This time the topic will be Fair and Ethical Algorithms. Dojocrea (
How do we go about designing fair and ethical Algorithms? It looks like the subject is gaining some attention, here are three that just showed up:
- Bart Selman, Moshe Vardi et Wendell Wallach on the societal impact of AI .
- The 2nd International Workshop on the future of AI, Phoenix, AZ: "What is the future of AI? And what should we be doing about it now?")
- How a flawed approach to a Random Forest algorithm is implemented ... with consequences.
The meetup will be streamed (see above) Here is our lineup:
Franck Bardol and Igor Carron, Introduction
Suresh Venkatasubramanian, An Axiomatic treatment of fairness
We propose a mathematical framework for reasoning about fairness in machine learning.Here is an Al Jazzera show on the general subject of fair algorithms (with Suresh in it). Suresh is coming from the world of TCS, so his presentation will provide an interesting take on fairness I am sure.
Michael Benesty Application of advanced NLP techniques to French legal decisions: Demonstration of a significant bias of some French court of appeal judges in decisions about the rights of asylum. (pdf)
The presentation will start with a brief overview of the French legal system and the legal decisions that have been analyzed. The main part will be dedicated to Word2vec and a custom multi input and multi task learning algorithm based on bi-directional GRU and classical deep learning used for extraction of information from public law decisions. In the last part, some basic descriptive statistics will be used to analyze the extracted information and reveal the apparent bias of some French court of appeal judges.Attendant website: http://www.supralegem.fr
Michel Blancard, CMAP / EtaLab, A sunny day in the CDO team of the French gov
The OpenSolarMap.org project aims to create a roof orientation map of the French territory. I will present how we crowdsourced a training dataset and how we used deep learning on it.Attendant website: OpenSolarMap
Pierre Saurel, "Ethics of Algorithms: still an oxymoron?"
Most algorithms especially those from mathematics are considered intrinsically neutral, objective and universal. Ethics is the science of judgments of appreciation. In that sense algorithms seem to have no ability to judge and Ethics of Algorithms seems to be an empty oxymoron. Nevertheless, and beyond this simplistic view, considering certain kinds of algorithms this expression is no more an oxymoron. We propose six categories of algorithms to establish an Ethics of Algorithms.
Let me finally note that one of the Phoenix workshop presenter is Mark Reidl, he will be a speaker of the Paris Machine Learnng meetup in June . His current paper is titled: "Using Stories to Teach Human Values to Artificial Agents"
Liked this entry ? subscribe to Nuit Blanche's feed, there's more where that came from. You can also subscribe to Nuit Blanche by Email, explore the Big Picture in Compressive Sensing or the Matrix Factorization Jungle and join the conversations on compressive sensing, advanced matrix factorization and calibration issues on Linkedin.