Ruolo delle specie reattive dell'ossigeno e di caveole/raft lipidici nella biosegnalazione in linee cellulari umane

Membrane lipid rafts are detergent-resistant microdomains containing glycosphingolipids, cholesterol and glycosylphosphatidylinositol-linked proteins; they seem to be actively involved in many cellular processes including signal transduction, apoptosis, cell adhesion and migration. Lipid rafts may represent important functional platforms where redox signals are produced and transmitted in response to various agonists or stimuli. In addition, a new concept is emerging that could be used to define the interactions or amplification of both redox signalling and lipid raft-associated signalling. This concept is characterized by redox-mediated feed forward amplification in lipid platforms. It is proposed that lipid rafts are formed in response to various stimuli; for instance, NAD(P)H oxidase (Nox) subunits are aggregated or recruited in these platforms, increasing Nox activity. Superoxide and hydrogen peroxide generation could induce various regulatory activities, such as the induction of glucose transport activity and proliferation in leukaemia cells. The aim of our study is to probe: i) the involvement of lipid rafts in the modulation of the glucose transporter Glut1 in human acute leukemia cells; ii) the involvement of plasma membrane caveolae/lipid rafts in VEGF-mediated redox signaling via Nox activation in human leukemic cells; iii) the role of p66shc, an adaptor protein, in VEGF signaling and ROS production in endothelial cells (ECs); iv) the role of Sindecan-2, a transmembrane heparan sulphate proteoglycan, in VEGF signaling and physiological response in ECs and v) the antioxidant and pro-apoptotic activities of simple dietary phenolic acids, i. e. caffeic, syringic and protocatechuic acids in leukemia cells, characterized by a very high ROS content. Our results suggest that the role played by NAD(P)H oxidase-derived ROS in the regulation of glucose uptake, proliferation and migration of leukaemia and endothelial cells could likely occur through the control of lipid raft-associated signalling.

