rossana

What is Machine Learning?

Florian Stevenart Meeûs will embrace the challenge to answer our basics questions on Machine Learning (ML).

What is ML, how does it work, and what are the advantages & disadvantages ?

We will see that ML is a great and powerful tool that can be widely used. Unfortunately, it presents the disadvantage of being hardly explainable. Can we trust a black box?

A perfect start for understanding what are the legal implications of machine learning.

 

Florian Stévenart Meeûs is a software & machine learning engineer at Macq, a company innovating in smart mobility and automation. Florian is working on a smart camera and develops deep neural networks to recognize vehicles and people. Previously, he made his Master thesis about Person Detection with Automatic Systems. He gratuated with magna cum laude from Université Catholique de Louvain.

Posted by rossana, 0 comments

Data protection cases: an exercise to visualise facts and legal issues

February 15, 2019

Today, we run a little experiment in class. After an introduction to the fundamental principles and rationale of data protection legislation, EITLab students where asked to solve a case, using a canvas as a support.

The canvas was a model, which was created with the aim of supporting students in the process of identification and visualisation of the main elements emerging in their case. To list all the dots would have supported the connection of them in order to draw significant conclusion.

The legal cases that students had to solve were the following.

 

  1. Winter is coming

Jon, an employee of the Winterfell municipality, asks to his colleagues to sign a petition and contribute with a small donation for a charity campaign for protecting direwolves.

Jon is also a passionate blogger and after few days he publishes on his blog the name of the colleagues participating to the crowdfunding campaign and titles the post “We will fight the tyrannical regime of King’s landing! The direwolves are coming!”

 

2. The condominium pillory

Mary is a building administrator and she has several difficulties in contacting all the owners of the apartments and in receiving the payment of condominium taxes and fees on time.

Before proceeding with the injunction, she tries a last recourse and she hangs on the building board (the one close to the elevator) the name of the people not paying the condominium fees and the corresponding amount of the debt.

 

3. A ride to the hospital

Valentino had a motorcycle accident. Nothing serious, but his wrist hurts. He calls the ambulance and taken to the nearest hospital. After the triage with the nurse on duty he is accompanied to the waiting room. There are approximately 20 people waiting there for their turn. Little by little, each patient is called via the intercom:

“KIM KAM, PLEASE GO TO THE MAXILLOFACIAL DIVISION, 3RD FLOOR!”

“LARRY SMITH, PLEASE GO TO THE CARDIOLOGY DIVISION, 2ND FLOOR!”

“JANE DOE, PLEASE GO TO THE OBSTETRIC AND GYNECOLOGY DIVISION, 5TH FLOOR”

“HARRY JOKE, PLEASE GO TO THE CENTER FOR COGNITIVE DISORDER, 4TH FLOOR!”

 

4. Tell me what you watch and I will tell you who you are

Lara loves to watch new TV series on her favourite platform for streaming (Letflix). The platform offers to Lara several recommendations for movies on the basis of her previous activity.

One day the platform launches a global challenge and offer 1,000,000 Euros to whom will be able to optimise the recommendation algorithm. To do so, Letflix releases a database containing: 1) a code number linked to a user; 2) the list of movies watched; 3) the rating given to the movies.

Few weeks later, a team of researchers at UCLouvain discovers that with a simple algorithm is possible to single out the identity of some individuals.

Lara is worried because she has always watched TV series classified by Letflix as Gay TV shows. Her friends and family do not know she is lesbian. She does not want to come out, yet.

(based on Jane Doe v. Netflix suit)

 

5. Re-use of medical data for research

Dan is participating to a research project for discovering a new treatment for cancer done by the Trust Research Institute.

The Trust Research Institute informs Dan about all the details of the processing in line with the GDPR.

Dan provides all his electronic medical records and he also accept to fill a questionnaire about his behaviours (if he goes to the gym, f he goes to work by bike, what does he eat, how many time per week he drinks alchool, etc.).

After few months, the Trust Research Institute uses Dan’s data for another research project for investigating mental health problems and potential remedies.

 

6. A tragedy of hashtag

Gigì wants to run for the next European Parliament elections with Macron’s party.

However, in the Facebook’s profile of Alessandro, Gigì’s friend, there are a couple of pictures portraying Gigì during a «gilets jaunes» demonstration.

Gigì asks Alessandro to remove the tag revealing his identity, but Alessandro objects « My dear Gigì, my profile is private! Only few friends can see it. Your secret is safe with me! »

 

Students were divided in couples and had approximately 20′ to solve the case. First, they had to read the case and note their answers on the post-it individually. After that, each point was discussed by the couple to reach a common position. Each group was then called to present their own solution and the steps to arrive to their conclusion.

This was the canvas we used during the class. It was a simple model for getting students acquainted with brainstorming techniques (and the use of post-it!). You are free to re-use it or to create a more detailed matrix of analaysis.

