2. AI Approaches
• Ai models have two approaches:
• Rule based(eg: decision Tree)
• Learning based
– Machine learning.
– Deep Learning.
3. Rule Based Approaches
• A rule based system uses rules as the
knowledge representation .
• The rules are coded into the system in
the form of a series of if-then-else
statement that guides a computer to
reach a conclusion or recommendation.
4.
5. • The rule based system s just fake
intelligence because of the missing
learning capability.
• The rule based system is said to be
limited in its ability to simulate
intelligence.
• It is always limited by the size of its
underlying rule base.
6. • A rule based system is built on two main
components:
– A set of facts about a situation ( also
known as knowledge base. Eg: these facts
are a combination of data).
– A set of rules for how to deal with those
facts( also known as the rules engine
which describe the relation between the IF
and THEN statement.
7. Decision Tree
• A decision tree is a series of nodes.
• A directional graph that starts at the
base with a single node and extends to
the many leaf nodes that represent the
categories that the tree can classify.
• It looks like an up side down tree.
8.
9. Components of decision tree
• Question/Condition are Nodes or Roots.
• Ye/No options represents Edge or
Branches
• End Actions are leaves of the tree.
10. • Advantages of decision tree:
– Interpretable:
– Any data can be used.
• Disadvantages of Decision Tree:
– Overfitting
– Predicting continuous variable
12. Limitations of Rule based
approach
• It is tough to add rules to an already large
knowledge base without introducing
contradiction rules.
• Maintenance Time consuming and
expensive.
• Not useful for solving problems in complex
domains or across multiple simple domains.
• Some times It is not possible to explicitly
define rules in a programmatic or declarative
way.
13. Learning Based Approach
• The ability to learn causes adaptive
intelligence.
• Adaptive intelligence : Existing
knowledge can be changed or
discarded.
• Hence , the system build its rule on the
fly.
• A neural network is an instance of a
learning system.
14. • The decision whether to go for a rule-
based system or learning based system
depends on the problem you want to
solve, and its always a trade-off among
efficiency , training cost and
understanding
16. What is Machine Learning
• ML is an application of AI
• Provides the system the ability to
automatically learn and improve from
experience without being explicitly
programmed.
• ML focus on the development of
computer programs that can access
data and use it to learn.
17. • Machine Learning is a field of study that
gives computers the ability to learn without
being explicitly programmed.
_Arthur Samuel(1959)
18. • Through machine learning computer
can be trained to automate tasks that
would be exhaustive or impossible for a
human being.
• ML enables people to performs tasks
such as:
• Predicting the future
• Classifying things in a meaningful way.
• Making rational decision in a given context.
19. Difference between AI And ML
AI
Concept of machines being
able to carry out tasks in a
SMART way
Goal : simulate natural
intelligence to solve complex
problems
Decision making
Leads to develop a system
to mimic human to respond
in certain circumstances
ML
Application of AI based on the idea
that machines access data and let
them learn for themselves.
Goal : learn from data ob certain
tasks to maximize the performance
of machine on the given task.
Allows systems to learn from data
It involve creating self learning
algorithms
20. The Machine learning Models
1. Takes data and learns from what
happened before(learn from
experience).
2.It then predicts what’s going to come
next.
3.Tries and improves to find new
solutions.
21. Steps in a general Machine
Learning model.
1. Define objective
2. Collect Data
3. Prepare data
4. Select the algorithm to be used
5. Train the model
6. Test the model
7. Predict the output
8. Deploy
22.
23. Types of Machine Learning
• Supervised Learning: is a method used to enable
machines to classify/predict objects, problems or
solutions based on labelled data fed into the machine
• Un-supervised Learning: the learning model is
handed an unlabelled dataset without explicit
instructions on what to do with it.
• Semi-supervised :It requires both supervised and
unsupervised methods in order to obtain useful result
• Reinforcement learning: the required value of the
output is not known explicitly, but the system
provides feedback on the provided output.
24.
25. Supervised Learning:
• In supervised learning we train an
algorithm and at the end pick a model
that best predict well-defined output
based on the in put data.
• In SL the system receives input and
output in the beginning.
• Then based on i/p and o/p it create
appropriate rules to map the input to
output.
26.
27. Steps to solve a problem of
supervised learning
1. Determine the type of training examples.
2. Gather a fully labelled training set.
3. Determine the input feature representation
of the learned function.
4. Determine the structure of the learning
function and learning algorithm.
5. Complete the design.
6. Evaluate the accuracy of the learning
function.
28. List of common Algorithms
• Nearest Neighbor
• Naive Bayes
• Decision Trees
• Linear Regression
• Support Vector Machine(SVM)
• Neural Networks
29. Types of Supervised Learning
• Classification: Classification separates
data. The variable output is a category,
such as “Red”, “Blue” or “animal” “bird”
• Regression: a technique to reproduce
the output value. The output variable is
real value. such as “Dollar” or “weight”
33. Unsupervised Learning
• The learning model is handed an
unlabelled dataset without explicit
instructions on what to do with it.
• Then attempts are made to
automatically find structure in the data
by extracting useful features and
analysing the structure.
34.
35.
36. • Depending on the problem ,the
unsupervised learning model can
handle data in different ways:
• Clustering
• Anomaly detection
• Association rule
37. Clustering
• A clustering problem is where you want
to find the inherent groups in the data.
• Clustering works on discrete dataset.
• It is used to find similarities and
differences.
• Machine generates its own rules or
algorithms to differentiate the given
dataset.
38.
39. Anomaly Detection
• It is the identification of rare items ,
events or observations differing from the
majority of the data.
• Anomalous data can be connected to
some kind of problem or rare events.
• Eg: bank fraud, medical problems ,
malfunctioning equipments etc
42. Association rule
• Is a procedure which aims to observe
frequently occurring patterns ,
correlations, or associations from
datasets found in various kinds of
databases.
• Eg: Market based analysis
43.
44. Semi-Supervised learning:
• Is a learning process in which lots of
output values are missing(the one we
want to predict).
• It requires both supervised and
unsupervised methods in order to obtain
useful result
45. Reinforcement learning:
• the required value of the output is not
known explicitly, but the system
provides feedback on the provided
output.
• As the agent takes action that goes
towards the goal, it receives a reward.
46.
47. Deep Learning
• Deep leaning is a machine learning
technique that teaches computer to do
what comes naturally to human.
• “Learn by Examples”.
• Deep learning is a technique that
mimics the network of neurons in a
brain.
48. • Deep learning is a subset of machine
learning.
• The machine uses different layers to
learn from the data.
• The dept of the model is represented by
the number of layers in the model.
• The learning phase is done through
neural network.
49. Neural Network
• A neural network is an architecture
where the layers are stacked on top of
each other.