Sumários
Artificial Intelligence and Machine Learning @Critical Software
23 Novembro 2018, 14:30 • Catia Luisa Santana Calisto Pesquita
Data Science Seminars@Ciências: Paulo Gomes (Head of AI and ML, Critical Software)
Title: Artificial Intelligence and Machine Learning @Critical Software
Abstract: A área de Inteligência Artificial tem tido um destaque bastante
grande hoje em dia, sendo uma das áreas mais promissoras em termos de
futuro. As grandes empresas como: a IBM, a Google, o Facebook, a
Microsoft e a Amazon, têm feito um grande investimento na área e estão a
tirar dividendos disso. A Critical Software como empresa inovadora
também está a apostar na área e tem já uma equipa com bastante
experiência na área, desenvolvendo vários projetos onde o valor
acrescentado é a IA. Nesta apresentação vão ser apresentadas as várias
áreas da IA onde a Critical atua, bem como a apresentação de
vários projetos que estão a ser desenvolvidos na Critical Software.
Bio:
Paulo Gomes has more than 20 years of experience in the areas of Artificial Intelligence (AI) and Machine Learning (ML). He now is responsible for the AI&ML area of Critical Software, where he is creating synergies with the projects and areas that Critical Software develops. A professional path that started in the University of Coimbra in 1996 where he was an Assistant Professor and Researcher until December 2015. Finished his PhD in 2004 in the AI area, and since then coordinated more than 30 industry projects within the University scope. Parallel to the academic career, he was the founder of three companies: Datastream in 1999, a company that did Data Mining (DM) and ML projects, where he was CFO and business manager; AKTWise in 2008, a company that worked in the area of Software Reuse using AI, where he was CEO; and Wizdee in 2009, a company that created a Natural Language Search Engine for Analytics using AI and ML, he was CEO and raised 1,5M€ in venture capital, the company was later acquired in 2017. He has coordinated more than 20 projects in the DM and ML area, with extensive experience in being project and technical manager. At a research level he has done research work on the AI and ML, with more than 120 peer- reviewed research articles published in International conferences and journals on areas such as Semantic Web, Natural Language Processing, Data Mining, Text Mining and Machine Learning. There is a complete list available at https://www.cisuc.uc.pt/people/show/118. Participated in more than 20 program committees of Interna- tional conferences and workshops. Lectured several courses in the PhD program and Masters in Informatics Engineering of the Department of Informatics Engineering of the University of Coimbra, from which it is highlighted: Business Intelligence (MSc, the Machine Learning part); Advanced Topics on Artificial Intelligence (PhD, the modules of: Natural Language Processing; Machine Learning; Intelligent Systems for Knowledge Management; Ontologies; and Semantic Web); Intelligent Systems for Knowledge Management (MSc); Semantic Web (MSc); and Artificial Intelligence (MSc).
Not Taught.
16 Novembro 2018, 14:30 • Catia Luisa Santana Calisto Pesquita
Docente em dispensa temporária de serviço para apresentar trabalho numa conferência. Aula substituída a 7 de Janeiro.
Finding cold mice in scientific literature - a machine learning story
9 Novembro 2018, 14:30 • Catia Luisa Santana Calisto Pesquita
Title: Finding cold mice in scientific literature - a machine learning story
Abstract: All new drugs must be first tested with animals – typically mice – before being included in clinical studies. But the stress levels of the lab mice affect how their tumors respond to the chemotherapy drugs tested on them. In fact, independent studies in 2013 and 2015 showed that mice housed in 22°C (72°F) bioteriums became resistant to cancer drugs while mice kept at 30°C (86°F) did not. So have scientists changed the housing temperature for mice following that discovery? I will talk about the NLP strategy we used to perform a semi-automated meta-analysis aimed at investigating whether scientific behavior changes after major discoveries.
Helena Deus received her PhD in Bioinformatics from Universidade Nova de Lisboa where she focused on Linked Data and Semantic Web applications for Health Care and Life Sciences, with an emphasis on Cancer Research. Helena specializes in data integration and data wrangling techniques including sparse data management, query parallelization, data reuse and data mining for the facilitation of medical knowledge insights. Helena is passionate about breaking the cancer research silos to allow researchers to derive more value from their data and publications. Prior to joining Elsevier, Helena's roles included directing a knowledge engineering and data science team at Foundation Medicine and leading projects and strategy for Health Care and Life Sciences at the Digital Enterprise Research Institute, National University of Ireland at Galway (DERI/NUIG). Helena has published over 30 peer reviewed papers and was one of the winners of the Big Data Track in the 2013 Semantic Web Challenge and of the Linked Data Cup with her work on linking data from The Cancer Genome Atlas.
How to prepare and present a seminar.
2 Novembro 2018, 14:30 • Catia Luisa Santana Calisto Pesquita
How to prepare and present a seminar. How to choose a topic. Engaging the audience. Outlining the presentation. Preparing the materials. Oral presentation. Questions and answers. Group discussion on these topics.
The good and the bad of satellite data for forest monitoring
26 Outubro 2018, 14:30 • Catia Luisa Santana Calisto Pesquita
Title:
The good and the bad of satellite data for forest monitoring
Abstract:
Forests constitute one of the major natural resources of Earth, and their destruction greatly contributes to global warming. Financing mechanisms such as REDD+ (www.un-redd.org) promote sustainable forest management while supporting economic development, but they require that forest monitoring be applied at national and local level. Remote Sensing (RS) satellite-based approaches are the most practical option for monitoring forests.
Relying on the big data repositories of optical and radar imagery from Copernicus (www.copernicus.eu), and the open-access cloud computing of Google Earth Engine (earthengine.google.com), my scientific challenge for the next 6 years is to develop robust algorithms that can learn the spatiotemporal relationships contained in the dense time-series multi-sensor data, in order to allow i) the production of enriched land cover maps, ii) the early detection of forest degradation, and iii) the monitoring of forest fires.
With this workplan in mind, I will take you on a tour through the good and the bad of satellite data, describing its format and meaning, how to use it for forest monitoring, related technical and human difficulties, and some pitfalls to avoid.
Bio:
Sara Silva is a principal investigator of BioISI (BioSystems and Integrative Sciences Institute) in FCUL. Her main research interests are nature-inspired machine learning methods, in particular Genetic Programming (GP), which she has applied in several interdisciplinary projects from remote sensing and forest science to epidemiology and medical informatics. Sara Silva obtained her PhD (2008) from the University of Coimbra. She is the creator and developer of GPLAB (A Genetic Programming Toolbox for MATLAB) since 2003, and a member of the editorial board of the GPEM (Genetic Programming and Evolvable Machines) journal since 2009. In 2015 she was Editor-in-Chief of the largest conference in Evolutionary Computation (GECCO) and in 2018 she received the "EvoStar Award for Outstanding Contribution to Evolutionary Computation in Europe".