def get_approximate_options(n, m, methods, losses):
#X, y = load_boston(return_X_y=True)
#x = X[:100]
#x_validate = X[420:]
x = [[y] * m for y in range(1, 25)]
x_validate = [[y + 0.5] * m for y in range(0, 25)]
#x = [[y] * m for y in range(1, 101)]
#x_validate = [[y + 0.5] * m for y in range(0, 100)]
#X, y = load_diabetes(return_X_y=True)
#x = X[:50]
#x_validate = X[400:]
options = []
for method in methods:
options.append({
"x": [convert_variables_without_trainable(x_tmp) for x_tmp in x],
"x_validate": [
convert_variables_without_trainable(x_tmp)
for x_tmp in x_validate
],
"params":
convert_variables([1 for x in range(n)]),
"loss_function":
losses,
"opt":
method,
"eps":
0.0001,
"max_steps":
100,
})
return options
from utils.generate_data import generate_x
from utils.utils import convert_variables, convert_variables_without_trainable
from utils.generate_data import generate_set
from sklearn.datasets import load_boston
X, y = load_boston(return_X_y=True)
x = X[:100]
x_validate = X[420:]
approximate_options9_3 = [
{
"x": [convert_variables_without_trainable(x_tmp) for x_tmp in x],
"x_validate": [convert_variables_without_trainable(x_tmp) for x_tmp in x_validate],
"params": convert_variables([5 for x in range(79)]),
"loss_function": tf.keras.losses.MSE,
"opt": tf.keras.optimizers.SGD(learning_rate=1),
"eps": 0.0001,
"max_steps": 30,
},
{
"x": [convert_variables_without_trainable(x_tmp) for x_tmp in x],
"x_validate": [convert_variables_without_trainable(x_tmp) for x_tmp in x_validate],
"params": convert_variables([5 for x in range(79)]),
"loss_function": tf.keras.losses.MSE,
"opt": tf.keras.optimizers.SGD(learning_rate=0.1),
"eps": 0.0001,
"max_steps": 30,
import tensorflow as tf
from utils.utils import convert_variables, convert_variables_without_trainable
"""
Approximate options
"""
approximate_options2 = [
{
"x": [
convert_variables_without_trainable([x, x])
for x in range(-10, 10, 1)
],
"x_validate": [
convert_variables_without_trainable([x, x])
for x in range(-11, 10, 1)
],
"params":
convert_variables([1 for x in range(11)]),
"loss_function":
tf.keras.losses.MAE,
"opt":
tf.keras.optimizers.Adam(learning_rate=0.001),
"eps":
0.0001,
"max_steps":
1000,
},
{
"x": [
convert_variables_without_trainable([x, x])
def __init__(self):
self.val = {}
X, y = load_diabetes(return_X_y=True)
for k, v in zip(X, y):
self.val[str(convert_variables_without_trainable(k))] = v
import numpy as np
import tensorflow as tf
from utils.utils import convert_variables, convert_variables_without_trainable
"""
Approximate options
"""
approximate_options1 = [
{
"x": [convert_variables_without_trainable([x, x]) for x in list(np.arange(0, 5, 0.2))],
"x_validate": [convert_variables_without_trainable([x, x]) for x in list(np.arange(0.1, 5, 0.02))],
"params": convert_variables([1 for x in range(11)]),
"loss_function": tf.keras.losses.MAE,
"opt": tf.keras.optimizers.Adam(learning_rate=0.001),
"eps": 0.0001,
"max_steps": 1000,
},
{
"x": [convert_variables_without_trainable([x, x]) for x in list(np.arange(0, 5, 0.2))],
"x_validate": [convert_variables_without_trainable([x, x]) for x in list(np.arange(0.1, 5, 0.02))],
"params": convert_variables([1 for x in range(11)]),
"loss_function": tf.keras.losses.MAE,
"opt": tf.keras.optimizers.Adam(learning_rate=0.01),
"eps": 0.0001,
"max_steps": 10000,
},
{
"x": [convert_variables_without_trainable([x, x]) for x in list(np.arange(0, 5, 0.2))],
"x_validate": [convert_variables_without_trainable([x, x]) for x in list(np.arange(0.1, 5, 0.02))],