Reinforcement Lernen mit Regularisierungsnetzwerken

Die Arbeit behandelt das Problem der Skalierbarkeit von Reinforcement Lernen auf hochdimensionale und komplexe Aufgabenstellungen. Unter Reinforcement Lernen versteht man dabei eine auf approximativem Dynamischen Programmieren basierende Klasse von Lernverfahren, die speziell Anwendung in der Künstlichen Intelligenz findet und zur autonomen Steuerung simulierter Agenten oder realer Hardwareroboter in dynamischen und unwägbaren Umwelten genutzt werden kann. Dazu wird mittels Regression aus Stichproben eine Funktion bestimmt, die die Lösung einer "Optimalitätsgleichung" (Bellman) ist und aus der sich näherungsweise optimale Entscheidungen ableiten lassen. Eine große Hürde stellt dabei die Dimensionalität des Zustandsraums dar, die häufig hoch und daher traditionellen gitterbasierten Approximationsverfahren wenig zugänglich ist. Das Ziel dieser Arbeit ist es, Reinforcement Lernen durch nichtparametrisierte Funktionsapproximation (genauer, Regularisierungsnetze) auf -- im Prinzip beliebig -- hochdimensionale Probleme anwendbar zu machen. Regularisierungsnetze sind eine Verallgemeinerung von gewöhnlichen Basisfunktionsnetzen, die die gesuchte Lösung durch die Daten parametrisieren, wodurch die explizite Wahl von Knoten/Basisfunktionen entfällt und so bei hochdimensionalen Eingaben der "Fluch der Dimension" umgangen werden kann. Gleichzeitig sind Regularisierungsnetze aber auch lineare Approximatoren, die technisch einfach handhabbar sind und für die die bestehenden Konvergenzaussagen von Reinforcement Lernen Gültigkeit behalten (anders als etwa bei Feed-Forward Neuronalen Netzen). Allen diesen theoretischen Vorteilen gegenüber steht allerdings ein sehr praktisches Problem: der Rechenaufwand bei der Verwendung von Regularisierungsnetzen skaliert von Natur aus wie O(n**3), wobei n die Anzahl der Daten ist. Das ist besonders deswegen problematisch, weil bei Reinforcement Lernen der Lernprozeß online erfolgt -- die Stichproben werden von einem Agenten/Roboter erzeugt, während er mit der Umwelt interagiert. Anpassungen an der Lösung müssen daher sofort und mit wenig Rechenaufwand vorgenommen werden. Der Beitrag dieser Arbeit gliedert sich daher in zwei Teile: Im ersten Teil der Arbeit formulieren wir für Regularisierungsnetze einen effizienten Lernalgorithmus zum Lösen allgemeiner Regressionsaufgaben, der speziell auf die Anforderungen von Online-Lernen zugeschnitten ist. Unser Ansatz basiert auf der Vorgehensweise von Recursive Least-Squares, kann aber mit konstantem Zeitaufwand nicht nur neue Daten sondern auch neue Basisfunktionen in das bestehende Modell einfügen. Ermöglicht wird das durch die "Subset of Regressors" Approximation, wodurch der Kern durch eine stark reduzierte Auswahl von Trainingsdaten approximiert wird, und einer gierigen Auswahlwahlprozedur, die diese Basiselemente direkt aus dem Datenstrom zur Laufzeit selektiert. Im zweiten Teil übertragen wir diesen Algorithmus auf approximative Politik-Evaluation mittels Least-Squares basiertem Temporal-Difference Lernen, und integrieren diesen Baustein in ein Gesamtsystem zum autonomen Lernen von optimalem Verhalten. Insgesamt entwickeln wir ein in hohem Maße dateneffizientes Verfahren, das insbesondere für Lernprobleme aus der Robotik mit kontinuierlichen und hochdimensionalen Zustandsräumen sowie stochastischen Zustandsübergängen geeignet ist. Dabei sind wir nicht auf ein Modell der Umwelt angewiesen, arbeiten weitestgehend unabhängig von der Dimension des Zustandsraums, erzielen Konvergenz bereits mit relativ wenigen Agent-Umwelt Interaktionen, und können dank des effizienten Online-Algorithmus auch im Kontext zeitkritischer Echtzeitanwendungen operieren. Wir demonstrieren die Leistungsfähigkeit unseres Ansatzes anhand von zwei realistischen und komplexen Anwendungsbeispielen: dem Problem RoboCup-Keepaway, sowie der Steuerung eines (simulierten) Oktopus-Tentakels.

Modularità: Dai sistemi biologici ai sistemi artificiali

I sistemi di intelligenza artificiale vengono spesso messi a confronto con gli aspetti biologici riguardanti il cervello umano. L’interesse per la modularità è in continua crescita, che sta portando a risultati davvero interessanti attraverso l’utilizzo di sistemi artificiali intelligenti, come le reti neuronali. Molte reti, sia biologiche sia artificiali sono organizzate in moduli, i quali rappresentano cluster densi di parti interconnesse fra loro all’interno di una rete complessa. Nel campo dell’ingegneria, si usano design modulari per spiegare come una macchina è costituita da parti separate. Lo studio della struttura e delle funzioni di organismi/processi complessi si basa implicitamente su un principio di organizzazione modulare, vale a dire si dà per acquisito il fatto che siano modulari, cioè composti da parti con forma e/o funzioni diverse. Questo elaborato si propone di esporre gli aspetti fondamentali riguardanti la modularità di reti neuronali, le sue origini evolutive, le condizioni necessarie o sufficienti che favoriscono l’emergere dei moduli e relativi vantaggi. Il primo capitolo fornisce alcune conoscenze di base che permettono di leggere gli esperimenti delle parti successive con consapevolezza teorica più profonda. Si descrivono reti neuronali artificiali come sistemi intelligenti ispirati alla struttura di reti neurali biologiche, soffermandosi in particolare sulla rete feed-forward, sull’algoritmo di backpropagation e su modelli di reti neurali modulari. Il secondo capitolo offre una visione delle origini evolutive della modularità e dei meccanismi evolutivi riguardanti sistemi biologici, una classificazione dei vati tipi di modularità, esplorando il concetto di modularità nell’ambito della psicologia cognitiva. Si analizzano i campi disciplinari a cui questa ricerca di modularità può portare vantaggi. Dal terzo capitolo che inizia a costituire il corpo centrale dell’elaborato, si dà importanza alla modularità nei sistemi computazionali, illustrando alcuni casi di studio e relativi metodi presenti in letteratura e fornendo anche una misura quantitativa della modularità. Si esaminano le varie possibilità di evoluzione della modularità: spontanea, da specializzazione, duplicazione, task-dependent, ecc. passando a emulare l’evoluzione di sistemi neurali modulari con applicazione al noto modello “What-Where” e a vari modelli con caratteristiche diverse. Si elencano i vantaggi che la modularità produce, soffermandosi sull’algoritmo di apprendimento, sugli ambienti che favoriscono l’evoluzione della modularità con una serie di confronti fra i vari tipi, statici e dinamici. In ultimo, come il vantaggio di avere connessioni corte possa portare a sviluppare modularità. L’obiettivo comune è l’emergere della modularità in sistemi neuronali artificiali, che sono usati per applicazioni in numerosi ambiti tecnologici.

