Summer school

Visuel summer school

Artificial Intelligence Summer School 2020

 22th June – 3rd July - INSA Rouen Normandy School of Engineering

A 2-week summer school focusing on a major challenge of today’s society : artificial intelligence and a discovery of Normandy, a breath-taking and historical French region.

Credits

4 ECTS

  • Artificial Intelligence – 42h (contact hours including visits of labs) / 3.5 credits 
  • French Culture and Language – 15H (contact hours including courses and cultural visits) / 0.5 credit

Program transcripts

Artificial Intelligence

Targeted Students: Science and Engineering Undergraduate students from 4th-5th Year, Graduate students and Professionals

Learning objectives: 

Contents: 

  • Introduction to AI 
  • Machine Learning and AI 
  • Introduction to Deep Learning 
  • Deep Learning & Vision 
  • Reinforcement Learning 
  • Symbolic AI 
  • AI in Chatbot 
  • SMA for AI 
  • Autonomous Vehicle and AI 

Pre-requisites: 

- a good level of English (European CECRL: B2) and a strong motivation

- cf. course description

French Culture and Language

Targeted Students: All university students

Learning objectives: 

Getting an insight into the French culture and traditions, and getting some notions of French 

Contents: 

  • French History and Culture + Survival French (A1-A2 levels)

Pre-requisites:

none 

 

Applying is Easy

2020 Fees Early Registration (December - 16th February) Standard Registration (17th February – 31st March) Late Registration** (1st April- 18th May)
  • Students from INSA Rouen Normandie’s partner universities *
  • Students from INSA group
800 € 850 € 950 €
Alumni from INSA group 900 € 950 € 1 050 €
All other standard delegates 1 000 € 1 050 € 1 150 €

Limited places

* Students from partner universities have to contact summerschool@insa-rouen.fr prior to registration
** Only if places left, contact summerschool@insa-rouen.fr to know whether it is still possible to apply before registration

 

The Summer School tuition fees are competitively priced and include: 

  • a set of goodies from INSA Rouen Normandy
  • a participation to the Opening Evening Cocktail
  • a participation to the Formal Farewell Dinner
  • a 2-week access to courses 
  • academic materials from our providers 
  • access to wifi in the school
  • a field-trip excursion to a popular French tourist destination 
  • a visit of the historical city of Rouen 
  • a 10-ride metro ticket
  • free access to the school facilities (gym, library etc.)
  • your lunch from Mondays to Fridays

 

Optional: 

  • Shuttle from Paris CDG airport to INSA Rouen Normandy 
    Arrivals on Monday 22ND June morning only (departure of shuttle at 1pm)- 45 eur
    Please let us know at summerschool@insa-rouen.fr your arrival time and flight number before May 22nd   
  • Shuttle from INSA Rouen Normandy to Paris CDG airport 
    Departures on Friday 3 at 2pm (arrival at the airport around 5-6pm) – 45 eur
  • Accommodation in residences
    Rooms on campus can be proposed on demand if available when you register. 

 

The Summer School tuition fees do not include: 

  • Travel to and from INSA Rouen Normandy, personal insurance, food (except lunch), personal expenses and anything not listed as included.

 

To Bring

  • Your laptop or tablet

Application Procedure

How to apply 

Step 1

Step 1

Step 1

Step 1

To book your place, you will need to:

  1. Fill out our online application form https://insarouen.moveonfr.com/locallogin/5dde8a7c82bf0508510098c3/eng
  2. You'll receive an email to confirm your application is being processed. We recommend not to book your flight and accommodation before you receive your registration confirmation as for April 6th
  3. As for April 6th, you will receive an email to confirm your acceptance at INSA with instructions on how to confirm your place and process payment (tuition fees + options). 

Final registration will close on 15th April 2020. 

Terms and Conditions

Cancellation possible 30 days prior to the beginning of courses (refund of 90 % of the tuition fees). 

Summer school programme description

Contact hours: 3h

 

This course is an introduction to artificial intelligence through use cases and applications of artificial intelligence such as those found in personal assistants or autonomous vehicles. It aims to help students understand the concepts and vocals of AI such as multi agent systems, machine learning, deep learning and neural networks. Various issues and concerns surrounding artificial intelligence such as ethics, bias and future uses will be discussed.

Contact hours : 6h

Pre-requisites : Python, Basics of Linear Algebra

 

Simple machine learning methods can lead to efficient prototypes to many real world problems. In this course, we will review simple and off-the-shelf methods to design a first machine learning model (Linear Regression, SVM, KMeans, XgBoost, etc.). The course will also focus on protocols and rules of thumbs to design an efficient and robust model.

Contact hours: 3h

 

The lecture will go through these new developments in deep learning covering basic motivations, ideas, models and optimization in deep learning for computer vision, natural language processing and gaming, identifying challenges and opportunities. It will focus on issues related with large scale learning that is: high dimensional features, large variety of classes, large number of examples and related neural architectures and frameworks including a practical session with Keras. 

