Loading presentation...

Present Remotely

Send the link below via email or IM


Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.


Neural Networks

No description

Devanjith Ganepola

on 9 November 2015

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of Neural Networks

Overview on Artificial Neural Networks
Technical Details
Current Real Life Applications
Future Revelance and Possiblities
Impact on Current/Future Society and Environment
ideas of ANNs were around before the creation of computers
it was that the learning method for ANNs would follow Hebbian Theory
first artifical neuron made in 1943 Warren McCulloch and Walter Pitts
next major thing was the created of the Perceptron
innovation came to halt due to a book published by Minsky and Papert
research still continued by some scientist
ART (Adaptive Resonance Theory) network by Steve Grossberg
back-propagation learning method by Paul Werbos in 1974
Neocognitron (handwritten character recognition) by Kunihiko Fukushima in 1975
the most likely frontier is medical analysis and AI development
overall increase in the standing of technology
five main problems with the current technology
model robustness - ability to adapted to any type of input
model transparency - ability to understand the thought process
knowledge extraction - ability to properly understand and analyze the given problem
model extrapolation - ability to predict well outside the range of its current data
model uncertainty - ability to determine where the system went wrong and by how much
Future Revelence and Possibilities
Technical Details
Artificial Neural Networks
By: Devanjith Ganepola
electronic replicas of the biological brain
biologically: dendrites to neurons to axon to synapse to other neurons
electronically: input to nodes/neurons to output
simple neuron: non weighted inputs
follows 'fire set' or 'non fire set' first
then follows the closest match
complex neuron: weighted inputs
only fires if the sum of the inputs is greater than the threshold value
X1W1 + X2W2 + X3W3 + ... XnWn > T
complex one is used more because it replicates the real brain and is more flexible
Impact on Current/Future Society and Environment
Current Real Life Applications
should be an overall benefit to society (ex. self driving cars)
ANNs come with the evrionmental problems of computers (ex. mineral extration, product production, and product disposal)
currently an unsustainable system
must become sustainable, resource friendly, and safely disposable
Divided into five main sections:
Function approximation/modeling - applying mainly physics and mathematical concepts to large number of inputs (ex. weather and business forecasting)
Classification - conducts pattern recognition and sequence recognition
Data processing - sorting data (ex. filtering, blind source separation, and clustering)
Impact on Current/Future Society and Environment
Technical Details (con.)
ANNs have three main layers
input layer
hidden layer
output layer
two types of ANNs
feed-forward ANN - data goes one way
feedback ANN - data goes in both direction between all the three layers
two learning methods for ANNs
associative mapping - matches input to output & uses nearest-neighbour recall
regularity mapping - anaylzes input for patterns and changes its own code with each input
Robotics - mimicking humans (ex. computer game-play, prosthetics, and artificial intelligence development)
Medical Analysis - diagonsis of various diseases/infections (ex. lung cancer detection system called HLND)
from a psychological, biological, and spiritual perspective...
"will artificial intelligence help humans understand themselves or will humans understand themselves first in order to create the perfect artificial intelligence?"
ANNs can become capable of doing everything
statistically improbable, but possible
no need for humans to have jobs
ethical issue of what defines us as humans
what do people life for?
Why go to school?
the financial issue of how people will get paid
ANNs have the ethical issues that surround AIs
are AIs human or some different being? do they have rights?
can they be trusted?
they were designed to replicate the biological brain so shouldn't they feel love, compassion, jealousy, or even hatred
summary: ANN are a branch of computer models/programs that are based on the biological brain
a history of up and downs
a complex technology, but simple basic concepts
current real applications in many fields
will most likely be very relevant in the future but must first overcome some technical problems
also has some environmental and ethical issues to address
despite its benefits and disadvantages, artificial neural networks are one of the emerging technologies of the future
Full transcript