OpenAI CartPole-v0 solved with a Support Vector Regressor - https://gym.openai.com/evaluations/eval_h4Cbis6SiSfwXHf7HuVnw#reproducibility

cartpole-v0_SVR.py

import gym
from gym import wrappers

import numpy as np
from sklearn import svm

env = gym.make('CartPole-v0')
env = wrappers.Monitor(env, '/tmp/cartpole-experiment-1', force=True)

observations = []
actions = []
rewards = []

model = None

def update_model():
    """Updates the  prediction model using all data available"""
    global model
    a_observations = np.array(observations)
    a_actions = np.array(actions).reshape((len(actions), 1))
    X = np.hstack([a_observations, a_actions])
    a_rewards = np.array(rewards)
    y = np.stack(rewards)
    model = svm.SVR()
    model.fit(X, y)

def predict(observation, exploring_ratio):
    """Predicts the best action using the current observation"""
    if model is None or np.random.random() < exploring_ratio:
        return np.random.choice(range(2),1)[0]
    X = np.hstack([np.array([observation, observation]), np.array([[0],[1]])])
    pred = model.predict(X)
    return list(range(2))[np.argmax(pred)]

for i_episode in range(200):
    observation = env.reset()
    reward = 0
    for t in range(1000):
        env.render()
        #print observation, reward
        observations.append(observation)
        action = predict(observation, 1/(i_episode/8.0 + 1))
        actions.append(action)
        observation, reward, done, info = env.step(action)
        if done:
            print("Episode {} finished after {} timesteps".format(i_episode, t+1))
            if t < 199: rewards += (np.array(list(reversed(range(t+1))))/float(t)).tolist()   #Add cumulative rewards
            else: rewards += [1]*(t+1)
            break
    update_model()

env.close()