Skip to content

Instantly share code, notes, and snippets.

Iaroslav Shcherbatyi iaroslav-ai

Block or report user

Report or block iaroslav-ai

Hide content and notifications from this user.

Learn more about blocking users

Contact Support about this user’s behavior.

Learn more about reporting abuse

Report abuse
View GitHub Profile
View Fit transformer on train partition only.ipynb
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
View SMOTE test.ipynb
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
View Disballanced Learning.ipynb
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
View skopt_obj_plot_experiment.py
from skopt import gp_minimize
from skopt.space import Real
from skopt.plots import plot_convergence
from skopt.plots import plot_objective, plot_evaluations
import numpy as np
np.random.seed(2)
# a dummy objective with 6 dimensions
View example_skopt_persistent_lie.py
"""
Comparison of parallel vs sequential optimization,
where constant lie appraoch is used.
"""
from threading import Thread
from copy import deepcopy
from skopt import Optimizer
from skopt.learning import ExtraTreesRegressor, GaussianProcessRegressor
from skopt.space import Real
import numpy as np
View example.py
from scipy.optimize import minimize, basinhopping, differential_evolution
import numpy as np
from copy import deepcopy
from skopt import Optimizer
from skopt.learning import GaussianProcessRegressor, RandomForestRegressor, ExtraTreesRegressor
from skopt.learning.gaussian_process.kernels import Matern
from skopt import space
from sklearn.svm import SVR
class MultiTaskOptProb():
@iaroslav-ai
iaroslav-ai / main.py
Created Mar 6, 2017
Knowledge transfer for sequential model based optimization
View main.py
from scipy.optimize import minimize, basinhopping, differential_evolution
import numpy as np
from autograd import numpy as np, grad
class SkoptProxy():
def __init__(self, ModelClass, bounds):
self.ModelClass = ModelClass
self.X = []
self.Y = []
@iaroslav-ai
iaroslav-ai / brain.py
Created Jun 19, 2016
Estimates the time needed to compute forward pass of neural net of size of human brain (10^15 synapses). Assumes that synapse implements function which can be well approximated by multiplication.
View brain.py
import numpy as np
import theano
from theano import tensor as T
# synapses to compute at once
N = 2 ** 13
M = N
Sym_rep = 16
x = np.random.randn(N).astype('float32')
View README.md

A slightly modified deep Q learning approach is used from this paper. Requires chainer.

To reproduce run code below with python 2.7; It will run training and monitor of the environment. Training data and some videos will be saved in "pendulum" folder near the script file.

Continuous space is discretized with 11 different actions.

It appears that there are some convergence problems; Maybe better selection of parameters would lead to a better objective value.

@iaroslav-ai
iaroslav-ai / README.md
Last active Jun 7, 2016
Cartpole RL experiment. Inspired by nature paper on atari game playing. Requires chainer.
View README.md

A slightly modified deep Q learning approach is used from this paper. Requires chainer. To reproduce run code below with python 2.7; It will run training and monitor of the environment. Training data and some videos will be saved in "cartpole" folder near the script file.

You can’t perform that action at this time.