ARTIFICIAL INTELLIGENCE
Artificial intelligence is concerned with how to make computers do things at which, at the
moment, people are better.
This is the ability of a digital computer or computer-controlled robot to perform tasks
commonly associated with intelligent beings.
Artificial intelligence involves developing systems that are able to carry out intellectual
processes characteristics of humans, such as the ability to reason, discover meaning, generalize,
or learn from past experiences.
Categories of AI
Narrow AI: This is when a machine has superior performance to a human when
doing one specific task.
General AI: This is when a machine is similar in its performance to a human in any
intellectual task.
Strong AI: This is when a machine has superior performance to a human in many
tasks.
Purpose and Structure of graphs
A graph is a collection of nodes or vertices between which there can be edges. Each node has a
name. An edge can have an associated label which is a numerical value. An example is
presented below;
Page 1 of 18
A graph can be used to represent a variety of scenarios. One common representation is that the
nodes represent places and the edge labels represent the distances between those places.
Edges are only included in the graph when there is a route available for direct travel between
the pair of nodes. Such graphs can, for example, find the shortest route between two nodes
that are not adjacent to each other.
•
When two nodes are connected by an edge, they are called neighbours
•
The degree of a node is the number of other nodes that it is connected to (i.e. the
number of neighbours that it has)
•
A loop is an edge that connects a node to itself
•
A path is a sequence of nodes that are connected by edges
Page 2 of 18
•
A cycle is a closed path, i.e. a path that starts and ends at the same node (and no node is
visited more than once)
A graph can be used to represent a wide variety of complex data relationships. For
example:
•
•
•
•
Social networking: the nodes could be individual people. An edge could
represent that two people are acquaintances.
Transport networks: the nodes could be towns. An edge could represent that a
train line connects two towns.
The internet: the nodes could be routers. An edge could represent that two
routers are connected.
The World Wide Web: the nodes could be webpages. An edge could represent
that the pages are linked.
TYPES OF GRAPHS
❖
UNDIRECTED GRAPHS:
An undirected graph allows you to move (traverse) in either direction between nodes.
Below is a diagram of an undirected graph. The edges are simple lines, not arrows.
The graph has 7 nodes and 11 edges. Dunwich is a neighbour of both Blaxhall and
Harwich. Clacton has 3 edges directly connecting it to 3 other vertices; it therefore has a
degree of 3.
There are several paths from Dunwich to Clacton. Examples are:
•
•
Dunwich → Harwich → Clacton
Dunwich → Blaxhall → Harwich → Clacton
Dunwich → Blaxhall → Harwich → Dunwich is a cycle.
Page 3 of 18
❖ DIRECTED GRAPHS (DIGRAPHS)
Here, the edges have direction, which means that you move between nodes in a
specified direction. In diagrams, arrows are used (instead of lines) to represent the
edges. If the edge is bidirectional, two arrows are used (although you may sometimes
see versions with double-headed arrows).
In the road map example, most roads will be bidirectional; however, there will also be
one-way streets.
Page 4 of 18
❖ WEIGHTED / LABELLED GRAPHS
This is a graph can have values associated with the edges. Weighted
graphs can be either directed or undirected.
Weights can be used to record information relating to the edges. For example, in
a mapping application, you may want to record distance between towns, or the
time needed to travel between one town and another. Weights will enable you
to find the shortest path between nodes.
Page 5 of 18
Weighted, Undirected graph
Weighted, Digraph
Page 6 of 18
We could use our intelligence to find the shortest route (using the first graph above) between
node A and node G by considering all of the possible routes and calculating the overall distance
for each route.
For A to B to C to D, overall distance is 40 + 10 + 40 = 90
For A to B to F to E to D, overall distance is 40 + 15 + 20 + 5 = 80, which is the shortest
For A to F to E to D overall distance is 60 + 20 + 5 = 85.
For A to F to B to C to Z overall distance is 60 + 15 + 10 + 40 = 125.
For each graph containing 100 nodes, this could be quite time consuming.
Fortunately, a number of artificial intelligence algorithms have been developed to solve this
type of problem.
