Elon Musk

CEO of Tesla & SpaceX

AI is a fundamental risk to the existence of human civilization.

 I, Robot
2004 ‧ Sci-fi/Action
 ?
?
What is AI?

?
Recognition Letter A

Procedure to recognition the image is Letter “A”

 Read the image pixel

 Make the logic

 Return result
Take a Deep Look

Letter “A” make by three lines

Different Front
Style Letter A
Baby Learn pattern to detect Object.

 So is your programming login is capable to

detect these type of letter?

 If No, then what is the lacking for your

logic?

 What should be add to detect object

dynamically?
 Design Dynamic Model

Programming Logic Dynamically Recognition

able, Self-learning Model.


What is exactly Artificial
Intelligence?

Artificial Intelligence is a
model/procedure/tool who has
capability for self learning,
dynamically detect the pattern/object
and take decision by own knowledge
just like human brain.

“So according to the definition, is it proved that AI is

really threat for human existence?”
Human Intelligence Curve
 Human Brain Learning Process

Input Image Feature Extraction Learning

Human Brain Neuron

Recognition
Artificial Neural Network vs human brain Neural
 Learning Vs Recognition

Learning Recognition
Learning is a search According to the
through the space of training dataset
possible hypotheses for learning process is
one that will perform performed and engine
well, even on new is updated. By pass
examples beyond the through the input
training set. To sample over the engine
measure the accuracy and it will return an
of a hypothesis we give output according to the
it a test set of examples learning accuracy.
that are distinct from
the training set.
 Humans recognize the
scene around them through
their eyes which create
images of that world.
AI Hierarchy
Machines are even
better at pattern
Image
Computer Vision falls
AI recognition because it Humans have the
under the symbolic
Processing
way for computers
can use more data
and dimensions of
ability to see patterns
to process such as grouping of
data. like objects.
information.

Symbolic Machine Pattern

The human brain is a network of
Learning Humans can
Learning Recognition
neurons and we use these to learn
Humans can Much of speech ANN’s are more
things if we replicate the structure
understand their see with their and function of the human brain
recognition is complex & deeper,
we might be able to get cognitive
environment and eyes & process we use those to
statistically based capabilities in machines.
move around fluidly. what they see. learn complex thing

Computer Statistical Deep

Robotics ANN
Vision Learning Learning
Humans can
To replicate the human
speak & listen to Human can brain if we get the
communicate write & read text network to scan images
Neural Network to
through language. in a language from left-right, top-
remember a
bottom.
limited past

Speech
NLP CNN RNN
Recognition

With accomplished
by CNN &
computer vision.

Object
Recognition
 Artificial Neural Network (ANN)
ANN, is a group of multiple perceptrons/ neurons at each
layer. ANN can be used to solve problems related to:
 Tabular data

 Image Data

 Text Data

ANN Application:
 Image Recognition

 Pattern Recognition

 Natural Language Processing

 Text to Speech
 Recurrent Neural Network (RNN)
RNN is a class of artificial neural networks where
connections between nodes form a directed graph along a
temporal sequence.
A looping constraint on the hidden layer of
ANN turns to RNN.

ANN can be used to solve problems related to:-

 Time Series Data
 Text Data
 Audio data

ANN Application:-
 Speech Recognition
 Text Processing(Chatbot)
 Face detection, OCR Applications as Image Recognition
 Music composition
 Convolution Neural Network (CNN)
A Convolutional Neural Network (ConvNet/CNN) is a
Deep Learning algorithm which can take in an input image,
assign importance (learnable weights and biases) to
various aspects/objects in the image and be able to
differentiate one from the other.

CNN can be used to solve problems related to:

 Image Data

CNN Application:
 Image Recognition

 Face Recognition

 Image Classification
 Reinforcement Learning
Reinforcement Learning(RL) is a type of machine learning
technique that enables an agent to learn in an interactive
environment by trial and error using feedback from its
own actions and experiences.
Application:
 Robot deciding its path

 Next move in a chess game

 Supervised Learning
Supervised Learning use of labeled datasets to train
algorithms that to classify data or predict outcomes
accurately. As input data is fed into the model, it adjusts
its weights through a reinforcement learning process,
which ensures that the model has been fitted
appropriately.
The model first learns from the given training data. The
training data contains different patterns, which the model
will learn.

Application:
 classifying spam in a separate folder from your inbox
 Image- and object-recognition
 Predictive analytics
 Unsupervised Learning
Unsupervised learning has no training phase; instead, the
algorithm is simply handed a dataset and uses the
variables within the data to identify and separate out
natural clusters.

Application:
 Finding customer segments
 Feature selection