Symbolic Knowledge Extraction from Trained Neural Networks Governed by Lukasiewicz Logics

This work describes a methodology to extract symbolic rules from trained neural networks. In our approach, patterns on the network are codified using formulas on a Lukasiewicz logic. For this we take advantage of the fact that every connective in this multi-valued logic can be evaluated by a neuron in an artificial network having, by activation function the identity truncated to zero and one. This fact simplifies symbolic rule extraction and allows the easy injection of formulas into a network architecture. We trained this type of neural network using a back-propagation algorithm based on Levenderg-Marquardt algorithm, where in each learning iteration, we restricted the knowledge dissemination in the network structure. This makes the descriptive power of produced neural networks similar to the descriptive power of Lukasiewicz logic language, minimizing the information loss on the translation between connectionist and symbolic structures. To avoid redundance on the generated network, the method simplifies them in a pruning phase, using the "Optimal Brain Surgeon" algorithm. We tested this method on the task of finding the formula used on the generation of a given truth table. For real data tests, we selected the Mushrooms data set, available on the UCI Machine Learning Repository.

An optimal scheduling problem in distribution networks considering V2G

This paper addresses the problem of energy resource scheduling. An aggregator will manage all distributed resources connected to its distribution network, including distributed generation based on renewable energy resources, demand response, storage systems, and electrical gridable vehicles. The use of gridable vehicles will have a significant impact on power systems management, especially in distribution networks. Therefore, the inclusion of vehicles in the optimal scheduling problem will be very important in future network management. The proposed particle swarm optimization approach is compared with a reference methodology based on mixed integer non-linear programming, implemented in GAMS, to evaluate the effectiveness of the proposed methodology. The paper includes a case study that consider a 32 bus distribution network with 66 distributed generators, 32 loads and 50 electric vehicles.

Implementação da técnica Silver In Situ Hibridization para avaliação do status do gene EGFR em doentes com CPNPC

Avaliação do estado do gene Epidermal Growth Factor Receptor (EGFR), por Silver In Situ Hibridization (SISH), tem-se destacado como biomarcador preditivo na resposta à terapêutica. O principal objectivo foi optimizar a etapa de recuperação por calor da metodologia automatizada SISH Dual-Colour, em carcinomas pulmonares fixados em formol durante 24 e 72 horas. A optimização levou a um aumento da preservação do contorno nuclear e da intensidade e contraste dos sinais para os dois tempos de fixação, permitindo avaliar o estado do EGFR em 83,3% dos casos em estudo. A SISH Dual-Colour é uma alternativa para avaliar o estado do EGFR.

Avaliação da expressão do gene BDNF em doenças neurodegenerativas

O Factor Neurotrófico Derivado do Cérebro (BDNF) está associado a processos de crescimento, diferenciação e sobrevivência das células neuronais. A expressão diferencial do BDNF, particularmente no hipocampo, está relacionada com a manifestação clínica de algumas doenças do foro psiquiátrico e cognitivo como a doença de Huntington, Alzheimer, depressão e esquizofrenia. Este trabalho pretende dar conhecimento das técnicas utilizadas para avaliar a expressão do gene BDNF. As técnicas de ELISA, IHC e Western blot, por permitirem a avaliação precisa da expressão de BDNF, são úteis para uma melhor compreensão, diagnóstico e tratamento de algumas doenças neurodegenerativas.

Can power laws help us understand gene and proteome information?

Proteins are biochemical entities consisting of one or more blocks typically folded in a 3D pattern. Each block (a polypeptide) is a single linear sequence of amino acids that are biochemically bonded together. The amino acid sequence in a protein is defined by the sequence of a gene or several genes encoded in the DNA-based genetic code. This genetic code typically uses twenty amino acids, but in certain organisms the genetic code can also include two other amino acids. After linking the amino acids during protein synthesis, each amino acid becomes a residue in a protein, which is then chemically modified, ultimately changing and defining the protein function. In this study, the authors analyze the amino acid sequence using alignment-free methods, aiming to identify structural patterns in sets of proteins and in the proteome, without any other previous assumptions. The paper starts by analyzing amino acid sequence data by means of histograms using fixed length amino acid words (tuples). After creating the initial relative frequency histograms, they are transformed and processed in order to generate quantitative results for information extraction and graphical visualization. Selected samples from two reference datasets are used, and results reveal that the proposed method is able to generate relevant outputs in accordance with current scientific knowledge in domains like protein sequence/proteome analysis.

