High-dimensional neural network¶
Introduction¶
SIMPLE-NN use High-Dimensional Neural Network(HDNN) [1] as a default machine learning model.
Parameters¶
Exponential decay¶
Some parameters in neural_network may need to decrease exponentially during the optimization process. In those cases, you can use this format instead of float value. More information can be found in Tensorflow homepage
parameter_name:
learning_rate: 1.
decay_rate: 0.95
decay_steps: 10000
staircase: false
Note
If continue: true
, global_step
(see the link above) of save points is also loaded.
Thus, you need to consider the global_step
to calculate the values from exponential_decay
.
On the contrary, global_step
is reset when continue: weights
Methods¶
-
__init__
(self)¶ Initiator of Neural_network class.
-
train
(self, user_optimizer=None, aw_modifier=None)¶ - Args:
user_optimizer
: User defined optimizer. Can be set in the script run.pyaw_modifier
: scale function for atomic weights.
Method for optimizing neural network potential.
References
[1] | J. Behler, M. Parrinello, Phys. Rev. Lett. 98 (2007) 146401 |