Seminars

MAP-tele Seminars 2014/2015

The MAP-Tele Seminars includes a series of invited talks, and a module on Research Methods on Engineering.

The talks are on telecommunications with an emphasis on Future Internet Technologies.

MAP-Tele Seminars grading takes into account the participation in both the talks and the module.

Talks scheduled for 2014/2015:

1 Speaker: João Álvaro Carvalho, UM/Algoritmi
TitleCreating scientific knowledge: challenges to the technological areas (Part I)
17.10.2014, 11h00
2 Speaker: Paulo Cortez, UM/Algoritmi
TitleData mining and computer networks
24.10.2014, 11h00
3 Speaker: João Álvaro Carvalho, UM/Algoritmi
TitleCreating scientific knowledge: challenges to the technological areas (Part II)
31.10.2014, 11h00
4 Speaker: Pétia Georgieva, UA
TitleMachine learning for Big Data processing
14.11.2014, 11h15
5 Speaker: Paulo Ferreira, UA
TitleAn introduction to probabilistic methods and network applications
20.11.2014, 11h15
6 Speaker: Pétia Georgieva, UA
TitleArtificial neural networks for advanced signal processing
21.11.2014, 14h30

.

Seminar 1

Speaker: João Álvaro Carvalho, UM/Algoritmi
TitleCreating scientific knowledge: challenges to the technological areas (Part I)
Abstract:
The seminar addresses several aspect related to the creation of technological knowledge, namely: the scientific method - a general process model for the creation of scientific knowledge - and its adaptation for areas that involve the creation of technological knowledge; main differences between the research process and the engineering design process (and other professions that also involve design); criteria for the validation of technological knowledge; different types of scientific research; broad view of research methods, i.e., methods for the validation of scientific knowledge.

Seminar 2

Speaker: Paulo Cortez, UM/Algoritmi
TitleData mining and computer networks
Abstract:
This seminar starts by briefly presenting the area of Data Mining. Then, data mining applications to the computer network field are presented (e.g., Internet Traffic Forecasting, Spam Email detection using Network Level properties).

Seminar 3

Speaker: João Álvaro Carvalho, UM/Algoritmi
TitleCreating scientific knowledge: challenges to the technological areas (Part II)
Abstract:
The seminar addresses several aspect related to the creation of technological knowledge, namely: the scientific method - a general process model for the creation of scientific knowledge - and its adaptation for areas that involve the creation of technological knowledge; main differences between the research process and the engineering design process (and other professions that also involve design); criteria for the validation of technological knowledge; different types of scientific research; broad view of research methods, i.e., methods for the validation of scientific knowledge.

Seminar 4

Speaker: Pétia Georgieva, UA
TitleMachine learning for Big Data processing 
Abstract:
Machine Learning (ML) is a subfield of artificial intelligence that is concerned with the design, analysis, implementation, and applications of programs that learn from examples
or experience. Learning from data is commercially and scientifically important. ML consists of methods and respective software that extract automatically interesting knowledge (patterns, models, relationships) in large databases of sometimes chaotic and redundant information. ML is a data-based knowledge-discovering process that has the potential not only to analyze events in retrospect but also to predict future events or important alterations.  The objective of this lecture is to introduce the PhD students to the research area of ML. The lecture will give a flavor of Supervised (classification and regression) versus Unsupervised (clustering) learning paradigms. Distance (similarity) measures, instance based classifiers, Bayesian statistical learning, and performance evaluation (confusion matrix). ML applications, such as text mining, image classification, brain data processing will be discussed.

Seminar 5

Speaker: Paulo Ferreira, UA
TitleAn introduction to probabilistic methods and network applications 
Abstract:
The goal of the talk is to overview some of the probabilistic method more often used in network applications, in a big data context, or in connection with the web graph. I will discuss probabilistic counting and cardinality estimation and some of the basic underlying ideas. I will also address set membership determination and some of its applications in networking. The problem of estimating the elements of a set as well as their frequencies of occurrence will also be debated. Finally, some attention will be given to fountain codes and the probabilistic ideas behind them. The talk will be of interest to students with a basic knowledge about probability and an interest in topics such as coding, peer-to-peer networks, big data, data mining, routing, query optimisation, data summarisation, page ranking and large-scale problems on the web.

Seminar 6

Speaker: Pétia Georgieva, UA
TitleArtificial neural networks for advanced signal processing
Abstract:
Artificial Neural Networks (ANN) is a computational and engineering methodology whose design is based on models taken from neurobiology and on the notion of learning, in particular, the brain's massively parallel and learning aspects. The use of neural networks is becoming increasingly widespread, with applications in many areas. The objective of this lecture is to introduce the PhD students to the research area of ANNs and to some of their most successful applications as in advanced signal processing. The lecture is going to consider typical ANN architectures (Feedforward ANN, Radial Basis ANN, Recurrent ANN) and associated computational algorithms for accomplishing learning (Gradient Descent and Back propagation error correction learning). ANN for classification and regression will also be discussed.