神经网络python包_实现连续时间递归神经网络(CTRNNs)的Python包

CTRNN

Python package that implements Continuous Time Recurrent Neural Networks (CTRNNs)

See Beer, R.D. (1995). On the dynamics of small continuous-time recurrent neural networks. Adaptive Behavior 3:469-509. for a study of CTRNNs.

Using this repo in conjunction with StochSearch allows you to optimize CTRNNs to perform different tasks. Stochsearch uses Python's multiprocessing framework to parallelize population based stochastic search optimization methods.

You can also find a Tensorflow version of building and implementing a population of CTRNNs here. Refer to this same repo for optimizing CTRNNs using Tensorflow on a GPU.

Installation instructions

$ pip install CTRNN

Requirements: numpy, scipy

Using the package

Importing the CTRNN package:

from CTRNN import CTRNN

Creating a CTRNN object:

cns = CTRNN(network_size,step_size=0.1)