The document summarizes the key topics covered so far in previous AI Club sessions, including how to represent images numerically, flattening images, multi-layer perceptrons, tuning MLPs, residual neural networks, and training models in Navigator. The upcoming session agenda includes learning how to tune residual neural networks, comparing results from MLPs and ResNet, discussing the pros and cons of each algorithm, and selecting a custom image project to work on.
3. What have we learnt so far?
1 0 0 0 0
0.5 1 0 0 0
0.5 0.5 1 0 0
0.5 0.5 0.5 1 00
0.5 0.5 0.5 0.5 1
• How to represent images as number? - pixels
grayscale
Color with Red, Green and
Blue channels
4. 1 0 0 0 0
0.5 1 0 0 0
0.5 0.5 1 0 0
0.5 0.5 0.5 1 00
0.5 0.5 0.5 0.5 1
1
0
0
0
0
0.5
1
0
0
0
0.5
0.5
1
0
0
0.5
0.5
0.5
1
0
0.5
0.5
0.5
0.5
1
• How to represent images as
number?
• How to flatten images
• We discussed python
code to do this
What have we learnt so far?
5. What have we learnt so far?
• How to represent images as number?
• How to flatten images
• Multi-layer Perceptron (MLP)
6. What have we learnt so far?
• How to represent images as number?
• How to flatten images
• Multi-layer Perceptron (MLP)
• How to tune MLP – Stochastic Gradient
Descent (SGD)
• Learning rate
• Number of layers and number of neurons in each
layer
• epochs
7. What have we learnt so far?
• How to represent images as number?
• How to flatten images
• Multi-layer Perceptron (MLP)
• How to tune MLP
• Residual neural network (ResNet)
8. What have we learnt so far?
• How to represent images as number?
• How to flatten images
• Multi-layer Perceptron (MLP)
• How to tune MLP
• Residual neural network (ResNet)
• How to train MLP and ResNet in Navigator
9. What have we learnt so far?
• How to represent images as number?
• How to flatten images
• Multi-layer Perceptron (MLP)
• How to tune MLP
• Residual neural network (ResNet)
• How to train MLP and ResNet in Navigator
• Cats and Dogs Exercise
10. What will we do today?
• Learn how to tune Residual Neural Network (ResNet)
11. What will we do today?
• Learn how to tune Residual Neural Network (ResNet)
• Compare results from MLP and ResNet
12. What will we do today?
• Learn how to tune Residual Neural Network (ResNet)
• Compare results from MLP and ResNet
• Discuss Pros and Cons of the two algorithms
13. What will we do today?
• Learn how to tune Residual Neural Network (ResNet)
• Compare results from MLP and ResNet
• Discuss Pros and Cons of the two algorithms
• Select a custom project for this course and collect
images for it