A multifactorial analysis of obesity as CVD risk factor: use of neural network based methods in a nutrigenetics context

Obesity is a multifactorial trait, which comprises an independent risk factor for cardiovascular disease (CVD). The aim of the current work is to study the complex etiology beneath obesity and identify genetic variations and/or factors related to nutrition that contribute to its variability. To this end, a set of more than 2300 white subjects who participated in a nutrigenetics study was used. For each subject a total of 63 factors describing genetic variants related to CVD (24 in total), gender, and nutrition (38 in total), e.g. average daily intake in calories and cholesterol, were measured. Each subject was categorized according to body mass index (BMI) as normal (BMI ≤ 25) or overweight (BMI > 25). Two artificial neural network (ANN) based methods were designed and used towards the analysis of the available data. These corresponded to i) a multi-layer feed-forward ANN combined with a parameter decreasing method (PDM-ANN), and ii) a multi-layer feed-forward ANN trained by a hybrid method (GA-ANN) which combines genetic algorithms and the popular back-propagation training algorithm.

Molecular mechanisms of muscle plasticity with exercise

The skeletal muscle phenotype is subject to considerable malleability depending on use. Low-intensity endurance type exercise leads to qualitative changes of muscle tissue characterized mainly by an increase in structures supporting oxygen delivery and consumption. High-load strength-type exercise leads to growth of muscle fibers dominated by an increase in contractile proteins. In low-intensity exercise, stress-induced signaling leads to transcriptional upregulation of a multitude of genes with Ca2+ signaling and the energy status of the muscle cells sensed through AMPK being major input determinants. Several parallel signaling pathways converge on the transcriptional co-activator PGC-1α, perceived as being the coordinator of much of the transcriptional and posttranscriptional processes. High-load training is dominated by a translational upregulation controlled by mTOR mainly influenced by an insulin/growth factor-dependent signaling cascade as well as mechanical and nutritional cues. Exercise-induced muscle growth is further supported by DNA recruitment through activation and incorporation of satellite cells. Crucial nodes of strength and endurance exercise signaling networks are shared making these training modes interdependent. Robustness of exercise-related signaling is the consequence of signaling being multiple parallel with feed-back and feed-forward control over single and multiple signaling levels. We currently have a good descriptive understanding of the molecular mechanisms controlling muscle phenotypic plasticity. We lack understanding of the precise interactions among partners of signaling networks and accordingly models to predict signaling outcome of entire networks. A major current challenge is to verify and apply available knowledge gained in model systems to predict human phenotypic plasticity.

