Simulating event related potentials in tasks involving conflict resolution: a neural network

In this thesis, the main Executive Control theories are exposed. Methods typical of Cognitive and Computational Neuroscience are introduced and the role of behavioural tasks involving conflict resolution in the response elaboration, after the presentation of a stimulus to the subject, are highlighted. In particular, the Eriksen Flanker Task and its variants are discussed. Behavioural data, from scientific literature, are illustrated in terms of response times and error rates. During experimental behavioural tasks, EEG is registered simultaneously. Thanks to this, event related potential, related with the current task, can be studied. Different theories regarding relevant event related potential in this field - such as N2, fERN (feedback Error Related Negativity) and ERN (Error Related Negativity) – are introduced. The aim of this thesis is to understand and simulate processes regarding Executive Control, including performance improvement, error detection mechanisms, post error adjustments and the role of selective attention, with the help of an original neural network model. The network described here has been built with the purpose to simulate behavioural results of a four choice Eriksen Flanker Task. Model results show that the neural network can simulate response times, error rates and event related potentials quite well. Finally, results are compared with behavioural data and discussed in light of the mentioned Executive Control theories. Future perspective for this new model are outlined.

Deep learning for computer vision: a comparison between convolutional neural networks and hierarchical temporal memories on object recognition tasks

In recent years, Deep Learning techniques have shown to perform well on a large variety of problems both in Computer Vision and Natural Language Processing, reaching and often surpassing the state of the art on many tasks. The rise of deep learning is also revolutionizing the entire field of Machine Learning and Pattern Recognition pushing forward the concepts of automatic feature extraction and unsupervised learning in general. However, despite the strong success both in science and business, deep learning has its own limitations. It is often questioned if such techniques are only some kind of brute-force statistical approaches and if they can only work in the context of High Performance Computing with tons of data. Another important question is whether they are really biologically inspired, as claimed in certain cases, and if they can scale well in terms of "intelligence". The dissertation is focused on trying to answer these key questions in the context of Computer Vision and, in particular, Object Recognition, a task that has been heavily revolutionized by recent advances in the field. Practically speaking, these answers are based on an exhaustive comparison between two, very different, deep learning techniques on the aforementioned task: Convolutional Neural Network (CNN) and Hierarchical Temporal memory (HTM). They stand for two different approaches and points of view within the big hat of deep learning and are the best choices to understand and point out strengths and weaknesses of each of them. CNN is considered one of the most classic and powerful supervised methods used today in machine learning and pattern recognition, especially in object recognition. CNNs are well received and accepted by the scientific community and are already deployed in large corporation like Google and Facebook for solving face recognition and image auto-tagging problems. HTM, on the other hand, is known as a new emerging paradigm and a new meanly-unsupervised method, that is more biologically inspired. It tries to gain more insights from the computational neuroscience community in order to incorporate concepts like time, context and attention during the learning process which are typical of the human brain. In the end, the thesis is supposed to prove that in certain cases, with a lower quantity of data, HTM can outperform CNN.

Parametric Sensitivity Analysis of the Most Recent Computational Models of Rabbit Cardiac Pacemaking

The cellular basis of cardiac pacemaking activity, and specifically the quantitative contributions of particular mechanisms, is still debated. Reliable computational models of sinoatrial nodal (SAN) cells may provide mechanistic insights, but competing models are built from different data sets and with different underlying assumptions. To understand quantitative differences between alternative models, we performed thorough parameter sensitivity analyses of the SAN models of Maltsev & Lakatta (2009) and Severi et al (2012). Model parameters were randomized to generate a population of cell models with different properties, simulations performed with each set of random parameters generated 14 quantitative outputs that characterized cellular activity, and regression methods were used to analyze the population behavior. Clear differences between the two models were observed at every step of the analysis. Specifically: (1) SR Ca2+ pump activity had a greater effect on SAN cell cycle length (CL) in the Maltsev model; (2) conversely, parameters describing the funny current (If) had a greater effect on CL in the Severi model; (3) changes in rapid delayed rectifier conductance (GKr) had opposite effects on action potential amplitude in the two models; (4) within the population, a greater percentage of model cells failed to exhibit action potentials in the Maltsev model (27%) compared with the Severi model (7%), implying greater robustness in the latter; (5) confirming this initial impression, bifurcation analyses indicated that smaller relative changes in GKr or Na+-K+ pump activity led to failed action potentials in the Maltsev model. Overall, the results suggest experimental tests that can distinguish between models and alternative hypotheses, and the analysis offers strategies for developing anti-arrhythmic pharmaceuticals by predicting their effect on the pacemaking activity.

