This presentation deals with the basics of AI and it's connection with neural network. Additionally, it explains the pros and cons of AI along with the applications.
AI Explained: Understanding Artificial Intelligence and its Applications
1.
2.
3.
4. Artificial Intelligence is the science and
engineering of making intelligent machines, especially
intelligent computer programs. It is related to the
similar task of using computers to understand human
intelligence, but AI does not have to confine itself to
methods that are biologically observable.
5. When a program makes observations of some kind, it
is often programmed to compare what it sees with a
pattern. For example, a vision program may try to
match a pattern of eyes and a nose in a scene in order
to find a face. More complex patterns, e.g. in a
natural language text, in a chess position, or in the
history of some event are also studied. These more
complex patterns require quite different methods
than do the simple patterns that have been studied
the most.
6. From some facts, others can be inferred.
Mathematical logical deduction is adequate for
some purposes.
Programs do that. The approaches to AI based on
connectionism and neural nets specialize in that.
There is also learning of laws expressed in logic. [
Mit97] is a comprehensive undergraduate text on
machine learning. Programs can only learn what facts
or behaviors their formalisms can represent, and
unfortunately learning systems are almost all based
on very limited abilities to represent information.
7. Genetic programming is a technique for getting
programs to solve a task by mating random Lisp
programs and selecting fittest in millions of
generations
AI programs often examine large numbers of
possibilities, e.g. moves in a chess game or inferences
by a theorem proving program. Discoveries are
continually made about how to do this more
efficiently in various domains.
8. Jobs - Depending on the level and type of intelligence these
machines receive in the future, it will obviously have an effect
on the type of work they can do, and how well they can do it
(they can become more efficient). As the level of AI increases
so will their competency to deal with difficult, complex even
dangerous tasks that are currently done by humans.
They don't stop - As they are machines there is no need for
sleep, they don't get ill , there is no need for breaks or
facebook, they are able to go, go, go! There obviously may be
the need for them to be charged or refueled, however the
point is they are definitely going to get a lot more work done
than we can.
9. No risk of harm - When we are exploring new undiscovered
land or even planets, when a machine gets broken or dies,
there is no harm done as they don't feel, they don't have
emotions. Where as going on the same type of expeditions a
machine does, may simply not be possible or they are
exposing themselves to high risk situations.
Act as aids - They can act as 24/7 aids to children with
disabilities or the elderly, they could even act as a source for
learning and teaching. They could even be part of security
alerting you to possible fires that you are in threat of, or
fending off crime.
10. Their functions are almost limitless - As the machines will
be able to do everything (but just better) essentially their use,
pretty much doesn't have any boundaries. They will make
fewer mistakes, they are emotionless, they are more efficient,
they are basically giving us more free time to do as we
please.
11. •Over reliance on AI - As you may have seen in many films such
as The Matrix, iRobot or even kids films such as WALL.E, if we
rely on machines to do almost everything for us we become
very dependent, so much so they have the potential to ruin our
lives if something were to go wrong. Although the films are
essentially just fiction, it wouldn't be too smart not to have
some sort of back up plan to potential issues on our part.
•Human Feel - As they are machines they obviously can't
provide you with that 'human touch and quality', the feeling of
a togetherness and emotional understanding, that machines
will lack the ability to sympathize and empathize with your
situations, and may act irrationally as a consequence.
23. Once an architecture has been selected and Input
signals are prepared then the next step is to train the
network. To start the training process initial weights
are chosen randomly. Neural Network can be trained
in two ways-
24. Learning is the process by which the
free parameters of neural network are adapted and
then simulated. There are five types of learning
processes-
25. (a)Decide the network architecture according to the problem
(b)Decide number of Input nodes
(c)Decide number of Output nodes
(d)Prepare training set. The training set must contain many
examples so that the network becomes familiarize with the
given problem
(e)If network training is supervised provide the network with
the desired output for input vectors
(f)Train the network using input vectors
(g)Finally test the network; If the network fails to provide the
desired output, then, repeat the above procedure until
optimal solution has been achieved
26. (1)Inheritably, massively parallel (Multi-process)
(2)It is designed to be adaptive
(3)It needs less effort for the characterization of the problems
(4)Artificial neural network cannot be programmed. It learns by its
previous examples
(5)Artificial neural network is robust in nature i.e. it can operate even if
the portions of the given problems are incorrect
(6)Artificial neural network may be fault tolerant because of parallelism
(7)Artificial neural network exhibits mapping capabilities i.e. they can
map their input patterns to their associated input patterns
(8)Neural network has the capability to generalize the input
(9)They can predict new outcomes from the past problems
(10)The neural network can process information simultaneously at high
speed and in a distributed manner
27. (1)No clear rules or guidelines can be defined in designing artificial
neural network
(2)There is no general way to access the internal operations of artificial
neural network
(3)Training may be difficult in an artificial neural network but it is not
impossible
28.
29. It attempts to predict the movement of stock from
the previous data by using linear model
It is used for comparing signature with the already
stored signature like in banks
(i) Toys- With the ability of neural network chip the
system is designed to recognize simple entities. For
example, simple commands like stop and go
30. (ii) Pen PCs- By Pen PCs one can write on a tablet. The
writing can be recognized and further translated into
text with the help ASCII codes
To monitor the state of air-craft by monitoring
vibration levels etc. Early warnings of the engine
problem can be predicted previously
It is used to improve marketing mail shots. In this
technique a test mail shot is run and it recognizes the
pattern of these mail shots. Finally, predictive
mapping of data is done and next mail shot is run
31. Bibliograph
y-
>> Neural Network
-S. Rajeshkaran >> www.wikipidea.com
>> Applications of AI
-John McCarthy
>> www.google.com
>> www.cnn.com
>> Call Centers of the Future
-L Venkata Subramaniam
>> www.infobarrel.com