Application and comparison of classification algorithms for recognition of Alzheimer's disease in electrical brain activity (EEG)

The early detection of subjects with probable Alzheimer's disease (AD) is crucial for effective appliance of treatment strategies. Here we explored the ability of a multitude of linear and non-linear classification algorithms to discriminate between the electroencephalograms (EEGs) of patients with varying degree of AD and their age-matched control subjects. Absolute and relative spectral power, distribution of spectral power, and measures of spatial synchronization were calculated from recordings of resting eyes-closed continuous EEGs of 45 healthy controls, 116 patients with mild AD and 81 patients with moderate AD, recruited in two different centers (Stockholm, New York). The applied classification algorithms were: principal component linear discriminant analysis (PC LDA), partial least squares LDA (PLS LDA), principal component logistic regression (PC LR), partial least squares logistic regression (PLS LR), bagging, random forest, support vector machines (SVM) and feed-forward neural network. Based on 10-fold cross-validation runs it could be demonstrated that even tough modern computer-intensive classification algorithms such as random forests, SVM and neural networks show a slight superiority, more classical classification algorithms performed nearly equally well. Using random forests classification a considerable sensitivity of up to 85% and a specificity of 78%, respectively for the test of even only mild AD patients has been reached, whereas for the comparison of moderate AD vs. controls, using SVM and neural networks, values of 89% and 88% for sensitivity and specificity were achieved. Such a remarkable performance proves the value of these classification algorithms for clinical diagnostics.

Evaluation of texture features in hepatic tissue characterization from non-enhanced CT images

Aim of this paper is to evaluate the diagnostic contribution of various types of texture features in discrimination of hepatic tissue in abdominal non-enhanced Computed Tomography (CT) images. Regions of Interest (ROIs) corresponding to the classes: normal liver, cyst, hemangioma, and hepatocellular carcinoma were drawn by an experienced radiologist. For each ROI, five distinct sets of texture features are extracted using First Order Statistics (FOS), Spatial Gray Level Dependence Matrix (SGLDM), Gray Level Difference Method (GLDM), Laws' Texture Energy Measures (TEM), and Fractal Dimension Measurements (FDM). In order to evaluate the ability of the texture features to discriminate the various types of hepatic tissue, each set of texture features, or its reduced version after genetic algorithm based feature selection, was fed to a feed-forward Neural Network (NN) classifier. For each NN, the area under Receiver Operating Characteristic (ROC) curves (Az) was calculated for all one-vs-all discriminations of hepatic tissue. Additionally, the total Az for the multi-class discrimination task was estimated. The results show that features derived from FOS perform better than other texture features (total Az: 0.802+/-0.083) in the discrimination of hepatic tissue.

Neural network based glucose - insulin metabolism models for children with Type 1 diabetes

In this paper two models for the simulation of glucose-insulin metabolism of children with Type 1 diabetes are presented. The models are based on the combined use of Compartmental Models (CMs) and artificial Neural Networks (NNs). Data from children with Type 1 diabetes, stored in a database, have been used as input to the models. The data are taken from four children with Type 1 diabetes and contain information about glucose levels taken from continuous glucose monitoring system, insulin intake and food intake, along with corresponding time. The influences of taken insulin on plasma insulin concentration, as well as the effect of food intake on glucose input into the blood from the gut, are estimated from the CMs. The outputs of CMs, along with previous glucose measurements, are fed to a NN, which provides short-term prediction of glucose values. For comparative reasons two different NN architectures have been tested: a Feed-Forward NN (FFNN) trained with the back-propagation algorithm with adaptive learning rate and momentum, and a Recurrent NN (RNN), trained with the Real Time Recurrent Learning (RTRL) algorithm. The results indicate that the best prediction performance can be achieved by the use of RNN.

A computer-aided diagnostic system to characterize CT focal liver lesions: design and optimization of a neural network classifier

In this paper, a computer-aided diagnostic (CAD) system for the classification of hepatic lesions from computed tomography (CT) images is presented. Regions of interest (ROIs) taken from nonenhanced CT images of normal liver, hepatic cysts, hemangiomas, and hepatocellular carcinomas have been used as input to the system. The proposed system consists of two modules: the feature extraction and the classification modules. The feature extraction module calculates the average gray level and 48 texture characteristics, which are derived from the spatial gray-level co-occurrence matrices, obtained from the ROIs. The classifier module consists of three sequentially placed feed-forward neural networks (NNs). The first NN classifies into normal or pathological liver regions. The pathological liver regions are characterized by the second NN as cyst or "other disease." The third NN classifies "other disease" into hemangioma or hepatocellular carcinoma. Three feature selection techniques have been applied to each individual NN: the sequential forward selection, the sequential floating forward selection, and a genetic algorithm for feature selection. The comparative study of the above dimensionality reduction methods shows that genetic algorithms result in lower dimension feature vectors and improved classification performance.

