You are on page 1of 22

Unsupervised learning

• With the help of patterns, the algorithm learn


itself is known as unsupervised learning
It is three types
• Clustering
• Dimensionality reduction
• Reinforcement learning
Clustering
Clustering falls under unsupervised learning
methods. In this, the machine is provided with a
set of unlabeled data, and the machine is
required to extract the structure from the data Just to know
from its own, without any external supervision. Four methods
It searches for similar patterns in the dataset
and then forms a cluster of such samples by
1.Density based
similar grouping attributes. In other words, 2. Hirerarchial
Clustering is a process of grouping related data based
samples, such that data points in one group are 3. Partitioning
of the same kind but are different from the data 4..Grid
points in another group. An assumption is made,
and based on that similarity of datapoints is
constituted.
Dimensionality reduction
As humans, we can visualise the 3D only. But lot of theories and
algorithms, there exists various entities which exist beyod 3D. In
NLP, we have lot of dimension which we cannot visualise because its
beyond.
If you click this picture of ball,it transforms to 2 dimension. But same
ball its there in your hands, it will become 3 dimension. As soon as
you reduce one dimension, at least 50% of the information is lost.
Ex of dimensionality:
• Simple email classification –email spam or not
• Features of email
• Large number of features
• Generic title
• Content of the email
• Template of the email
• Features of overlap
Reinforcement learning
• Actually learning from mistakes
• Ex: Video games. Complete the level and
earn a badge
• Else play the game again.
• This is to accomplish a particular goal or
improve performance on a specific task.
As the agent takes action that goes
towards the goal, it receives a reward.

• This technique especially useful for training


robots which makes a series of decisions in
tasks like steering autonomous vehicle or
managing inventory in a ware house.
• Ex: chess games using robots, self driving cars
AI-Neural networks
WALT:
To introduce the basics of neural network
To understand the working of neural network
Reading time and discussion

• What is neural network?


• How is it connected human brain?
• Why is the neural network used?
brain
Given are the images of a Human Neuron and its relation with the Neural
Network. The axon from a neuron sends an impulse to the synapse of another
neuron. The impulse received is then sent to the cell body (nucleus) through
dendrites. The cell body performs an activation function on the impulse received
and then gives it to the output axon which passes the same to the next neuron in
the system. Now as we relate this process with an Artificial Neural Network, we
can see that the input layer gets data which is passes on to the nodes in the
hidden layer. The nodes perform specific actions on the data and pass the
processed information to the next layer. In the end, the final processed data
reaches the output of the system.
Artificial Neural Network Human Brain and Nervous System
Input layer Axon terminals
Nodes Axon(impulses carried away from and
towards to cell body)
Output layer Nucleus(cell body)
Functioning of neurons in Human nervous system

Dendrites:
Responsible for receiving the information from other
neurons
Soma:
cell body of the neuron and responsible for processing the
information
Axon:
just like a cable through which neurons send information.
Synapses:
connection between the axon and other neuron.
Neural networks -defined
Neural networks are loosely modelled after how
neurons in the human brain behave. The key
advantage of neural networks are that they are able
to extract data features automatically without
needing the input of the programmer. A neural
network is essentially a system of organizing
machine learning algorithms to perform certain tasks.
It is a fast and efficient way to solve problems for
which the dataset is very large, such as in images.
Compare the
Biological nn and artificial nn

Biological neural Artificial Neural


network Network
Soma Node
Dendrites Input connection
Synapse Weights or
Interconnections
Axon Output/input information
Compare the Biological neural network
and artificial neural network
CRITERIA BNN ANN

PROCESSING Parallel, slow but is superior Parallel, fast but is inferior


than ANN than BNN

LEARNING Anbiguity can be tolerated Structured, precise data is


required to tolerate ambiguity

FAULT TOLERANCE Performance degrades even Has the potential to be fault


with parital change to the tolerant due to the capability
system of robust performance
STORAGE Stores the information Stores the information in
contagious memory
features of neural network
Advantages of neural network

Storing information on the entire network


information stored in the entire network not in a database. The
disappearance of a few pieces of information in one place does not
restrict the network.
Ability to work with inadequate knowledge:
After ANN, the data may produce output with incomplete
information. So lack of performance may appear.
It has fault tolerance
corruption of one or more cells of ANN does not prevent it from
generating output. This feature make the network fault tolerant.
Ability to train machine:
learns events and make decisions by commenting on similar events
Parallel processing ability:
These have numerical strength that can perform more than one job
at the same time.
disAdvantages of neural network

Hardware dependence:
require processors with parallel processing power, by their
structure.
Unexplained functioning of the network:
this is the most important problem of ANN. When it gives a probing
solution, it does not give a clue as to why and how.
Assurance of proper network structure:
There is no specific rule for determining the structure of artificial
network, It is achieved the result of trial and error method.
Difficulty of showing the problem to the network:
ANNs can work with numerical information. Problems have
translated into numerical values before being introduced to ANN.
Applications of neural network
• Fraud detection – Credit card fraud, bomb detection
• Speech recognition- humans are still need to difficult to learn the
different languages. ANN will be installed with many languages which
helps here
• Pattern recognition- Automatic recognition of handwritten
characters,either letters or digits.
• Human face recognition- one of the biometric methods. It is a typical
task.
• Monitoring robotic factories- it controls the machinery settings, adjust
temperature.
• Marketing-NN are well equipped to carry this out by segmenting
customers according to basic characteristics including
demographics,location,economic status.
• Banking and finance- NN have been applied successfully to problems like
future price forcast, exchange forecast and stock performance.
• Medicine- modelling the parts of the human body and recognising
diseases.
Deep neural network
Key concepts of deep neural networks:
Earlier versions of neural networks such as the perceptrons were
shallow, composed of one input and one output layer and at most
one hidden layer in between. If more than three layers qualifies as
deep learning.
A deep neural network (DNN) is an artificial neural network (ANN)
with multiple layers between the one input and output layers.
There are different types of neural networks but they always
consist of the same components: neurons, synapses, weights,
biases, and functions.
UNDERSTAND ABOUT IT….
How they (Neural Networks learn)
• There are two methods for training ANN
depending on the problem to solve
• Self organising learn – exposed to large amounts of
data and tends to discover patterns and relationship in
that data.
• Back propagation ANN- It is trained by humans to
perform specific tasks. During this training, the
teacher evaluates whether the ANN output is correct.
If its correct, the neural weightings that produce the
output are reinforced. If the output is incorrect, those
weightings responsible are diminished

You might also like