Hyderabad, , India
ElectricalEngineering,ComputerKnowledge,EnglishLanguage
0 टिप्पणी करें | 1 राय | 18 अक्तूबर 18  | Timmaraju Sai Saurabh
Project on Game Bot using Reinforcement Learning
Reinforcement Learning is a popular and intuitive way of solving problems wherein an agent has no prior knowledge about the environment and owns two characteristics: trial-and-error and delayed rewards to improve its gameplay. A Game Bot is a Reinforcement Learning agent that derives an optimal policy by directly interacting with the environment and getting information about it in the form of rewards and penalties. The bot runs on Flash and Atari environments provided by the OpenAI (Artificial Intelligence) Gym and Universe software platforms. Game Bots help demonstrate Reinforcement Learning and Q-Learning techniques and can learn to master a game without explicitly being told how to play the game. The games are action limited and use a convolution neural network for processing image inputs and fully connected layers for estimating actions according to the input where the idea of taking action is based on Q-learning (model-free reinforcement learning), yet modified it for our policy and usage. We try to maximize the score and the screen is RGB (Red Green Blue) which means the agent has to process it to understand it.
  • अन्य प्रासंगिक दस्तावेज

  • अन्य प्रासंगिक छवियां


    • इस ब्लॉग के लिए सामाजिक शेयर

पोर्फोलिओ और ब्लॉग
Timmaraju Sai Saurabh विभिन्न कंपनियों का अनुसरण करता है, ये कंपनियां और नियोक्ता Timmaraju के फिर से शुरू देख सकते हैं
सबसे अच्छा नौकरी के अवसर पाने के लिए अपना फिर से शुरू करें अपलोड करें

मुफ्त रजिस्टर करें!