Zukunftspotential durch Design und RM

Im Bereich Medical Design spielen eine hohe Funktionalität und eine angenehme Handhabung eine grosse Rolle. In den Engineering-Projekten der Erdmann Design AG werden kühne Ideen in den Raum gesetzt die das Klima für Innovationen fördern. „Innovation entsteht aus elementarer Neugier und Lust auf eine bessere Zukunft“, sagt Raimund Erdmann. „Wir beraten mit unseren Feed Forward Methoden Start-up Firmen, Kleinunternehmen wie auch etliche, börsenkotierte, international aktive Unternehmen. Und das für excellentes Industrial Design, Corporate Design und Design Management“ mit den überzeugenden Möglichkeiten des Rapid Manufacturing.Im Bereich Medical Design spielen eine hohe Funktionalität und eine angenehme Handhabung eine grosse Rolle. In den Engineering-Projekten der Erdmann Design AG werden kühne Ideen in den Raum gesetzt die das Klima für Innovationen fördern. „Innovation entsteht aus elementarer Neugier und Lust auf eine bessere Zukunft“, sagt Raimund Erdmann. „Wir beraten mit unseren Feed Forward Methoden Start-up Firmen, Kleinunternehmen wie auch etliche, börsenkotierte, international aktive Unternehmen. Und das für excellentes Industrial Design, Corporate Design und Design Management“ mit den überzeugenden Möglichkeiten des Rapid Manufacturing.

Outline of a theory of efficient teams in Sport.

Introduction So far, social psychology in sport has preliminary focused on team cohesion, and many studies and meta-analyses tried to demonstrate a relation between cohesiveness of a team and its performance. How a team really co-operates and how the individual actions are integrated towards a team action is a question that has received relatively little attention in research. This may, at least in part, be due to a lack of a theoretical framework for collective actions, a dearth that has only recently begun to challenge sport psychologists. Objectives In this presentation a framework for a comprehensive theory of teams in sport is outlined and its potential to integrate research in the domain of team performance and, more specifically, the following presentations, is put up for discussion. Method Based on a model developed by von Cranach, Ochsenbein and Valach (1986), teams are considered to be information processing organisms, and team actions need to be investigated on two levels: the individual team member and the group as an entity. Elements to be considered are the task, the social structure, the information processing structure and the execution structure. Obviously, different task require different social structures, communication processes and co-ordination of individual movements. Especially in rapid interactive sports planning and execution of movements based on feedback loops are not possible. Deliberate planning may be a solution mainly for offensive actions, whereas defensive actions have to adjust to the opponent team's actions. Consequently, mental representations must be developed to allow a feed-forward regulation of team member's actions. Results and Conclusions Some preliminary findings based on this conceptual framework as well as further consequences for empirical investigations will be presented. References Cranach, M.v., Ochsenbein, G. & Valach, L. (1986). The group as a self-active system: Outline of a theory of group action. European Journal of Social Psychology, 16, 193-229.

Fibulin-2 stabilizes tumor extracellular matrix and drives malignant progression of lung adenocarcinoma

The ECM of epithelial carcinomas undergoes structural remodeling during periods of uncontrolled growth, creating regional heterogeneity and torsional stress. How tumors maintain ECM integrity in the face of dynamic biophysical forces is still largely unclear. This study addresses these deficiencies using mouse models of human lung adenocarcinoma. Spontaneous lung tumors were marked by disorganized basement membranes, dense collagen networks, and increased tissue stiffness. Metastasis-prone lung adenocarcinoma cells secreted fibulin-2 (Fbln2), a matrix glycoprotein involved in ECM supra-molecular assembly. Fibulin-2 depletion in tumor cells decreased the intra-tumoral abundance of matrix metalloproteinases and reduced collagen cross-linking and tumor compressive properties resulting in inhibited tumor growth and metastasis. Fbln2 deposition within intra-tumoral fibrotic bands was a predictor of poor clinical outcome in patients. Collectively, these findings support a feed-forward model in which tumor cells secrete matrix-stabilizing factors required for the assembly of ECM that preferentially favors malignant progression. To our knowledge, this is the first evidence that tumor cells directly regulate the integrity of their surrounding matrix through the secretion of matrix-stabilizing factors such as fibulin-2. These findings open a new avenue of research into matrix assembly molecules as potential therapeutic targets in cancer patients.

