HackyHour Würzburg 35

• When: May 23^rd, 2019 at 5:00pm
• Where: Center for Computational and Theoretical Biology (CCTB)
• Info: HackyHour Website

Topic Suggestions

Add your :+1: to the end of a line you are interested in

• Say it challenge (hacker.org)
• organizing a fun coding competition (fbctf?)
• Make the computer play computer games (Reinforcement Learning, gym)
• Text Mining / Fact Extraction / Knowledge Graphs / WikiData
• Proteomics
• natural language processing

Ideas for another time

• revisit GANs maybe with this tutorial
• Quantum computing (for the very curious)
• AutoML (PennAI)
• GNU Guix

Participants

• Markus :pizza:
• Matthias :sushi:
• Franzi :sunflower: :sushi:
• Michaela

Cross Links

• previous pad
• next pad

Manual solution

import gym
env = gym.make('CartPole-v0')
for i_episode in range(20):
    observation = env.reset()
    last_wiggle = 0
    for t in range(1000):

        env.render()
        #print(observation)
        if observation[2] > 0.1:
            action = 1
            last_wiggle = 1
        elif observation[2] < -0.1:
            action = 0
            last_wiggle = 0
        else:
            last_wiggle = 1 - last_wiggle
            action = last_wiggle

        observation, reward, done, info = env.step(action)
        if done:
            print("Episode finished after {} timesteps".format(t+1))

            break
env.close()

Link to the DQN approach I found

Cartpole - Introduction to Reinforcement Learning (DQN - Deep Q-Learning)