Contact hours : 3h

Pre-requisites : Basic knowledge in computer programming, algorithmics and probabilities

 

The course will present different techniques to implement decision making processes in a multi-agent system. The goal of these techniques is to allow a software agent to decide which actions it should perform according to its current situation observed by its perception modules. The course will give a broad survey of the field ranging from deliberative agent architecture, such as the BDI framework, to probabilistic planning using markovian decision processes. It will also cover single agent decision making where only one agent has to decide which action to perform to multi-agent decision making where several agents have to perform a coordinated group of actions, with potential interferences.

Contact hours : 3h

Pre-requisites : image processing, data fusion, machine learning

 

This course will be the opportunity to present a review of the different sensors used for the intelligent vehicle perception and navigation. We will present each modality used and its application (pedestrian detection, road scene detection, road lines detection, etc.) as well as the different ways for data fusion. We will draw a particular attention to the use of polarization encoded sensors to deal with the vehicle perception in bad weather conditions. In all these cases, we will show how AI successes to mix multimodality and efficient decision in intelligent systems navigation. 

Contact hours: 3h

 

While the volume of information available to users is growing drastically, the interaction capabilities of systems evolve slowly. The goal of this course is to introduce the concept of intelligent agent as interactive partner, in order to design personalized interaction.

Context is a key factor to personalize human-agent interaction. Interaction should be adapted to humans (by opposition to other agents), to categories of humans, to individuals and to the current activity. To recognize such contexts, prototypical situations can be extracted from activities or dialogues as recurrent patterns of behavior and exploited as the interactive and reasoning models of an intelligent agent.

Embodied conversational agents (ECA -virtual agents and robots-) are proven to be intuitive and interactive interfaces, with promising results in several applications such as assistance, e-learning, remediation and so on. In particular, taking into account the user’s emotions and social relations improve the user’s engagement in interaction and task performance.

The Human-Agent interaction course will therefore deal with:  formal models for interaction (automata, HMM, etc.), virtual and augmented reality (human-agent/robot interaction, mixed communities, etc.), multi-modal interaction (gesture recognition, emotion detection, etc.), dialogue (ECA, affective computing, etc.) and their applications.

Contact hours : 6h

Pre-requisites : Programming Skills - Knowledge about First Order Logic (FOL).

 

Knowledge Representation and Reasoning (KRR) is the field of Symbolic Artificial Intelligence (AI) that focuses on building representations about the world in a form that a computer system can use to solve complex tasks. It incorporates findings from psychology about how humans solve problems and from logic to automate various kinds of reasoning, such as the application of rules. To enable the encoding of semantics with the data, Description Logics (DLs) are an important family of logic-based formalism that have been developed for the representation of conceptual knowledge as a set of ontological axioms. Recently, DLs have attracted increased interest since they form the logical basis of ontology languages such as OWL (Web Ontology Language). These technologies formally represent the meaning involved in information. For example, an ontology can describe concepts, relationships among things, and categories of things. These embedded semantics with data offer significant advantages such as reasoning over data and dealing with heterogeneous data sources.
This course covers the principles of knowledge representation and reasoning procedures. Additionally, Symbolic AI is at the heart of the new research field of Explainable AI (XAI), that intends to propose intelligent systems that can be understood, validated and controlled by human beings. This new exciting research field will be also discussed in this course.

Contact hours : 6h

 
After recapitulating some basic neural network concepts such as error backpropagation, a focus will be placed on neural networks that have been successful in computer vision, in particular, convolutional neural networks. A second part of the presentation will look at the problem of assessing the quality of a trained network, in particular, methods that explain neural network predictions in a human-interpretable manner will be presented. The lecture will also include hands-on examples of how to program and train a simple neural network, and how to implement recent explanation procedures such as LRP on state-of-the-art convolution networks.

Contact hours : 6h

 
Reinforcement learning is a domain of Machine Learning. RL belongs to the field of sequential decision making under uncertainty. The RL problem is for an agent to learn to behave in order to solve a certain task (learning to play a game, learning to speak in natural language, learning to manage resources, learning to control a robot, …). During this tutorial, I will present the reinforcement learning problem, formalize it, present the main algorithms to solve it, illustrate it on use cases, explain some best practices to use RL to solve your problem.

Academic staff

Samia AINOUZ

Stéphane CANU

Benoit GAUZERE

Franco GIUSTOZZI

Grégoire MONTAVON

Alexandre PAUCHET

Philippe PREUX

Laurent VERCOUTER

Cecilia ZANNI-MERK

Program schedule (to be confirmed)

June - July

Day

Morning

Afternoon

22

Monday

Arrival at the airport

Arrival in Rouen – Set up in accomodation

Opening Evening Cocktail at INSA 

23

Tuesday

Introduction to AI

French Culture and Language Initiation 

24

Wednesday

Machine learning and AI

Machine learning and AI

25

Thursday

French Culture and Language Initiation

Visit of the city of Rouen

26

Friday

Visit of labs (CORIA/ LITIS) 

+ CRIANN data center

SMA for AI 

27

Saturday

 Visit of Mont Saint-Michel

28

Sunday

Free day

29

Monday

Introduction to deep learning

Autonomous vehicle and AI

30

Tuesday

Symbolic AI

Symbolic AI

1

Wednesday

Deep learning & vision

Deep learning& vision

2

Thursday

Reinforcement Learning

Reinforcement Learning

Formal Farewell Diner

3

Friday

AI in chatbot

Departure at 2pm