The Role Of IL-8-Mediated Src Family Kinase Activation In Tumor-Tumor And Tumor-Stromal Interactions In Metastasis Of Prostate Cancer

Men with localized prostate cancer (PCa) have a 100% five-year survival rate, but this rate drops to 33% for men with metastatic disease. A better understanding of the metastatic process is needed to develop better therapies for PCa. Aberrant activation of protein tyrosine kinases, including Src Family Kinases (SFKs) contribute to metastasis through numerous functions, one of which leads to increased expression of cytokines, such as IL-8. However, the relationship between Src activity and IL-8 regulation is not completely understood. In cell line models, I determined that IL-8 activates Src and in turn Src activates IL-8 demonstrating a feed forward loop contributing to the migration and invasion of PCa cells. However, IL-8 is also produced by tumor-associated stromal cells. In bone marrow derived stromal cells (HS5), I demonstrated a feed forward loop occurs as was observed in tumor cells. HS5 conditioned media increased Src activity in PCa cells. By silencing IL-8 in HS5 cells, Src activity was decreased to control levels in PCa cells as was migration and invasion. Thus, stromal cells producing IL-8 contribute to metastatic properties of PCa by a paracrine mechanism. To examine the effect of stromal cells on tumor growth and metastatic potential of PCa in vivo, I mixed HS5 and PCa cells and co-injected them intraprostatically. I determined that tumor growth and metastases were increased. By silencing IL-8 in HS5 cells and co-injecting them with PCa cells intraprostatically, tumor growth and metastases were still increased relative to injection of PCa cells alone, but decreased relative to co-injections with PCa cells and HS5 cells. These studies demonstrated: (1) a feed forward loop in both tumor and stromal cells, whereby IL-8 activates Src, derepressing IL-8 expression in PCa cells in vitro; (2) stromal produced IL-8 activates Src and contributes to the migration and invasion of PCa cells in vitro; and (3) stromal produced IL-8 is responsible, in part, for increases in PCa tumor growth and metastatic potential. Together, these studies demonstrated that IL-8-mediated Src activity increases the metastatic potential of PCa and therapeutic agents interfering with the IL-8/SFK signaling axis may be useful for prevention and treatment of metastases.

Acceleration of the PanIN Development in Mice Expressing Oncogenic K-Ras Due to a High Fat Diet

Obesity is postulated to be one of the major risk factors for pancreatic cancer, and recently it was indicated that an elevated body mass index (BMI correlates strongly with a decrease in patient survival. Despite the evident relationship, the molecular mechanisms involved are unclear. Oncogenic mutation of K-Ras is found early and is universal in pancreatic cancer. Extensive evidence indicates oncogenic K-Ras is not entirely active and it requires a triggering event to surpass the activity of Ras beyond the threshold necessary for a Ras-inflammation feed-forward loop. We hypothesize that high fat intake induces a persistent low level inflammatory response triggering increased K-Ras activity and that Cox-2 is essential for this inflammatory reaction. To determine this, LSL-K-Ras mice were crossed with Ela-CreER (Acinar-specific) or Pdx-1-Cre (Pancreas-specific) to “knock-in” oncogenic K-Ras. Additionally, these animals were crossed with Cox-2 conditional knockout mice to access the importance of Cox-2 in the inflammatory loop present. The mice were fed isocaloric diets containing 60% energy or 10% energy from fat. We found that a high fat diet increased K-Ras activity, PanIN formation, and fibrotic stroma significantly compared to a control diet. Genetic deletion of Cox-2 prevented high fat diet induced fibrosis and PanIN formation in oncogenic K-Ras expressing mice. Additionally, long term consumption of high fat diet, increased the progression of PanIN lesions leading to invasive cancer and decreased overall survival rate. These findings indicate that a high fat diet can stimulate the activation of oncogenic K-Ras and initiate an inflammatory feed forward loop requiring Cox-2 leading to inflammation, fibrosis, and PanINs. This mechanism could explain the relationship between a high fat diet and elevated risk for pancreatic cancer.

