Sumários
Machine Learning from Systems to Biology
7 Janeiro 2019, 14:30 • Catia Luisa Santana Calisto Pesquita
Title: Machine Learning from Systems to Biology
Presenter: Alcides Fonseca
Short Bio:
Alcides Fonseca is an Assistant Professor at FCUL and a researcher at LASIGE. Alcides holds a PhD from the University of Coimbra, where he taught as an Assistant and as an Invited Professor at ISCTEM, Mozambique. His research crosses the areas of Programming Languages, Parallel Systems and Machine Learning. One example of such research was the CMU|Portugal AEminium project, from which a new concurrent-by-default programming language was developed, targeting multicore and GPU processors.
http://alcidesfonseca.com
Abstract:
The talk will cover two types of work. The first half of the talk will cover how Machine Learning was applied to automate complex decisions that authors of parallel programs have to make in order to have faster programs. Two examples will be provided: CPU vs GPU decision and choosing the best dynamic granularity mechanism.
The second half will cover how to apply the same classification algorithms to a completely different scenario: predicting the pathogenicity of genetic variants. Despite using the same algorithms, the challenges are completely different: what is the ideal dataset to use, how to integrate this pipeline into commercial clinical diagnosis, how to convince doctors to trust these results.
Support to essay writing and seminar presentation
14 Dezembro 2018, 14:30 • Catia Luisa Santana Calisto Pesquita
Support to essay writing and seminar presentation
Capacity Advisor – Network optimization using performance forecasts.
13 Dezembro 2018, 14:30 • Catia Luisa Santana Calisto Pesquita
Data Science Seminars@Ciências: João Martins (Head of Nokia Software's Digital Operations Assurance)
Title: Capacity Advisor – Network optimization using performance forecasts.
Date: December 13th, 14:30h
Where: C6.3.38
=========================================
Título
Capacity Advisor – Network optimization using performance forecasts.
Resumo
Radio network performance optimization is an area of great importance to telecommunication service providers. The intent is to both ensure that existing equipment is configured and operating at maximum capacity, as well as respond and adapt to changing traffic patterns, as well as plan the introduction of new equipment or technologies.
To support this process, Nokia introduced a functionality in its network performance tool (Nokia Performance Manager), called Capacity Advisor, that relies on time-series forecasting algorithms to provide a forecast of the network’s resource usage in the future; it then combines this information with a set of expert-built rules to provide recommendations for resolving or preventing performance problems.
Bio curta
João has more than 15 years of experience in telecommunications and OSS. He started as a software developer at Siemens for optical transport networks, and has held architect positions for products in the IPTV, BSS, Service Monitoring and Telecom Analytics areas.
He is currently head of Nokia Software's Digital Operations Assurance architecture team.
The FAIR Data Principles in practice: an application to the Plant Science domain
7 Dezembro 2018, 14:30 • Catia Luisa Santana Calisto Pesquita
Title: The FAIR Data Principles in practice: an application to the Plant Science domain
Summary:
The increasing scale of data production in several scientific domains poses challenges to knowledge discovery. It is critical to ensure that data is published in a manner that facilitates discovery, integration and analysis by both humans and machines. To this effect, the FORCE11 community published the FAIR Data Principles, which detail the criteria that data must meet to be Findable, Accessible, Interoperable and Reusable, and thus enable knowledge discovery. These principles have since been endorsed by several research infrastructures, namely by ELIXIR, Europe’s distributed infrastructure for biological data.
Plant Science is one of the scientific domains that raises more challenges to data FAIRness due to its heterogeneity and complexity. It is also one of ELIXIR’s communities, which is precisely tackling the challenge of making plant phenotypic data FAIR and enabling integration between phenotypic and genotypic data.
In this talk, I will dissect the FAIR Data Principles, identify the main hurdles to implementing them, then delve into the efforts of ELIXIR’s Plant Science community towards doing so in practice.
Biography:
Daniel Faria is a post-doctoral researcher in the Instituto Gulbenkian de Ciência under the ELIXIR-EXCELERATE project, and a member of BioData.pt, the Portuguese infrastructure for biological data and national ELIXIR node. In his work, he is putting his semantic web and knowledge management expertise to the service of ELIXIR’s Plant Science community and aiding in its goal of attaining data FAIRness.
Prior to his work on ELIXIR, he worked as a post-doctoral researcher at FCUL under the SOMER project, wherein he was (and remains) the lead developer of the ontology matching tool AgreementMakerLight, which continues to be highly successful to this day.
He obtained his PhD in Informatics (specialty Bioinformatics) at FCUL in 2012, and his graduate degree in Biological Engineering at the IST in 2004.
Seminar and essay support
30 Novembro 2018, 14:30 • Catia Luisa Santana Calisto Pesquita
Support on preparing seminars and writing essays.