DJIKSTRA’S ALGORITHM
Dijkstra’s algorithm has one motivation: to find the shortest paths from a start node to all
other nodes on the graph.
•
•
The cost of a path that connects two nodes is calculated by adding the weights
of all the edges that belong to the path.
The shortest path is the sequence of nodes, in the order they are visited, which
results in the minimum cost to travel between the start and end node.
When the algorithm has finished running, it produces a list that holds the following
information for each node:
•
•
•
The node label
The cost of the shortest path to that node (from the start node)
The label of the previous node in the path
Using the information in this list you can backtrack through the previous nodes back to
the start node. This will give you the shortest path (sequence of visited nodes) from the
start node to each node and the cost of each path.
Forexample:
1)
Page 7 of 18
The graph above illustrates the connections between five nodes. Using Dijkstra's algorithm, find
the shortest path from the start node A to all other nodes.
2) Given the following set of information produced by running Dijkstra's algorithm, in the
form of a visited list, what is the route of the shortest path from A to G?
Node
A
C
F
E
B
D
G
Visited list
Cost (from start)
0
3
9
10
12
13
15
Assignment:
Page 8 of 18
Previous
none
A
C
F
A
E
E
Carryout researches and make short notes on the following;
a) Writing Dijkstra’s algorithm using structured English
b) Writing Dijkstra’s algorithm using code
c) Limitations of the Dijkstra’s algorithm
A* ALGORITHM
This is a searching algorithm that searches for the shortest path between the initial and final
stage.
A* algorithm has 3 parameters:
-
(g): It is the cost of moving from the initial cell to the current cell. Basically, it is the sum
of all the cells that have been visited since leaving the first cell.
-
(h): Heuristic value: It is the estimated cost of moving from the current sale to the final
sell. The actual cost cannot be calculated until the final sale is reached. Hence h is the
estimated cost.
NB: We must make sure that there is never an overestimation of the cost.
-
(f): It is used to find the least cost from one node to another.
-
{ f = g + h } : It is responsible for finding the optimal path between source and
destination.
The way they are going to make this decision is by taking the F value into account. The
algorithm selects the smallest value cells and moves to that cell. The process continues until the
algorithm reaches its goal cell.
Page 9 of 18
Example:
Find the shortest path from A to F using A* algorithm using the chart below
Assignment:
Carryout researches and make short notes on the following;
a) Writing A* algorithm using structured English
b) Writing A* algorithm using code
c) Comparing the A* and Dijkstra’s algorithm
MACHINE LEARNING
Page 10 of 18
When machine learning, systems learn without being programmed to learn.
The requirements for machine learning can be summarized as follows
1. Database system has a defined task or tasks to perform.
2. Knowledge is acquired through the experience of performing the tasks
3. As a result of this experience and knowledge gained, the performance of future taxes
improved.
a) SUPERVISED LEARNING
The system is fed knowledge with associated classification.
For example, an AI program might be under development for making exam
paper questions. In the supervised learning, answers to examination questions
could be provided together with a grade for each one or with categorized
comments.
A special case of supervised learning is where an expert system is being
developed.
An expert system, always focus always has a focus on a narrowly defined domain
of knowledge. In this case, human experts are given samples of data requiring
analysis.
Page 11 of 18
The experts provide the conclusions to be drawn from the data. The data and
conclusions are input into and input to the knowledge base.
The effectiveness of the system can be tested by a human expert providing
sample data and checking the accuracy of the conclusions provided by the expert
system.
If performance is poor, then further data and conclusions are input into the
system.
,n jjj
b) UNSUPERVISED LEARNING
The system has to draw its own conclusions from its experience of the results of
the tax it has performed, for these algorithms are needed that can organize or
categorize the knowledge acquired.
An example is where conceptual clusters are identified which are based on a
hierarchical framework.
In this approach, the knowledge is initially all placed in the root of a tree
structure.
Then, depending on attributes of the knowledge, selected groups are moved into
branches of the tree.
