You are on page 1of 8

Q Learning in Machine Learning

Presented by,
Shobitha AS
R22DE138
INTRODUCTION TO Q LEARNING

What is Q Learning?
• Q Learning is a reinforcement learning algorithm.
• That aims to find the optimal action-selection policy for an agent in a
Markov decision process (MDP).
• It learns by iteratively updating the Q-values of state-action pairs based
on the rewards received from the environment.

2
IMPORTANCE OF Q LEARNING

Q Learning is important in enabling agents to learn and make decisions in


complex environments. It allows agents to adapt to changing situations
and make decisions based on the expected outcomes of their actions. This
is particularly useful in applications such as robotics, where agents need to
navigate and interact with their environment in a dynamic and uncertain
way.

3
THE Q LEARNING ALGORITHM
The Q Learning algorithm is a reinforcement learning technique used in machine
learning. It is a model-free approach that allows an agent to learn optimal actions in a
given environment through trial and error.
The algorithm works by maintaining a Q-table, which stores the expected rewards for
each state-action pair. The Q-table is updated iteratively based on the agent's
interactions with the environment.
At each step, the agent selects an action based on the current state and the values in the
Q-table. The selected action leads to a new state and the agent receives a reward. The
Q-table is then updated using the following formula:
Q(s, a) = Q(s, a) + α * (R + γ * max(Q(s', a')) - Q(s, a))

4
THE Q LEARNING ALGORITHM
Where

5
APPLICATIONS OF Q LEARNING
Game Playing Robotics Resource Finance
Management
Q Learning has been Q Learning can be Q Learning can be
successfully applied used in robotics to Q Learning can be used in finance for
to game playing, train autonomous applied to optimize portfolio
such as in the famous agents to navigate resource allocation management,
case of AlphaGo, and perform tasks in and management in algorithmic trading,
where it was used to dynamic various domains, and risk assessment.
train the AI agent to environments. It such as energy It can learn optimal
play the game of Go enables the robot to management, traffic trading strategies
at a superhuman learn from trial and control, and based on historical
level. error and make inventory data and market
decisions based on management. It helps conditions.
rewards and in making efficient
penalties. decisions to
maximize rewards
and minimize costs. 6
Conclusion
Q Learning is a powerful algorithm for training agents to
make decisions in complex environments. However, it is not
without its challenges and limitations. By understanding
these challenges and limitations, researchers and
practitioners can develop more effective strategies for using
Q Learning in real-world scenarios.

7
THANK YOU

You might also like