ThyrixYang

{"lastUpload":"2020-07-01T16:36:25.427Z","extensionVersion":"v3.4.3"}

ThyrixYang

import getpass
import io
import os
import requests
import retro.data
import subprocess
import sys
import tarfile
import tempfile
import zipfile

ThyrixYang

'''
author: Thyrix Yang
github: https://github.com/ThyrixYang
'''

# Deep DPG (DDPG)

# I didn't do much experiment on the model and hyper-parameters,
# and this is a implementation without some tricks mentioned in the paper.
# The result on this env is well, but worse than my CEM implementation...

ThyrixYang

import numpy as np
import gym
from gym import wrappers
import tensorflow as tf
import json, sys, os
from os import path
import random
from collections import deque

#####################################################################################################

ThyrixYang

'''
author: Thyrix Yang
github: https://github.com/ThyrixYang
'''


import numpy as np 
import gym

# Cross Entropy Method

ThyrixYang

'''
author: Thyrix Yang
github: https://github.com/ThyrixYang
'''


import numpy as np 
import gym

# Cross Entropy Method

ThyrixYang

'''
author: Thyrix Yang
github: https://github.com/ThyrixYang
'''



import numpy as np 
import gym

ThyrixYang

-- The C compiler identification is GNU 5.4.0
-- The CXX compiler identification is GNU 5.4.0
-- Check for working C compiler: /usr/bin/cc
-- Check for working C compiler: /usr/bin/cc -- works
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Detecting C compile features
-- Detecting C compile features - done
-- Check for working CXX compiler: /usr/bin/c++
-- Check for working CXX compiler: /usr/bin/c++ -- works

ThyrixYang

original test
average fail count on each model
at round 200
------------------------
0 0 0.0000000000
0 1 0.0000000000
0 2 0.0000000000
1 0 1.8900000000
1 1 0.0000000000
1 2 0.0000000000

ThyrixYang

The test result on mlpack_krann
I have tried four different threshold(5, 12, 20, 28), with the RAModelTest
Each test is run 200 times, 


round 200 fail 200 (run 200 times, the test failed 200 times)
------------------------
model id, j, failure count
0 0 0
0 1 0