This is an interdisciplinary workshop involving speakers from both industry and academia. The workshop revolves around the following fairness-related key aspects:
How to measure fairness? Many fairness metric exist across different applications: is one better than the other?
Transparency in marketing and e-commerce: How to define an optimization problem and impression allocation mechanism that can be considered fair to all parties and that permits algorithmic fairness?
Fairness over online platforms: Ranking algorithms have recently come under scrutiny for preventing minority groups from reaching higher ranking slots in applications like search and recommendation, thus reducing their visibility. How do algorithms bias against minority groups?
Bias and privacy preservation for decision making: Automated decision making systems influence critical decisions, raising concerns about fairness and representation. The cultural and social implications of these systems are profound, as biases embedded in training data and model design can reinforce stereotypes and perpetuate inequalities. How do different training objectives and architectures exacherbate biases?
The workshop talks will provide an overview of cutting-edge technical tools that can be leveraged for fairness guarantees in automated decision making systems. Attendees will have the opportunity to discuss the potential benefits and risks of using control theory for the societal good. The understanding of feedback loops and system dynamics plays a key role in unfolding the fundamental aspects underlying bias exacerbation.
The talks will be brought together in a final panel discussion allowing attendees to reflect on the (also ethical) implications of applying control theory to cope with societally relevant problems, stimulating new research into socially aware and fair control and potentially fostering new collaborations with industry and across different disciplines.
The workshop is aimed at graduate students and researchers interested in broadening their knowledge about algorithmic fairness metrics and automated decision-making systems algorithms. In particular, the talks of our one-day workshop aim to showcase how these algorithms work, what they optimize for, what the fundamental aspects behind bias propagation are, and how control theory helps designing bias mitigation solutions.
At the same time, as the topics and frameworks touched by the talks are various but self-contained, the workshop is also intended for researchers who aim to approach these disciplines in their research, having a glance at possible research directions and existing techniques, by taking inspiration from the approaches used in industry and across different disciplines.
By pairing experienced and young researchers with diverse expertise, the workshop brings together a rather varied group of internationally recognized scientists affiliated with institutions in Europe and the United States of America. Follow here for more details on the talks.