Francisco S. Melo
Short Personal Interview
Francisco S. Melo was born in 1977, in Guarda, Portugal. He is an INESC-ID Researcher since 2009, integrating the Scientic Area Artificial Intelligence for People and Society (AIPS).
How did you get to INESC-ID?
I started in 2009 as an Associated Laboratory Researcher, and in 2010 I joined IST as a faculty. At that time, I became a senior researcher at INESC-ID.
A research institution that fosters state-of-the-art research in electrical engineering and computer science and facilitates knowledge and technology transfer between academia and industry in Portugal.
Research project(s) under development
How would you explain in the most accessible and least technical language possible, what is the application / expected results of this (these) project (s)?
HOTSPOT seeks to develop robots that can successfully collaborate with humans across multiple tasks.
ANIMATAS is a European initiative that seeks to build a network of PhD researchers in the area of human-robot interaction.
ILU seeks to improve traffic conditions in the city of Lisbon using Artificial Intelligence.
Tell us about your favorite project so far (or one of them)?
My favorite project so far was project INSIDE, which was a collaboration with several research institutions in Portugal and the USA, Portuguese companies and a Hospital, and investigated the use of robots in the therapy of children with autism spectrum disorders.
What are the biggest challenges of working in research in this area?
I work in artificial intelligence, specifically, a subarea of artificial intelligence known as machine learning. Currently, research in machine learning poses challenges along three orthogonal “perspectives”:
– Scientific: In spite of the big successes of research in machine learning and artificial intelligence, current methods require absurd amounts of data and computation. One of the challenges is, therefore, to reduce the requirements of such methods. Another important challenge is related with interpretability: the aforementioned successes rely on complex models that are hard to interpret by humans, which brings forth several issues (e.g., trust).
– Ethical: Relying on large amounts of data, machine learning algorithms are naturally subject to the biases that humans exhibit, and thus the algorithms tend to exhibit the same sort of discriminative outputs that are present in the data. Pursuing research in machine learning that mitigates the effect of such biases is, therefore, a very relevant challenge in this area.
– Methodological: The large amount of computation required by state-of-the-art algorithms requires that institutions performing research in these areas have large computational budgets available, with two important consequences: institutions with humbler budgets find it challenging to compete with larger institutions/companies, and must therefore find research “niches” where the computation is a less central requirement; on the other hand, the results portrayed in many articles are hard to reproduce without access to such computation, which brings about the challenge of reproducibility.
What book are you currently reading?
- Sapkowski, “Blood of Elves”.
How would you explain to your child (or your parents or grandparents) what your job is? How do you explain what means to be a researcher in this area?
My job is to make machines more intelligent and able to learn.
How do you see the mission of INESC-ID “to produce added value to people and society, supporting the response of public policies to scientific, health, environmental, cultural, social, economic and political challenges, in the fields of Computer Science and Electrical and Computer Engineering”?
I think that, in light of the technological advances of recent years, if INESC-ID is able to keep this mission in sight during its internal and external strategic decision-making, it has a unique opportunity to position itself as a top research institute at the national, European and even at the world level.
Linkedin Profile Link
PhD in Electrical and Computer Engineering
Training / Research Area(s)
Artificial Intelligence (Machine Learning/Reinforcement Learning)
INESC-ID Scientific Area *
ESSIR 2022: 13th European Summer School in Information Retrieval
This year’s edition of ESSIR, a 5-day teaching event on Information Retrieval (IR) and related research topics, will take place between 18-23 July 2022. The event is organized by NOVA LINCS (FCT UNL), INESC ID and Universidad Autónoma de Madrid with the support of the CMU Portugal Program and BCS, the Chartered Institute for IT.
The event will also incorporate the new edition of the Symposium on Future Directions in Information Access (FDIA), which will provide a forum for early researchers to present their research in a friendly environment, while among senior researchers.
The summer school is addressed to students, Master’s students, Ph.D. Researchers, Industry and Academia practitioners. No deep previous knowledge of IR or ML is required, but the attendants are assumed to have some basic computer science and mathematics background.
The Summer School will be held at the Faculty of Social and Human Sciences (FCSH) of the Universidade Nova de Lisboa.
The early registration deadline is June the 15th 2022, and the application form is available online.
EPIA’22: EPIA Conference on Artificial Intelligence
The EPIA Conference on Artificial Intelligence is a well-established European conference. The 21st edition of the EPIA conference will take place at Instituto Superior Técnico by 2022.
As in previous editions, this international conference is hosted with the patronage of the Portuguese Association for Artificial Intelligence (APPIA).
The purpose of this conference is to promote research in all areas of AI, covering both theoretical/foundational issues and applications, and the scientific exchange among researchers, engineers and practitioners in related disciplines.
For further details please visit the EPIA’22 website.