Page 12 of 18
Nowadays, unsupervised learning is a dominant activity. Powerful computer
systems having access to massive data banks are regularly used to make
decisions based on previous actions recorded.
We all have our activity on the World Wide Web recorded and stored.
This stored data is then used to make decisions about what products or services
should be recommended to us in future internet use.
Systems are able to identify hidden patterns from the data provided. They are
not trained using the ‘right’ answer.
c) REINFORCEMENT LEARNING
This has some features similar to supervised and unsupervised learning.
The system is not trained. It learns on the basis of ‘reward and punishment’
when carrying out an action. That is, it uses trial and error in algorithms to
determine which action gives the highest / optimal outcome.
Examples include search engines, online games and robotics.
Page 13 of 18
For example;
An agent is learning how best to perform in an environment.
The environment has many defined states.
At each step, the agent takes an action
An agent has a policy that guides its actions
The policy is influenced by the recorded history and the knowledge of the
current state of the environment.
An action changes the environment to a new state.
The agent receives a reward that is a measure of how effective the action
was in relation to the achievement of the overall goal.
The policy will guide the agent in deciding whether the next action should be
exploiting knowledge already known or exploring a new avenue.
ARTIFICIAL NEURAL NETWORKS
The neural networks in our brains provide our intelligence. It therefore seems obvious that
artificial neural networks should be considered as a foundation for AI systems.
Page 14 of 18
An artificial neural network could be created in software or hardware.
The triangles are the nodes in the network which represent artificial neurons.
In general, a node can receive one or more inputs and can provide an output to one or more of
the other nodes.
Layers of artificial neural networks
-
The Input layer: the data’s entry point into the system
The Hidden layer: where the information gets processed
The Output layer: where the system decides how to proceed based on the data.
The column of three nodes on the left receive input.
The column on the right provides output
The nodes in between form a hidden layer. Some artificial neural networks will contain several
hidden layers.
Page 15 of 18
The neural network functions via a collection of nodes or connected units, just like artificial
neurons.
These nodes loosely model the neuron network in the animal brain. An artificial neuron
receives a signal in the form of a stimulus, processes it and signals other neurons connected to
it.
The neural workings of an artificial neural network.
An artificial neuron receives a stimulus in the form of a signal that is a real number. Then;
-
The output of each neuron is computed by a nonlinear function of the sum of its inputs.
The connections among the neurons are called edges.
Both neurons and edges have a weight.
This parameter adjusts and changes as the learning proceeds.
The weight increases or decreases the strength of the signal at the connection.
Neurons may have a threshold. A signal is sent onward only if the aggregate signal
crosses this threshold.
DEEP LEARNING
Here, machines think in a way similar to the human brain. They handle huge amounts of data
using artificial neural networks.
Page 16 of 18
Backpropagation of Neural Network
This is the central mechanism by which artificial neural networks learn. It is the messenger
telling the neural network whether or not it made a mistake when it made a prediction.
To propagate is to transmit something (light, sound, motion or information) in a particular
direction or through a particular medium.
To backpropagate is to transmit something in response, to send information back upstream – in
this case, with the purpose of correcting an error.
With backpropagation in deep learning, we deal the transmission of information and that
information relates to the error produced by the neural network when it makes a guess about
data. Backpropagation is synonymous with correction.
Algorithms experience the world through data. So by training a neural network on a relevant
dataset, we seek to decrease its ignorance.
The knowledge of a neural network with regard to the world is captured by its weights, the
parameters that alter input data as its signal flows through the neural network towards the
net’s final layer, which will make a decision about that input.
Page 17 of 18
Those decisions are often wrong, because the parameters transforming the signal into a
decision are poorly calibrated; they haven’t learned enough yet.
Forward propagation is when a data instance sends its signals through a network’s
parameters toward the prediction at the end.
Once the prediction is made, its distance from the ground truth (error) can be
measured.
So, the parameters of the neural network have a relationship with the error the net produces.
When the parameters change, the error changes too.
We change the parameters using optimization algorithms such as gradient descent.
Assignment:
Make short notes on the Uses of Error Backpropagation.
Page 18 of 18