Archive

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...

The performance of non-invasive electroencephalogram-based (EEG) brain–computer interfaces (BCIs) has improved significantly in recent years. However, remaining challenges include the non-stationarity and the low signal-to-noise ratio of the EEG, which limit the bandwidth and hence the available applications. Optimization of both individual components of BCIs and the interrelationship between them is crucial to enhance bandwidth. In other words, neuroscientific knowledge and machine learning need to...

Melanoma (skin cancer) is considered one of the most concerning forms of cancer due to its great potential to metastasize. Furthermore, statistical data show that the incidence rates of melanoma have been steadily growing in the past decades, leading to an increase in the number of deaths related with it. However, if melanoma is diagnosed on an early stage, the life span of the patient...

In the first part of this talk, I will present AD^3 (Alternating Direction Dual Decomposition), a new decoding algorithm for approximate LP-MAP inference in constrained factor graphs. The LP-MAP approximation consists in ignoring global effects caused by the cycles of the graph, and can be seen as a linear relaxation of the original problem. The proposed algorithm can handle arbitrary first-order logic constraints and is...

The structure control system design consists mainly of two steps: input/output (I/O) selection and control configuration (CC) selection. The first one is devoted to the problem of computing how many actuators/sensors are needed and where should be placed in the plant to obtain some desired property. Control configuration is related to the decentralized control problem and is dedicated to the task of selecting which outputs...

In this seminar I will talk about the work of Hsu and Kakade (2009) on spectral methods for leaning a Hidden Markov Model (HMM). I will introduce and exemplify this method and describe how we can estimate the best set of state transitions or how we can predict a sequence of future observations. Hidden Markov Models are one of the most fundamental tools for modeling...

In this talk I will describe our recent efforts within the PT-STAR project for speech translation across languages. I will begin with brief descriptions about the component systems for speech recognition, machine translation and speech synthesis and talk in greater detail about modeling and conversion of prosodic aspects of speech across languages, a major part of speaking style. Illustrating with example demos for the case...

Biometrics, as an authentication tool, provides several advantages over conventional what you know (e.g., password, PIN) and what you possess (e.g., keys, tokens) authentication methods. However, a biometrics is an irrevocable password as we can’t change the biometrics easily. If it is compromised digitally, it is compromised for ever. Secondly, a biometrics can be easily matched against multiple databases to link identities. In order to...

Moraglio et al. have recently introduced new genetic operators for genetic programming, called geometric semantic operators. It is possible to prove that these operators induce a unimodal fitness landscape for all the problems consisting in matching input data with known target outputs (like regression and classification). This feature facilitates genetic programming evolvability, which makes these operators extremely promising. Nevertheless, Moraglio et al. leave one big...

The Internet is an essential tool but can be disruptive distraction. Using wifi usage and student grade data from the Faculdade de Engeneharia da Universidade do Porto (FEUP) in Portugal, we find a consistent, positive correlation between Internet usage and student performance. This result is robust for several models and and for different metrics. Models controlling for usage specific to major and curricular year, and...