Codificador de voz MIDI

El presente proyecto tiene el objetivo de facilitar la composición de canciones mediante la creación de las distintas pistas MIDI que la forman. Se implementan dos controladores. El primero, con objeto de transcribir la parte melódica, convierte la voz cantada o tarareada a eventos MIDI. Para ello, y tras el estudio de las distintas técnicas del cálculo del tono (pitch), se implementará una técnica con ciertas variaciones basada en la autocorrelación. También se profundiza en el segmentado de eventos, en particular, una técnica basada en el análisis de la derivada de la envolvente. El segundo, dedicado a la base rítmica de la canción, permite la creación de la percusión mediante el golpe rítmico de objetos que disponga el usuario, que serán asignados a los distintos elementos de percusión elegidos. Los resultados de la grabación de estos impactos serán señales de corta duración, no lineales y no armónicas, dificultando su discriminación. La herramienta elegida para la clasificación de los distintos patrones serán las redes neuronales artificiales (RNA). Se realizara un estudio de la metodología de diseño de redes neuronales especifico para este tipo de señales, evaluando la importancia de las variables de diseño como son el número de capas ocultas y neuronas en cada una de ellas, algoritmo de entrenamiento y funciones de activación. El estudio concluirá con la implementación de dos redes de diferente naturaleza. Una red de Elman, cuyas propiedades de memoria permiten la clasificación de patrones temporales, procesará las cualidades temporales analizando el ataque de su forma de onda. Una red de propagación hacia adelante feed-forward, que necesitará de robustas características espectrales y temporales para su clasificación. Se proponen 26 descriptores como los derivados de los momentos del espectro: centroide, curtosis y simetría, los coeficientes cepstrales de la escala de Mel (MFCCs), y algunos temporales como son la tasa de cruces por cero y el centroide de la envolvente temporal. Las capacidades de discriminación inter e intra clase de estas características serán evaluadas mediante un algoritmo de selección, habiéndose elegido RELIEF, un método basado en el algoritmo de los k vecinos mas próximos (KNN). Ambos controladores tendrán función de trabajar en tiempo real y offline, permitiendo tanto la composición de canciones, como su utilización como un instrumento más junto con mas músicos. ABSTRACT. The aim of this project is to make song composition easier by creating each MIDI track that builds it. Two controllers are implemented. In order to transcribe the melody, the first controler converts singing voice or humming into MIDI files. To do this a technique based on autocorrelation is implemented after having studied different pitch detection methods. Event segmentation has also been dealt with, to be more precise a technique based on the analysis of the signal's envelope and it's derivative have been used. The second one, can be used to make the song's rhythm . It allows the user, to create percussive patterns by hitting different objects of his environment. These recordings results in short duration, non-linear and non-harmonic signals. Which makes the classification process more complicated in the traditional way. The tools to used are the artificial neural networks (ANN). We will study the neural network design to deal with this kind of signals. The goal is to get a design methodology, paying attention to the variables involved, as the number of hidden layers and neurons in each, transfer functions and training algorithm. The study will end implementing two neural networks with different nature. Elman network, which has memory properties, is capable to recognize sequences of data and analyse the impact's waveform, precisely, the attack portion. A feed-forward network, needs strong spectral and temporal features extracted from the hit. Some descriptors are proposed as the derivates from the spectrum moment as centroid, kurtosis and skewness, the Mel-frequency cepstral coefficients, and some temporal features as the zero crossing rate (zcr) and the temporal envelope's centroid. Intra and inter class discrimination abilities of those descriptors will be weighted using the selection algorithm RELIEF, a Knn (K-nearest neighbor) based algorithm. Both MIDI controllers can be used to compose, or play with other musicians as it works on real-time and offline.