steffen/self-replicating-neural-networks
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Refactor

Browse Source
This commit is contained in:
Si11ium
2019-03-04 15:24:01 +01:00
parent 415c7e9d41
commit f33db614b0
1 changed files with 57 additions and 51 deletions
Download Patch File Download Diff File Expand all files Collapse all files

52
code/network.py
View File

@@ -1,15 +1,18 @@
import math
import copy
import os
import numpy as np
import tensorflow as tf
from keras.models import Sequential, Model
from keras.models import Sequential
from keras.layers import SimpleRNN, Dense
from keras.layers import Input, TimeDistributed
from tqdm import tqdm
from experiment import FixpointExperiment, IdentLearningExperiment
# Supress warnings and info messages
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
def normalize_id(value, norm):
if norm > 1:
@@ -33,7 +36,7 @@ def are_weights_within(network_weights, lower_bound, upper_bound):
for layer_id, layer in enumerate(network_weights):
for cell_id, cell in enumerate(layer):
for weight_id, weight in enumerate(cell):
if not (lower_bound <= weight and weight <= upper_bound):
if not (lower_bound <= weight <= upper_bound):
return False
return True
@@ -58,7 +61,6 @@ class NeuralNetwork:
self.params.update(params)
self.keras_params = dict(activation='linear', use_bias=False)
self.silent = True
self.model = None
def silence(self):
self.silent = True
@@ -151,8 +153,8 @@ class WeightwiseNeuralNetwork(NeuralNetwork):
self.model.add(Dense(units=self.width, **self.keras_params))
self.model.add(Dense(units=1, **self.keras_params))
def apply(self, *input):
stuff = np.transpose(np.array([[input[0]], [input[1]], [input[2]], [input[3]]]))
def apply(self, *inputs):
stuff = np.transpose(np.array([[inputs[0]], [inputs[1]], [inputs[2]], [inputs[3]]]))
return self.model.predict(stuff)[0][0]
def apply_to_weights(self, old_weights):
@@ -174,7 +176,10 @@ class WeightwiseNeuralNetwork(NeuralNetwork):
new_weights[layer_id][cell_id][weight_id] = new_weight
if self.params.get("print_all_weight_updates", False) and not self.silent:
print("updated old weight " + str(weight) + "\t @ (" + str(layer_id) + "," + str(cell_id) + "," + str(weight_id) + ") to new value " + str(new_weight) + "\t calling @ (" + str(normal_layer_id) + "," + str(normal_cell_id) + "," + str(normal_weight_id) + ")")
print("updated old weight {weight}\t @ ({layer},{cell},{weight_id}) "
"to new value {new_weight}\t calling @ ({n_layer},{n_cell},{n_weight_id})").format(
weight=weight, layer=layer_id, cell=cell_id, weight_id=weight_id, new_weight=new_weight,
n_layer=normal_layer_id, n_cell=normal_cell_id, n_weight_id=normal_weight_id)
return new_weights
@@ -237,8 +242,8 @@ class AggregatingNeuralNetwork(NeuralNetwork):
total_weights += 1
return total_weights
def apply(self, *input):
stuff = np.transpose(np.array([[input[i]] for i in range(self.aggregates)]))
def apply(self, *inputs):
stuff = np.transpose(np.array([[inputs[i]] for i in range(self.aggregates)]))
return self.model.predict(stuff)[0]
def apply_to_weights(self, old_weights):
@@ -298,8 +303,8 @@ class RecurrentNeuralNetwork(NeuralNetwork):
self.model.add(SimpleRNN(units=width, return_sequences=True, **self.keras_params))
self.model.add(SimpleRNN(units=self.features, return_sequences=True, **self.keras_params))
def apply(self, *input):
stuff = np.transpose(np.array([[[input[i]] for i in range(len(input))]]))
def apply(self, *inputs):
stuff = np.transpose(np.array([[[inputs[i]] for i in range(len(inputs))]]))
return self.model.predict(stuff)[0].flatten()
def apply_to_weights(self, old_weights):
@@ -329,7 +334,7 @@ class LearningNeuralNetwork(NeuralNetwork):
def mean_reduction(weights, features):
single_dim_weights = np.hstack([w.flatten() for w in weights])
shaped_weights = np.reshape(single_dim_weights, (1, features, -1))
x = np.mean(shaped_weights, axis=1)
x = np.mean(shaped_weights, axis=-1)
return x
@staticmethod
@@ -339,16 +344,15 @@ class LearningNeuralNetwork(NeuralNetwork):
return x
@staticmethod
def random_reduction(weights, features):
def random_reduction(_, features):
x = np.random.rand(features)[None, ...]
return x
def __init__(self, width, depth, features, reduction, **kwargs):
def __init__(self, width, depth, features, **kwargs):
super().__init__(**kwargs)
self.width = width
self.depth = depth
self.features = features
self.reduction = reduction
self.compile_params = dict(loss='mse', optimizer='sgd')
self.model.add(Dense(units=self.width, input_dim=self.features, **self.keras_params))
for _ in range(self.depth-1):
@@ -360,13 +364,13 @@ class LearningNeuralNetwork(NeuralNetwork):
self.compile_params.update(kwargs)
return self
def learn(self, epochs, batchsize=1):
def learn(self, epochs, reduction, batchsize=1):
with tqdm(total=epochs, ascii=True,
desc='Type: {t} @ Epoch:'.format(t=self.__class__.__name__),
postfix=["Loss", dict(value=0)]) as bar:
for epoch in range(epochs):
old_weights = self.get_weights()
x = self.reduction(old_weights)
x = reduction(old_weights, self.features)
history = self.model.fit(x=x, y=x, verbose=0, batch_size=batchsize)
bar.postfix[1]["value"] = history.history['loss'][-1]
bar.update()
@@ -377,8 +381,11 @@ if __name__ == '__main__':
with FixpointExperiment() as exp:
for run_id in tqdm(range(100)):
# net = WeightwiseNeuralNetwork(width=2, depth=2).with_keras_params(activation='linear')
net = AggregatingNeuralNetwork(aggregates=4, width=2, depth=2).with_keras_params(activation='linear').with_params(shuffler=AggregatingNeuralNetwork.shuffle_random, print_all_weight_updates=False, use_bias=True)
# net = RecurrentNeuralNetwork(width=2, depth=2).with_keras_params(activation='linear').with_params(print_all_weight_updates=True)
net = AggregatingNeuralNetwork(aggregates=4, width=2, depth=2).with_keras_params(activation='linear')\
.with_params(shuffler=AggregatingNeuralNetwork.shuffle_random,
print_all_weight_updates=False, use_bias=True)
# net = RecurrentNeuralNetwork(width=2, depth=2).with_keras_params(activation='linear')\
# .with_params(print_all_weight_updates=True)
# net.print_weights()
exp.run_net(net, 100)
@@ -386,8 +393,7 @@ if __name__ == '__main__':
if True:
with IdentLearningExperiment() as exp:
net = LearningNeuralNetwork(width=2, depth=2, features=2, reduction=LearningNeuralNetwork.random_reduction)\
net = LearningNeuralNetwork(width=2, depth=2, features=2, )\
.with_keras_params(activation='linear') \
.with_params(print_all_weight_updates=False)
net.learn(1000)
net.learn(1000, reduction=LearningNeuralNetwork.mean_reduction)