7 Steps of Machine Learning

1. Gathering Data
2. Preparing that Data
3. Choosing a Model: CNN
4. Training
5. Evaluation
6. 2 more

MNIST

Handwritten digits. From 0 to 9. It’s a dataset. Famous one. Great to play with Convolutional Neural Network (CNN).

Yann LeCun (ex Meta and NYU). He has been interviewed or being in articles. He states that the scaling Law wont make AI smarter.

Ilya Sustskever - He made Chat GPT4 happen. Former supporter of scaling law but now thinks is not the solution either.

Quick Draw dataset. Human data. 345 categories.

Convolutional Neural Networks

They are computationally efficient. Have superhuman accuracy. Started with AlexNet, MobileNet and StyleTransfer. Anything related to images has some convolution in the network.

Structure

Dense layers → Normal layers

Convolution layers → Sips the raw information of the pixel.

Clasification layer.