Intro to Traditional Deep Learning

This session provides an overview of the foundational elements of deep learning, including its historical context, key concepts, and practical applications. The course will delve into various types of neural networks, outlining their advantages and disadvantages. It will specifically focus on convolutional neural networks (CNNs), recurrent neural networks (RNNs), and long short-term memory (LSTM) networks, highlighting their unique characteristics and applicability to a range of problem-solving scenarios, including those in economics.

Auto Insurance in Sweden

Swedish Committee on Analysis of Risk Premium in Motor Insurance. read more

Overview

In the dataset:

• X = number of claims
• Y = total payment for all the claims in thousands of Swedish Kronor for geographical zones in Sweden

Reference: Swedish Committee on Analysis of Risk Premium in Motor Insurance

Tasks

• First step: We will create a simple Artificial Neural Network with 1 node and training with 1 sample of data
• Second step: The simple Artificial Neural Network will be trained through the dataset

Notebooks

• Intro to Traditional Deep Learning - ANN
• Intro to Traditional Deep Learning - RNN
• Intro to Traditional Deep Learning - LSTM
• Intro to Traditional Deep Learning - CNN
• Intro to Activation Functions
• Intro to Gradient Descent

Exercises

• ANN Exercise
• RNN Exercise
• LSTM Exercise

Solutions

• ANN Exercise and Solutions

• LSTM Exercise and Solutions

Recommended Datacamp exercises

• Deep Learning with PyTorch

Slides

Use arrows keys on keyboard to navigate. Alternatively use fullscreen slides.

Resources

• PyTorch
• PyTorch NN Linear