Seminário Doutoral

Explainable semantic similarity for biomedical supervised learning

Sala 6.3.27, Ciências ULisboa

Por Rita Sousa (orientadoras: Cátia Pesquita e Sara Silva).

Explainable artificial intelligence approaches ensure algorithmic fairness, identify potential bias in the data, ensure that the algorithms perform as expected, and bridge the gap between the machine learning community and other scientific disciplines. They are key to promoting the adoption of machine learning as a tool for scientific discovery. Explanations in the biomedical domain should be grounded in domain knowledge which can be achieved by using ontologies and knowledge graphs. However, the most popular way to explore knowledge graphs with machine learning is through embeddings, which are not explainable. In this work, the use of ontology-based semantic similarity that captures different semantic aspects represented in a knowledge graph as a tool to support both supervised learning and explainability is investigated. The underlying hypothesis is that using more semantic aspects to compute similarity and more interpretable models can make machine learning over ontologies more explainable with minimal losses in predictive performance. The experiments for protein-protein interaction prediction revealed that interpretable machine learning models coupled with semantic similarity, although performing worse than black-box ones, produce global models relevant to the biological phenomena and have a high prediction agreement with black-box models. This work represents a step towards demonstrating the potential of explainable artificial intelligence for scientific discovery.

12h00-13h00
Departamento de Informática | Ciências ULisboa

Um programa dirigido aos alunos de Licenciatura e de 1.º ano de Mestrado.

Philosophical metatheories have had a great development in recent times. They have been used to reconstruct theories in many different scientific fields, from physics to biology to the social sciences. However, they lack means to systematically test the adequacy of their products.

Seminário do Centro de Física Teórica e Computacional, por Artem Ryabov (Charles University, Prague, Czech Republic).

.

Cognitive sciences have approached social cognition from two complementary perspectives. On one hand, social psychology and behavioural economics have mostly emphasized mentalizing, social perception, and our ability to model other minds. On the other hand, developmental psychology and motor neuroscience have more focused on imitating, social interaction, and our propensity to coordinate with others’ behaviour. 

Logótipo do evento

O conceito de One Health assenta no reconhecimento da necessidade de abordar as políticas de saúde numa perspetiva de interdependência humano-animal-ambiente.

The course aims at enabling the participants to use different methods to measure the impacts of pollutants on ecosystems. Basic knowledge will be provided through theoretical and practical lessons on how to select and use the most suitable metrics based on the analysis of multiple compartments of the ecosystems.

O evento visa reunir investigadores interessados na aplicação de isótopos estáveis em estudos ecológicos e na utilização de dados históricos de biodiversidade disponíveis em coleções museológicas.

Logótipo e data do evento

“There is often a considerable difference between what we think people need and what people want to use”. This is our perspective on how technology should be developed and will serve as the motto of this event.

Título e data do evento, sobre uma fotografia da Ponte Vasco da Gama (Lisboa)

O evento decorre de 12 a 14 de julho de 2022, sob o lema CHEMISTRY: Forging bonds, e destina-se a todos os investigadores e alunos da ULisboa.

Banner do evento

The program will consist of plenary lectures, oral, flash and poster communications, and a round table about Science Communication.

This course offers an overview of the different ways to measure biodiversity, and provides tips for the stratification of primary biodiversity data and the construction of variables that describe its various facets. It also includes an in-depth review of the different types of data used to measure biodiversity and their problems and limitations.

Banner do evento

Estaremos hoje habilitados a prever o impacto da atividade humana no clima e a inferir daí a insustentabilidade do nosso modo de vida?

Logótipo do programa

Após 2 anos de interregno, o programa Ser Cientista está de regresso!

Imagem alusiva ao prémio

Candidaturas até 31 de julho.

Fotografia representativa de agricultura sustentável

Submissão de iniciativas até 30 de agosto de 2022.

Fotografia de trator em campo agrícola

Submissão de iniciativas até 30 de agosto de 2022.

Fotografia de campo agrícola e parque eólico

Submissão de iniciativas até 30 de agosto de 2022.

Programa a disponibilizar futuramente.