Polimorfismo I/D do gene da enzima conversora da angiotensina

Em Portugal, no ano de 2009, existiam 983 mil indivíduos com Diabetes mellitus (DM), dos quais 90% DM tipo 2 (DM2). Um dos principais factores de risco para o desenvolvimento da DM2 é a obesidade. A DM2 está, desta forma, associada a um estilo de vida sedentário, quer pela diminuição do metabolismo da glicose, quer pela sua associação à obesidade devido ao aumento dos níveis de glicose como consequência da sobrealimentação. Outros factores como a lipotoxicidade e o stress oxidativo também têm sido considerados como responsáveis pela disfunção das células-beta pancreáticas. O gene da enzima conversora da angiotensina (ECA) é altamente polimórfico, sendo associado como factor predisponente para diferentes patologias como a DM. O polimorfismo melhor descrito até à actualidade é o de Inserção/Delecção (I/D), consistindo na inserção de um fragmento Alu de 287bp no intrão 16 do gene DCP1. A actividade pró-inflamatória e pró-oxidativa desta enzima sobre o pâncreas, bem como a sua actuação nos processos de fibrose, podem em parte auxiliar na compreensão do processo que origina esta patologia. O polimorfismo I/D torna-se assim um óptimo candidato para a detecção de indivíduos susceptíveis.Este estudo tem como objectivo analisar a distribuição do polimorfismo I/D, bem como a sua possível relação com a incidência de DM e com os níveis de glicose.

VirtualECare: group support in collaborative networks organizations for digital homecare

Collaborative Work plays an important role in today’s organizations, especially in areas where decisions must be made. However, any decision that involves a collective or group of decision makers is, by itself complex, but is becoming recurrent in recent years. In this work we present the VirtualECare project, an intelligent multi-agent system able to monitor, interact and serve its customers, which are, normally, in need of care services. In last year’s there has been a substantially increase on the number of people needed of intensive care, especially among the elderly, a phenomenon that is related to population ageing. However, this is becoming not exclusive of the elderly, as diseases like obesity, diabetes and blood pressure have been increasing among young adults. This is a new reality that needs to be dealt by the health sector, particularly by the public one. Given this scenarios, the importance of finding new and cost effective ways for health care delivery are of particular importance, especially when we believe they should not to be removed from their natural “habitat”. Following this line of thinking, the VirtualECare project will be presented, like similar ones that preceded it. Recently we have also assisted to a growing interest in combining the advances in information society - computing, telecommunications and presentation – in order to create Group Decision Support Systems (GDSS). Indeed, the new economy, along with increased competition in today’s complex business environments, takes the companies to seek complementarities in order to increase competitiveness and reduce risks. Under these scenarios, planning takes a major role in a company life. However, effective planning depends on the generation and analysis of ideas (innovative or not) and, as a result, the idea generation and management processes are crucial. Our objective is to apply the above presented GDSS to a new area. We believe that the use of GDSS in the healthcare arena will allow professionals to achieve better results in the analysis of one’s Electronically Clinical Profile (ECP). This achievement is vital, regarding the explosion of knowledge and skills, together with the need to use limited resources and get better results.

Optimization, high availability and redundancy in information communications and technology using networks and systems virtualization architectures

Mestrado em Engenharia Informática

Non-coding RNAs: multi-tasking molecules in the cell

In the last years it has become increasingly clear that the mammalian transcriptome is highly complex and includes a large number of small non-coding RNAs (sncRNAs) and long noncoding RNAs (lncRNAs). Here we review the biogenesis pathways of the three classes of sncRNAs, namely short interfering RNAs (siRNAs), microRNAs (miRNAs) and PIWI-interacting RNAs (piRNAs). These ncRNAs have been extensively studied and are involved in pathways leading to specific gene silencing and the protection of genomes against virus and transposons, for example. Also, lncRNAs have emerged as pivotal molecules for the transcriptional and post-transcriptional regulation of gene expression which is supported by their tissue-specific expression patterns, subcellular distribution, and developmental regulation. Therefore, we also focus our attention on their role in differentiation and development. SncRNAs and lncRNAs play critical roles in defining DNA methylation patterns, as well as chromatin remodeling thus having a substantial effect in epigenetics. The identification of some overlaps in their biogenesis pathways and functional roles raises the hypothesis that these molecules play concerted functions in vivo, creating complex regulatory networks where cooperation with regulatory proteins is necessary. We also highlighted the implications of biogenesis and gene expression deregulation of sncRNAs and lncRNAs in human diseases like cancer.

A model for heterogeneous networks management and Performance evaluation

In general, modern networks are analysed by taking several Key Performance Indicators (KPIs) into account, their proper balance being required in order to guarantee a desired Quality of Service (QoS), particularly, cellular wireless heterogeneous networks. A model to integrate a set of KPIs into a single one is presented, by using a Cost Function that includes these KPIs, providing for each network node a single evaluation parameter as output, and reflecting network conditions and common radio resource management strategies performance. The proposed model enables the implementation of different network management policies, by manipulating KPIs according to users' or operators' perspectives, allowing for a better QoS. Results show that different policies can in fact be established, with a different impact on the network, e.g., with median values ranging by a factor higher than two.

Distributed process execution in collaborative networks

Processes are a central entity in enterprise collaboration. Collaborative processes need to be executed and coordinated in a distributed Computational platform where computers are connected through heterogeneous networks and systems. Life cycle management of such collaborative processes requires a framework able to handle their diversity based on different computational and communication requirements. This paper proposes a rational for such framework, points out key requirements and proposes it strategy for a supporting technological infrastructure. Beyond the portability of collaborative process definitions among different technological bindings, a framework to handle different life cycle phases of those definitions is presented and discussed. (c) 2007 Elsevier Ltd. All rights reserved.

