Fairness In Machine-Learning Algorithms

Fairness In Machine-Learning Algorithms
Borja Sanz
Candidates should have a Ph.D. in a related subject, and preferably have published high-impact journals and at the top computer architecture and/or machine learning conferences. Basic knowledge of: Scientific computing (Python/C++/Matlab)
  • Information Sciences and Engineering (ENG)
DeustoTech (www.deustotech.eu) is a private non-profit institution of the Faculty of Engineering at the University of Deusto for applied research in new technologies. Since 2005 DeustoTech mission is to support the ICT activity in business and society through research, the development of technologies, innovation and knowledge transfer. We focus our activity around TRLs 2-7 and articulate it into four applied fields: Industry, Mobility, Energy and Society, having a fifth, the Chair of Applied Mathematics, as a transversal activity and support for the previous four. We are characterized for working with data of heterogeneous nature, throughout its life cycle and in compliance with ethical principles and humanists who define the University of Deusto. Applied Machine Learning and Data Analytics research area is focused in the improve of the machine and deep learning technique, with the focus in sentiment analysis detection in text, data pre-processing, data gathering and clustering. We have also expertise in artificial vision.


● Laorden, C., Galán-García, P., Santos, I., Sanz, B., Nieves, J., Bringas, P. G., & Gómez Hidalgo, J. M. (2014). Negobot: Detecting paedophile activity with a conversational agent based on game theory. Logic Journal of IGPL, 23(1), 17-30. ● Laorden, C., Ugarte-Pedrero, X., Santos, I., Sanz, B., Nieves, J., & Bringas, P. G. (2014). Study on the effectiveness of anomaly detection for spam filtering. Information Sciences, 277, 421-444. ● Sanz, B., Nieves, J., Laorden, C., Santos, I., & Bringas, P. G. (2013). ONLINE SEARCHES: from information recovery to knowledge generation. DYNA 88 (4), 392-394 ● Lorido-Botran, T., Huerta, S., Tomás, L., Tordsson, J., & Sanz, B. (2017). An unsupervised approach to online noisy-neighbor detection in cloud data centers. Expert Systems with Applications, 89, 188-204. ● Bilbao-Jayo, A., & Almeida, A. (2018). Automatic political discourse analysis with multi-scale convolutional neural networks and contextual data. International Journal of Distributed Sensor Networks, 14(11), 1550147718811827.
● Electronic Regional Manifestos Project. Analysis of the political discourse in social networks using convolutional neural networks. ● Aztarnet: Analysis of the online footprint of the companies using opinion mining, emotion classification and social network analysis. ● Toleraction (under evaluation): Text classification to detect hate-speech in social networks to create suitable counter-narratives. ● StandByMe (in preparation): Sexual harassment prevention and awareness by changing attitudes and behavior of young adults. Analysis of the social networks to detect harassment online.


The hosting team is a group integrated by research staff that has expertise with collaborating with other research areas. For example, several proposals are under evaluation with the Deusto Social Values Team, which has obtained on two consecutive occasions the recognition as the team of excellence in the calls of the Department of Education of the Basque Government. The central aim of the group is the analysis of social values, which is focused on two main lines of research: ● Breakdown of society and democracy through the study of political fact and ● Welfare and Social Policy through the study of social inclusion-exclusion processes and intervention systems. On the other hand, there are under evaluation projects with the Center of Applied Ethics of the university, that approaches the problem of trust in a straightforward and precise sequence of steps looking at a variety of actual trust problems, collecting citizens answers, locating biases in public media discourse and suggesting solutions.
With a growing understanding the social determinants of the bias in this type of systems, and the complexity of inter-relationships between the processed data, means that the control and identification of these biases cannot be tackled by the computer sciences sector alone—many factors lie within other research fields’ responsibilities. Aligned with the Europe 2020 Strategy and the Basque Country Smart Specialisation Strategy, DEUSTO research blends competitiveness, innovation and technology to tackle challenges for communities, companies and public bodies in the region. Ageing and Well-being, Creative Industries and Cities, Gender dimension, Social Justice and Inclusion and Strengthening Participation are the focus of the 5 Interdisciplinary Research Platforms. Besides, advanced research units contribute to generating knowledge and innovative solutions in the fields of efficient and clean Energy, Industry 4.0 (for example, advanced manufacturing or data analytics), Health and Territory. Collaboration with stakeholders is crucial for Deusto researchers to define research questions promoting social impact, sustainable development, innovative smart city solutions, efficient use of Big Data and well-being.


