Seminars


The Priberam Machine Learning Lunch Seminars are a series of informal meetings which occur every two weeks at Instituto Superior Técnico, in Lisbon. It works as a discussion forum involving different research groups, from IST and elsewhere. Its participants are interested in areas such as (but not limited to): statistical machine learning, signal processing, pattern recognition, computer vision, natural language processing, computational biology, neural networks, control systems, reinforcement learning, or anything related (even if vaguely) with machine learning.

The seminars last for about one hour (including time for discussion and questions) and revolve around the general topic of Machine Learning. The speaker is a volunteer who decides the topic of her presentation. Past seminars have included presentations about state-of-the-art research, surveys and tutorials, practicing a conference talk, presenting a challenging problem and asking for help, and illustrating an interesting application of Machine Learning such as a prototype or finished product.

Presenters can have any background: undergrads, graduate students, academic researchers, company staff, etc. Anyone is welcome both to attend the seminar as well as to present it. Ocasionally we will have invited speakers. See below for a list of all seminars, including the speakers, titles and abstracts.

Note: The seminars are held at lunch-time, and include delicious free food.

Feel free to join our mailing list, where seminar topics are announced beforehand. You may also visit the mailing list webpage. Anyone can attend the seminars; no registration is necessary. If you would like to present something, please send us an email.

The seminars are usually held every other Tuesday, from 1 PM to 2 PM, at the IST campus in Alameda. This sometimes changes due to availability of the speakers, so check regularly!

Tuesday, June 4th 2013, 13h00 - 14h00

Stefano Ruberto (INESC-ID)

Semantic Approach in Genetic Programming

Room PA2 (Edifício de Pós-Graduação)

Instituto Superior Técnico - Alameda

Abstract:

Evolutionary algorithms are stochastic optimization techniques based on the principles of natural evolution, and Genetic Programming (GP) belongs to this family. In recent years the study of GP systems has been extended to phenotypic aspects, while previously it was mainly focused on genotypic and syntactic aspects. Phenotypic properties, or semantics, is used with the aim of optimizing the ability of GP algorithms to explore the solution space in an effective way, increasing the probability of finding an optimal solution and escaping local optima. Currently, semantic GP is strictly related to the evaluation of individual behavior in the candidate population: this kind of evaluation is mainly obtained through the fitness function itself. This work introduces a new way of measuring semantic similarity between individuals that is more independent from the fitness itself, allowing a fair comparison even when the fitness values involved are far away from each other. This new measure enables a new series of techniques to be used to tackle the open problems in GP, like bloat and overfitting, and also targeting the phenotype variety preservation, thereby enhancing performance. Preliminary results will be provided. A new theoretical GP algorithm based on this new semantic measure is also introduced, showing the potential advantages. Very early results coming from a first naive implementation show interesting insight on this potential, when compared with other cutting edge algorithms.

--

Bio: Stefano Ruberto was born in Firenze, Italia (1974). He earned his bachelor degree with honours from Università dell'Insubria in 2005, while working as an IT consultant in the fashion industry where, his interests in fractal and automatic image composition were greatly appreciated. Stefano's interest in Artificial Intelligence and modelling were adequately nourished while he was working at KFT, a multi-national company in security and telecommunication. There, he could not only develop real world physics model of radio transmission for telephone triangulation, but also develop models for interpreting data from low cost MEMES accelerometers, gyroscopeand compass, that are applied to innovative automotive. Models aimed at vehicular traffic prediction were also developed. Finally, Stefano graduated with honours from University of Milano Bicocca in 2013. Currently he is a research fellow at INESC-ID, working on Genetic Programming with special focus on new semantic approaches for the novel field of Semantic GP.