2], dtype=Float32, value=
[[-2.00000000e+00, -5.02492237e+00],
 [-8.04984474e+00, -1.10747662e+01]])