Roadmap to Membership of RICS - Pathways and Routes
Introduction to artificial neural networks
1. Introduction to Artificial
Neural Networks
Object Recognition Problem
Piyush Mishra (Mtech Production, 217ME2221)
Department of Mechanical Engineering, NIT Rourkela
3. • A child doesn't know whether the object in front of him is a CAT or a
DOG…we need to tell him that (a) is Dog and (b) is Cat.
In other words we TRAIN him/he
LEARNS about various objects.(a)
(b)
4. Neural Networks are also trained
By showing them many images of cats and dogs. But will they give the correct answer in first attempt?
8. Summary- The Big Picture
• Imagine the box below to be our brain. Our brain contains billions of neuron.
• Each neurons contains weights and thresholds (W & T).
W
T
T
W
W
W
W
W
T
T
T
T
T
W
W
T T
T
W
W
W
Ref.[3]
X1
Xm
Z1
Zm
Z=f(x,w,t)
• Output Z is a function of inputs ‘x’, weights ‘w’ and thresholds ‘t’.
• We have to adjust the weights and thresholds so that what we get out is what we want.
9. References
[1] The art of neural networks, Mike Tyka, TEDXTUM,
https://youtu.be/0qVOUD76JOg
[2] Introduction to Deep Learning: What Are Convolutional Neural
Networks? MATLAB, https://youtu.be/ixF5WNpTzCA
[3] 6.034, Fall 2015, Artificial Intelligence, Patrick H. Winston, Lec 12a,
MIT OCW, https://youtu.be/uXt8qF2Zzfo