 

Canvas 1. MIND MAP

Download the Canvas here

 

After the use of the first canvas, thanks to the interaction with students, we have also created a second canvas in the form of a flow chart, representing the legal reasoning and steps followed for de-constructing and re-constructing the case. Credits to: Xavier Darmstaedter.

 

Canvas 2. FLOW CHART

Download the Canvas here

How would you have solved (visually and not) the cases? Write your answer in the comment!

 

Posted by rossana, 0 comments

Welcome to the EITLab

8 February 2019

Today, the EITLab has started!

EITLab will be one of the first courses in Europe dedicated to the study of IT Law through the lenses of legal design.

The course is a Jean Monnet Module, co-funded by the Erasmus+ Program of the EU Commission.

 

Our timeline

 

 

Posted by rossana, 2 comments

Transparency by Legal Design. Panel at CPDP2019

On January 30, the CRIDES (UCLouvain) and the Université Saint-Louis organised the first panel centred on legal design at the 12th International Conference “Computer, Privacy and Data Protection” (CPDP2019).

Within one of the most important events in the data protection panorama, we had a fruitful discussion on legal design and its interplay with the principle of transparency.

The GDPR has expressly codified transparency as a fundamental data protection principle. In particular, the Regulation requires that information notices, communications to the data subjects and requests for consent must be drafted “in a concise, transparent, intelligible and easily accessible form, using clear and plain language”, highlighting the importance of taking the target audience into consideration (e.g. in case the information is addressed to children). Along the same line, the proposed ePrivacy Regulation moves towards new transparency obligations. But how can such a principle be implemented in practice? Legal Design is an innovative approach that can shed some lights on this challenge. It is an interdisciplinary field that combines design methodology, insights from behavioural law and economics, and IT to pursue a legal goal. With our panelists we discussed the foundations of Legal Design and shown some of its practical applications in the field of data protection.

 

Monica Palmirani, Professor of Legal Informatics at the University of Bologna (Italy)

What is Legal Design? What are its scientific foundations?

The Legal Design Manifesto: scope and rationale

 

 

“Communication is not a trivial task: legal theory and philosophy of law are two pillars we can rely on for communicating and explaining legal information”

 

 

Antti Innanen, Co-founder and CEO at Dot. (legal design consultancy firm, Helsinki)

Antti presented the practical experience of Dot. where they are helping clients to develop new and interactive ways to communicate privacy related information. He shown some examples they elaborated for the privacy policies of Estée Lauder.

 

 

“At some point we are going to laugh at the situation where are we now: do you remember the time when there used to be websites with privacy policies that were over 30 pages long? It not going to be like this forever, but it is difficult to make those systemic changes. Legal design can be very powerful to support such a transition”

 

 

Arianna Rossi, Ph.D. Candidate at University of Bologna and University of Luxembourg

Arianna summarised in 5 minutes her 3-years PhD project dedicated to the elaboration of a legal design methodology for creating machine-readable icons to accompany privacy policies.

 

 

“There are at least two problems in the current design of privacy policies: first, privacy policies are usually presented as a “wall of text”, impenetrable to the human eye, which is something that works against the will of people of being informed about their rights; second, these documents are usually very long, even if you are looking for a specific information… you lose your patience before you find it. Why don’t we use icons to present the information in a more meaningful way?”

 

 

Tom Norton,  Executive Director at Center of Law and Information Policy  at Fordham Law

Privacy indicators can play a crucial role in enhancing the awareness of users. However, there are several obstacles to their success. Tom explored what are the main hurdles for privacy indicators, drawing some recommendations for their successful deployment.

 

“Based on our research, we have elaborated a set of recommendations for making privacy indicators meaningful. First, policy makers should establish standardised set of evaluation criteria: so that indicators systems are not comparing apples and oranges. Users can go for platforms and services and get a consistent evaluation of what they are using”.

 

 

Christoph Schmon, Senior Legal Officer, Leader of the Consumer Rights team, BEUC

If there is a field where the need of transparency is felt more urgently, this is undoubtedly AI. Christoph critically discussed how AI is changing the consumer data market, how we are moving toward an increasing personalisation of information (e.g. dynamic pricing) and what will be the role for legal design.

 

“We have discussed that in data protection law information must be intelligible, submitted in a clear and plain language. I have to disappoint many of you perhaps, but this is not new, it is not a revolution. It is a copy and paste of consumer law that we had for decades in the EU and consumer law has struggled ever since to give meaning to this concept. Behavioural studies and legal design can help to realise the information paradigm in practice”.

 

 

 

Chair

Professor of Media Law at Université Saint-Louis – Bruxelles and avocat at Daldewolf

Moderator

Researcher at UCLouvain and Université Saint-Louis – Bruxelles. Fellow of the Trento LawTech group

Posted by rossana, 0 comments