FMRI 3D registration based on Fourier space subsets using neural networks

In this work, we present a neural network (NN) based method designed for 3D rigid-body registration of FMRI time series, which relies on a limited number of Fourier coefficients of the images to be aligned. These coefficients, which are comprised in a small cubic neighborhood located at the first octant of a 3D Fourier space (including the DC component), are then fed into six NN during the learning stage. Each NN yields the estimates of a registration parameter. The proposed method was assessed for 3D rigid-body transformations, using DC neighborhoods of different sizes. The mean absolute registration errors are of approximately 0.030 mm in translations and 0.030 deg in rotations, for the typical motion amplitudes encountered in FMRI studies. The construction of the training set and the learning stage are fast requiring, respectively, 90 s and 1 to 12 s, depending on the number of input and hidden units of the NN. We believe that NN-based approaches to the problem of FMRI registration can be of great interest in the future. For instance, NN relying on limited K-space data (possibly in navigation echoes) can be a valid solution to the problem of prospective (in frame) FMRI registration.

Regulation of antioxidant enzymes gene expression in the yeast Saccharomyces cerevisiae during stationary phase

Gene expression of three antioxidant enzymes, Mn superoxide dismutase (MnSOD), Cu,Zn superoxide dismutase (Cu,ZnSOD), and glutathione reductase (GR) was investigated in stationary phase Saccharomyces cerevisiae during menadione-induced oxidative stress. Both GR and Cu,ZnSOD mRNA steady state levels increased, reaching a plateau at about 90 min exposure to menadione. GR mRNA induction was higher than that of Cu,ZnSOD (about 14-fold and 9-fold after 90 min, respectively). A different pattern of response was obtained for MnSOD mRNA, with a peak at about 15 min (about 8-fold higher) followed by a decrease to a plateau approximately 4-fold higher than the control value. However, these increased mRNA levels did not result in increased protein levels and activities of these enzymes. Furthermore, exposure to menadione decreased MnSOD activity to half its value, indicating that the enzyme is partially inactivated due to oxidative damage. Cu,ZnSOD protein levels were increased 2-fold, but MnSOD protein levels were unchanged after exposure to menadione in the presence of the proteolysis inhibitor phenylmethylsulfonyl fluoride. These results indicate that the rates of Cu,ZnSOD synthesis and proteolysis are increased, while the rates of MnSOD synthesis and proteolysis are unchanged by exposure to menadione. Also, the translational efficiency for both enzymes is probably decreased, since increases in protein levels when proteolysis is inhibited do not reflect the increases in mRNA levels. Our results indicate that oxidative stress modifies MnSOD, Cu,ZnSOD, and GR gene expression in a complex way, not only at the transcription level but also at the post-transcriptional, translational, and post-translational levels.

Tubulin cofactor A gene silencing in mammalian cells induces changes in microtubule cytoskeleton, cell cycle arrest and cell death

Microtubules are polymers of alpha/beta-tubulin participating in essential cell functions. A multistep process involving distinct molecular chaperones and cofactors produces new tubulin heterodimers competent to polymerise. In vitro cofactor A (TBCA) interacts with beta-tubulin in a quasi-native state behaving as a molecular chaperone. We have used siRNA to silence TBCA expression in HeLa and MCF-7 mammalian cell lines. TBCA is essential for cell viability and its knockdown produces a decrease in the amount of soluble tubulin, modifications in microtubules and G1 cell cycle arrest. In MCF-7 cells, cell death was preceded by a change in cell shape resembling differentiation.

Q192R polymorphism of the paraoxonase-1 gene as a risk factor for obesity in Portuguese women

Introduction - Obesity became a major public health problem as a result of its increasing prevalence worldwide. Paraoxonase-1 (PON1) is an esterase able to protect membranes and lipoproteins from oxidative modifications. At the PON1 gene, several polymorphisms in the promoter and coding regions have been identified. The aims of this study were i) to assess PON1 L55M and Q192R polymorphisms as a risk factor for obesity in women; ii) to compare PON1 activity according to the expression of each allele in L55M and Q192R polymorphisms; iii) to compare PON1 activity between obese and normal-weight women. Materials and methods - We studied 75 healthy (35.9±8.2 years) and 81 obese women (34.3±8.2 years). Inclusion criteria for obese subjects were body mass index ≥30 kg/m2 and absence of inflammatory/neoplasic conditions or kidney/hepatic dysfunction. The two PON1 polymorphisms were assessed by real-time PCR with TaqMan probes. PON1 enzymatic activity was assessed by spectrophotometric methods, using paraoxon as a substrate. Results - No significant differences were found for PON1 activity between normal and obese women. Nevertheless, PON1 activity was greater (P<0.01) for the RR genotype (in Q192R polymorphism) and for the LL genotype (in L55M polymorphism). The frequency of allele R of Q192R polymorphism was significantly higher in obese women (P<0.05) and was associated with an increased risk of obesity (odds ratio=2.0 – 95% confidence interval (1.04; 3.87)). Conclusion - 55M and Q192R polymorphisms influence PON1 activity. The allele R of the Q192R polymorphism is associated with an increased risk for development of obesity among Portuguese Caucasian premenopausal women.