This presentation describes various categories of machine learning techniques.It starts with importance of Machine learning and difference between ML and traditional AI. Examples and in-depth explanation of different learning techniques in ML.
2. Machine Learning
Supervised, Unsupervised and
Reinforcement Learning
A. What is Machine Learning?
B. Applications of Machine Learning
C. Supervised Learning
D. Unsupervised Learning
E. Reinforcement Learning
F. Conclusion
7. Machine Learning
Supervised, Unsupervised and
Reinforcement Learning
A. What is Machine Learning?
B. Applications of Machine Learning
C. Supervised Learning
D. Unsupervised Learning
E. Reinforcement Learning
F. Conclusion
9. Machine Learning
Supervised, Unsupervised and
Reinforcement Learning
A. What is Machine Learning?
B. Applications of Machine Learning
C. Supervised Learning
D. Unsupervised Learning
E. Reinforcement Learning
F. Conclusion
10. Supervised Learning
Learning from the association between Input and Output
Input:
Output:
1 3 4 7
1 9 16 49
10
?
F(x) = x2
(Function Approximation)
18. Machine Learning
Supervised, Unsupervised and
Reinforcement Learning
A. What is Machine Learning?
B. Applications of Machine Learning
C. Supervised Learning
D. Unsupervised Learning
E. Reinforcement Learning
F. Conclusion
20. Unsupervised Learning
We derive the structure from input by just looking at relation between
Input themselves
Finding structure within the data without labels
Low earning group
High earning group
21. What we try to Achieve
We try to organize something
Example: We have 20 million data items and we want to group them
We make up some sort to criteria to find a cluster
that scores well
23. Machine Learning
Supervised, Unsupervised and
Reinforcement Learning
A. What is Machine Learning?
B. Applications of Machine Learning
C. Supervised Learning
D. Unsupervised Learning
E. Reinforcement Learning
F. Conclusion
27. Reinforcement Learning
• Feedback after several steps
• We try to find the behavior which scores well
• Computation happens within the agent.
• No idea about the environment beforehand
• Learns about the environment through interaction with
the environment
28. Conclusion
Supervised Learning
• Learning through delayed feedback by interacting with environment
Reinforcement Learning
Unsupervised Learning
• We are trying to find association between input values and grouping them
• We are performing function approximation based on input and output values