"Prediction of Values Using Machine Learning": Dhole Patil College of Engineering Department of Information Technology

You might also like

Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 19

Dhole Patil College of Engineering

Department of Information Technology

Project Based Seminar (Oral) Presentation


On
“Prediction of Values using Machine Learning”
By
T150608505 Shubham Derhgawen

Guide
Prof. Pooja Kadam

Day and Date of Exam : Monday, 23 April 2019


Contents
• Project Group
• Project Motivation
• Introduction
• Literature Survey
• Actual Seminar related topics
• Applications/advantages
• Conclusions
• References

Prediction of Values using Machine Learning 1


Project Details
• Project Title : Prediction using Machine Learning
• Project Domain: Machine Learning
• Project Group Members:
– T150608505, Derhgawen Shubham
– T150608504, Chatterjee Subhasish
– T150608507, Gogineni Himaja

Prediction of Values using Machine Learning 3


Motivation

• Machine learning is a trending topic today.


• Many new application domains are
discovered each day.
• Historical data can help us predict future
data which is very valuable.
• Fields like Artificial networks are bridging
the gap between human and computers.

Prediction of Values using Machine Learning 2


Problem statement
• Prediction of values depends on data.
• Thus understanding the data is the first hurdle.
• Then gaining insightful information from the data is a
problem solve able by machine learning algorithms.
• Again finding which algorithm works best and on what
basis should a algorithm be judged.

Prediction of Values using Machine Learning 5


Aim and Objectives of the Project
• The aim of this project is to improve understanding of
machine learning algorithms and their performance in
predicting values from historical data.
• Objectives:-
• To understand the algorithms we will refer previous
implementations of them.
• To improve performance we will implement the
algorithms on our own to find any missed out
optimizations.

Prediction of Values using Machine Learning 6


Project and Seminar Topics
• Project Title : Prediction of Values using
Machine Learning
• Shubham Derhgawen :Prediction using Artificial
Neural Networks.
• Subhasish Chatterjee: Prediction using Decision
Trees.
• Himaja Gogineni: Prediction using Linear
Regression.
Prediction of Values using Machine Learning 7
Introduction
• Artificial Neural Networks were originally
developed by researchers trying to mimic the
neurophysiology of the human brain.
• The feedforward network is the simplest and
most popular type of neural network.
• Backpropagation is the most common method
for computing the error gradient for a
feedforward network.

Artificial Neural Network 2


• Artificial Neural Networks can be implemented
as MLP(Multi-layer perceptron).
• A perceptron is a mathematical equivalent of a
neuron in human brain.
• A simple perceptron is not intelligent but when
connected in a network can be used for
nonlinear regression.

9
Perceptron
 The perceptron is an
algorithm.
 It is a type of linear
classifier.
 If the weighted sum is
less than or equal to our
threshold, or bias, b,
then our output will be
0.
 If the weighted sum is
greater than our
threshold, or bias, b,
then our output will be
1.
Multilayer Perceptron(MLP)
• Neural networks perform well in applications
when the functional form is nonlinear.
• Backpropagation is used by the gradient
descent optimization algorithm to adjust the
weight of neurons by calculating the gradient of
the loss function.
• The number of hidden layers applied is
determined by estimating the generalization
error of each network. For the current
application to electricity energy consumption
data, it is found that one hidden layer network is
adequate.

12
Literature Survey
• Neural Networks in Practice
• Neural networks in medicine
• Modelling and Diagnosing the Cardiovascular System
• Electronic noses
• Neural Networks in business
• Marketing
• Credit Evaluation
Prediction using Artificial Neural Networks. 13
Literature Survey/Related work

Sr. No. Reference Name
Minimum FIVE referencesSeed Idea/ Work description
should Problems found
be there from literature. Any other criteria

1 Predicting Predicting electricity energy The historical


electricity energy Consumption of Hong Kong data had to be
consumption: A divided because
comparison of of the inherent
feature of the
regression
data
analysis, decision
tree and neural
networks

4/14/2019 SCOE, Information Technology (2014-15) 14


Contents of the Seminar
• Actual seminar contents can be of 5 to 6 slides
• Make use of multimedia technology for
presenting your ideas
• Avoid using abbreviations
• Explanation of concepts using animations are
welcome
• Choose good design for your presentation
• Make slides content ‘justfy’
Not more than 5 to 6 slides
Seminar Title Goes here 15
Applications
• Highlight some of the areas where your seminar
topic has a good shadow/effect in your project.
• Explain few applications in details if time
permits
• Show live demonstration of any application if
required using internet/executable

Only one slide


16
Seminar Title Goes here
Conclusions
• In this seminar we studied….
• This study has found that generally .......
• This work would be implemented in my final
year project…

17
References
List all the material used from various sources for making this seminar
[1] Journal article – A. A. Author of article. "Title of article," Title of Journal,
vol. #, no. #, pp. page number/s, Month year.
[2] Books- Author's last name, first initial. (Publication date). Book title.
Additional information. City of publication: Publishing company.
[3] Magazine - Author's last name, first initial. (Publication date). Article
title. Periodical title, volume number(issue number if available),
inclusive pages
[4] Website or Webpage Author's name. (Date of publication). Title of
article. Title of Periodical, volume number, Retrieved month day, year,
from full URL
one or two slides
Seminar Title Goes here 18
Question & Answer

Seminar Title Goes here 19

You might also like