Europe in a changing world – inclusive, innovative and reflective societies. Still, it is essential to find ways to make equality real in daily life. Our societies undergo socio-cultural changes connected to migration, globalization, democracy crisis that we can either study and act upon or neglect and remain unprepared. Diversity should be seen as a potential rather than a drawback. Moreover, there is potential to improve the implementation of existing policies that tackle these changes on national and European level. There is a need to develop new models and innovative approaches for social inclusion. In this line, the use of machine learning could help to improve all these situations or could make the differences even more significant. This project aims to contribute to reducing all these differences. On the other hand, the Sustainable Development Goals are the blueprint for achieving a better and more sustainable future for all. They address the global challenges we face, including those related to poverty, inequality, climate, environmental degradation, prosperity, and peace and justice. In particular, this project is aligned with SDG 5: Gender equality. By identifying and correcting for gender bias, further automat­ing/augmenting tasks, AI is empowering women for growth and new opportunities.


Artificial intelligence is not merely a new technology but a key driver of economic transformation. Thus, it is necessary to include all stakeholders to ensure that the evolution of the AI system is being done in the right way. To achieve this goal, it is also essential an intense collaboration with the industry and social actors. Deusto includes a lot of collaboration with several key actors in the industry (for example, Mercedes, Sidenor, Vidrala or Progenika (a Griffols company)) but also in the humanitarian sector (for example, Cruz Roja) or other academic leaders (for example University of Granada or Centre of Research and Technology Hellas (CERTH)).


Technological instruments govern society. The Information Society is based on access to information and characterized by speed and instantaneity. Thus, the information society is categorically conditioned by electronic devices or ICT (information and communication technologies). One of the hardest problems that we have found in the last years is the vast amount of data that is necessary to train the system. We are working in several hybrid approaches to reduce the needs of this amount of data. The main goal is to integrate prior knowledge to the models, also reducing the biases of the system. And finally, we want all these technologies to be in real life. In particular, we have been working closely with human sciences researches, using machine learning as a help tool to mitigate some problems of society. We have been working with different social agents using these techniques to improve the research (e.g., terrorist activity in social networks, human trafficking, etc.).


The ethics of artificial intelligence are of growing importance. Artificial intelligence (AI) is changing societies and economies around the world. Principles like privacy protection, the fairness, transparency, accountability and explainability (among other aspects, like legal and regulatory compliance) will be critical elements that will define the way that these systems will impact in the society. Artificial intelligence holds enormous potential to improve the community. While we are building a “general AI” that replicates human knowledge, there are numerous “specific AI” technologies which are already incredibly sophisticated at handling specific tasks (for example, Medical AI technologies and autonomous vehicles). But all these benefits come with risks. All these systems can limit issues associated with human bias. To mitigate these problems, we have to use the existing ethics in context, not reinvented, and it is necessary to create systems that ensure that fit with all the principles beforementioned. To ensure these new principles is essential to understand how all these systems work, how they make their predictions, but also understand what the social and ethical implications of the outputs of these systems are. It will be necessary to develop countermeasures to remove (or at least mitigate) the problems.


One of the University of Deusto’s key duties is to be fully aware of problems within the institution itself and the society we live in. For this reason, it should take specific steps to boost integration and real equality of opportunity for people with specific support needs. Timely specific action is required to enable them to enter higher education in equal conditions and ensure their full integration in the university community. DeustoTech, as one of its institutions, is included into this service of social action and inclusion. The main aims consist of achieving full normalisation, equal opportunities and gradually adopting the steps needed to ensure that the University of Deusto is an inclusive educational institution. Furthermore, the University of Deusto provides them with guidance and support on the transition to the labour market jointly with special job centres and companies at large.