Scribd Logo
Upload

Welcome to Scribd!

  • 23 views

    Immediate Reward Discount

    Uploaded by

    John Green
    The document discusses reinforcement learning and how an agent learns to take actions. It explains that when an agent takes an action in a state, it receives an immediate reward but the maximum future reward is discounted by γ to account for delayed rewards. The agent is trained to aim for maximum reward in the least time. A neural network is created to take states as input and output Q(s,a) values to choose the highest valued action.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Immediate Reward Discount

    Uploaded by

    John Green
    23 views1 page
    The document discusses reinforcement learning and how an agent learns to take actions. It explains that when an agent takes an action in a state, it receives an immediate reward but the maximum future reward is discounted by γ to account for delayed rewards. The agent is trained to aim for maximum reward in the least time. A neural network is created to take states as input and output Q(s,a) values to choose the highest valued action.

    Original Description:

    learning

    Original Title

    Ret

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document discusses reinforcement learning and how an agent learns to take actions. It explains that when an agent takes an action in a state, it receives an immediate reward but the maximum future reward is discounted by γ to account for delayed rewards. The agent is trained to aim for maximum reward in the least time. A neural network is created to take states as input and output Q(s,a) values to choose the highest valued action.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    23 views1 page

    Immediate Reward Discount

    Uploaded by

    John Green
    The document discusses reinforcement learning and how an agent learns to take actions. It explains that when an agent takes an action in a state, it receives an immediate reward but the maximum future reward is discounted by γ to account for delayed rewards. The agent is trained to aim for maximum reward in the least time. A neural network is created to take states as input and output Q(s,a) values to choose the highest valued action.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 1

    immediate reward

    discount
    the maximum future reward coming to the agent if it takes action a in state s.
    However, this value is discounted by 'γ' to take
    into account that it isn’t ideal for the agent to wait
    the learning rate during the updating
    forever for a future reward – it is best for the agent
    delayed reward
    to aim for the maximum award in the least period
    of time.

    a neural network is created which takes


    the state s as its input, and then the network is
    trained to output appropriate Q(s,a) values for
    each action in state s.

    The action a of the agent can then be chosen by


    taking the action with the greatest Q(s,a) value

    You might also like