Workshop – Finance and Insurance

MAY 13-17, 2019 | MONTRÉAL (English)

JUNE 10-14, 2019 | MONTRÉAL (French, IVADO members only)

About

The Fin-ML/IVADO Workshop is a one-week practical training in machine learning, applied to concrete problems in finance and insurance. This workshop will consist of theory in the morning, followed by problem-solving workshops in finance and insurance in the afternoon.

The workshop will be held in English. Participants will have to bring their laptops for the practical part, no special installation is required.

Objectives

  • Train professionals in new technologies in data science, machine learning and operations research;
  • Develop an understanding of the challenges and issues of data science applied to a specific field;
  • Learn to use computer tools to solve concrete problems;
  • Foster knowledge sharing and facilitate networking among specialists in a particular field;
  • Encourage interdisciplinary knowledge sharing.

Organizers

  • Manuel Morales, Université de Montréal
  • Brian Moore, IVADO
  • Rheia Khalaf, Université de Montréal / IVADO

Contact

For any inquiries, please contact us at formations@ivado.ca.

Dates and place

May 13-17, 2019: The workshop will be held in English at HEC Montréal (PWC, Côte Ste-Catherine).

June 10-14, 2019: The workshop will be held in French at HEC Montréal (PWC, Côte Ste-Catherine).

Program

Introduction to machine learning

Theory (3 h)

1. Introduction

  • Logistic regression and regression: Machine learning vs. statistical approaches

2. Types of learning: supervised, unsupervised and reinforced

3. Good practices: Overlearning

4. Good practices: Cross-validation and experimental design

Tutorial (2h30)

1. Framework presentation : Python, Keras, Pytorch.

2. Some illustrative examples.

Supervised and unsupervised learning.

Theory (3 h)

1. Introduction : Classification problem

  • Traditional approaches : SVM, Random Forests, etc
  • Modern approaches : neural networks
  • Good practices

2. Introduction : Clustering problem

  • Traditional approaches : K-means
  • Modern approaches : Embeddings

Tutorial (2h30)

1. Keras Tutorial : Data mining in insurance

Social event/cocktail

Neural networks

Theory (3 h)

1. Introduction : Forecasting problem

2. Multilayered Perceptron

3. Introduction to recurrent neural networks (RNN)

4. Good practices

Tutorial (2h30)

1. Pytorch tutorial : Data mining in finance

Introduction to NLP

Theory (3 h)

1. Text processing

2. Sentiment analysis

  • Application of convolutional neural networks (CNN)

3. Embeddings and anomaly detection

  • Generative models

Tutorial (2h30)

1. Case study: Spam detection and sentiment analysis

Reinforcement learning

Theory (3 h)

1. Introduction: learning by reinforcement

  • Q-learning

Tutorial (2h30)

1. Case study: Reinforcement learning in finance