Oxidation and pyrolysis of methane and methyl formate in a plug flow reactor: computational and experimenal studies

Biodiesel represents a possible substitute to the fossil fuels; for this reason a good comprehension of the kinetics involved is important. Due to the complexity of the biodiesel mixture a common practice is the use of surrogate molecules to study its reactivity. In this work are presented the experimental and computational results obtained for the oxidation and pyrolysis of methane and methyl formate conducted in a plug flow reactor. The work was divided into two parts: the first one was the setup assembly whilst, in the second one, was realized a comparison between the experimental and model results; these last was obtained using models available in literature. It was started studying the methane since, a validate model was available, in this way was possible to verify the reliability of the experimental results. After this first study the attention was focused on the methyl formate investigation. All the analysis were conducted at different temperatures, pressures and, for the oxidation, at different equivalence ratios. The results shown that, a good comprehension of the kinetics is reach but efforts are necessary to better evaluate kinetics parameters such as activation energy. The results even point out that the realized setup is adapt to study the oxidation and pyrolysis and, for this reason, it will be employed to study a longer chain esters with the aim to better understand the kinetic of the molecules that are part of the biodiesel mixture.

Experimental and computational study of methane and methyl formate pyrolysis and oxidation in a flow reactor

A study of the pyrolysis and oxidation (phi 0.5-1-2) of methane and methyl formate (phi 0.5) in a laboratory flow reactor (Length = 50 cm, inner diameter = 2.5 cm) has been carried out at 1-4 atm and 300-1300 K temperature range. Exhaust gaseous species analysis was realized using a gas chromatographic system, Varian CP-4900 PRO Mirco-GC, with a TCD detector and using helium as carrier for a Molecular Sieve 5Å column and nitrogen for a COX column, whose temperatures and pressures were respectively of 65°C and 150kPa. Model simulations using NTUA [1], Fisher et al. [12], Grana [13] and Dooley [14] kinetic mechanisms have been performed with CHEMKIN. The work provides a basis for further development and optimization of existing detailed chemical kinetic schemes.

"Experimental and computational approaches to historical masonry structures: A study in the direction of filling the gap"

The assessment of historical structures is a significant need for the next generations, as historical monuments represent the community’s identity and have an important cultural value to society. Most of historical structures built by using masonry which is one of the oldest and most common construction materials used in the building sector since the ancient time. Also it is considered a complex material, as it is a composition of brick units and mortar, which affects the structural performance of the building by having different mechanical behaviour with respect to different geometry and qualities given by the components.

A 2-dimensional computational model to analyze the effects of cellular heterogeinity on cardiac pacemaking

The mechanical action of the heart is made possible in response to electrical events that involve the cardiac cells, a property that classifies the heart tissue between the excitable tissues. At the cellular level, the electrical event is the signal that triggers the mechanical contraction, inducing a transient increase in intracellular calcium which, in turn, carries the message of contraction to the contractile proteins of the cell. The primary goal of my project was to implement in CUDA (Compute Unified Device Architecture, an hardware architecture for parallel processing created by NVIDIA) a tissue model of the rabbit sinoatrial node to evaluate the heterogeneity of its structure and how that variability influences the behavior of the cells. In particular, each cell has an intrinsic discharge frequency, thus different from that of every other cell of the tissue and it is interesting to study the process of synchronization of the cells and look at the value of the last discharge frequency if they synchronized.

Aggregate Programming in Scala: a Core Library and Actor-Based Platform for Distributed Computational Fields

La programmazione aggregata è un paradigma che supporta la programmazione di sistemi di dispositivi, adattativi ed eventualmente a larga scala, nel loro insieme -- come aggregati. L'approccio prevalente in questo contesto è basato sul field calculus, un calcolo formale che consente di definire programmi aggregati attraverso la composizione funzionale di campi computazionali, creando i presupposti per la specifica di pattern di auto-organizzazione robusti. La programmazione aggregata è attualmente supportata, in modo più o meno parziale e principalmente per la simulazione, da DSL dedicati (cf., Protelis), ma non esistono framework per linguaggi mainstream finalizzati allo sviluppo di applicazioni. Eppure, un simile supporto sarebbe auspicabile per ridurre tempi e sforzi d'adozione e per semplificare l'accesso al paradigma nella costruzione di sistemi reali, nonché per favorire la ricerca stessa nel campo. Il presente lavoro consiste nello sviluppo, a partire da un prototipo della semantica operazionale del field calculus, di un framework per la programmazione aggregata in Scala. La scelta di Scala come linguaggio host nasce da motivi tecnici e pratici. Scala è un linguaggio moderno, interoperabile con Java, che ben integra i paradigmi ad oggetti e funzionale, ha un sistema di tipi espressivo, e fornisce funzionalità avanzate per lo sviluppo di librerie e DSL. Inoltre, la possibilità di appoggiarsi, su Scala, ad un framework ad attori solido come Akka, costituisce un altro fattore trainante, data la necessità di colmare l'abstraction gap inerente allo sviluppo di un middleware distribuito. Nell'elaborato di tesi si presenta un framework che raggiunge il triplice obiettivo: la costruzione di una libreria Scala che realizza la semantica del field calculus in modo corretto e completo, la realizzazione di una piattaforma distribuita Akka-based su cui sviluppare applicazioni, e l'esposizione di un'API generale e flessibile in grado di supportare diversi scenari.