Bayesian Learning with Dependency Structures via Latent Factors, Mixtures, and Copulas

Bayesian methods offer a flexible and convenient probabilistic learning framework to extract interpretable knowledge from complex and structured data. Such methods can characterize dependencies among multiple levels of hidden variables and share statistical strength across heterogeneous sources. In the first part of this dissertation, we develop two dependent variational inference methods for full posterior approximation in non-conjugate Bayesian models through hierarchical mixture- and copula-based variational proposals, respectively. The proposed methods move beyond the widely used factorized approximation to the posterior and provide generic applicability to a broad class of probabilistic models with minimal model-specific derivations. In the second part of this dissertation, we design probabilistic graphical models to accommodate multimodal data, describe dynamical behaviors and account for task heterogeneity. In particular, the sparse latent factor model is able to reveal common low-dimensional structures from high-dimensional data. We demonstrate the effectiveness of the proposed statistical learning methods on both synthetic and real-world data.

Anomaly Detection in Noisy Images

Finding rare events in multidimensional data is an important detection problem that has applications in many fields, such as risk estimation in insurance industry, finance, flood prediction, medical diagnosis, quality assurance, security, or safety in transportation. The occurrence of such anomalies is so infrequent that there is usually not enough training data to learn an accurate statistical model of the anomaly class. In some cases, such events may have never been observed, so the only information that is available is a set of normal samples and an assumed pairwise similarity function. Such metric may only be known up to a certain number of unspecified parameters, which would either need to be learned from training data, or fixed by a domain expert. Sometimes, the anomalous condition may be formulated algebraically, such as a measure exceeding a predefined threshold, but nuisance variables may complicate the estimation of such a measure. Change detection methods used in time series analysis are not easily extendable to the multidimensional case, where discontinuities are not localized to a single point. On the other hand, in higher dimensions, data exhibits more complex interdependencies, and there is redundancy that could be exploited to adaptively model the normal data. In the first part of this dissertation, we review the theoretical framework for anomaly detection in images and previous anomaly detection work done in the context of crack detection and detection of anomalous components in railway tracks. In the second part, we propose new anomaly detection algorithms. The fact that curvilinear discontinuities in images are sparse with respect to the frame of shearlets, allows us to pose this anomaly detection problem as basis pursuit optimization. Therefore, we pose the problem of detecting curvilinear anomalies in noisy textured images as a blind source separation problem under sparsity constraints, and propose an iterative shrinkage algorithm to solve it. Taking advantage of the parallel nature of this algorithm, we describe how this method can be accelerated using graphical processing units (GPU). Then, we propose a new method for finding defective components on railway tracks using cameras mounted on a train. We describe how to extract features and use a combination of classifiers to solve this problem. Then, we scale anomaly detection to bigger datasets with complex interdependencies. We show that the anomaly detection problem naturally fits in the multitask learning framework. The first task consists of learning a compact representation of the good samples, while the second task consists of learning the anomaly detector. Using deep convolutional neural networks, we show that it is possible to train a deep model with a limited number of anomalous examples. In sequential detection problems, the presence of time-variant nuisance parameters affect the detection performance. In the last part of this dissertation, we present a method for adaptively estimating the threshold of sequential detectors using Extreme Value Theory on a Bayesian framework. Finally, conclusions on the results obtained are provided, followed by a discussion of possible future work.

Robotic Sensing and Manipulation of Deformable Linear Objects with Learning-based methods

Nowadays robotic applications are widespread and most of the manipulation tasks are efficiently solved. However, Deformable-Objects (DOs) still represent a huge limitation for robots. The main difficulty in DOs manipulation is dealing with the shape and dynamics uncertainties, which prevents the use of model-based approaches (since they are excessively computationally complex) and makes sensory data difficult to interpret. This thesis reports the research activities aimed to address some applications in robotic manipulation and sensing of Deformable-Linear-Objects (DLOs), with particular focus to electric wires. In all the works, a significant effort was made in the study of an effective strategy for analyzing sensory signals with various machine learning algorithms. In the former part of the document, the main focus concerns the wire terminals, i.e. detection, grasping, and insertion. First, a pipeline that integrates vision and tactile sensing is developed, then further improvements are proposed for each module. A novel procedure is proposed to gather and label massive amounts of training images for object detection with minimal human intervention. Together with this strategy, we extend a generic object detector based on Convolutional-Neural-Networks for orientation prediction. The insertion task is also extended by developing a closed-loop control capable to guide the insertion of a longer and curved segment of wire through a hole, where the contact forces are estimated by means of a Recurrent-Neural-Network. In the latter part of the thesis, the interest shifts to the DLO shape. Robotic reshaping of a DLO is addressed by means of a sequence of pick-and-place primitives, while a decision making process driven by visual data learns the optimal grasping locations exploiting Deep Q-learning and finds the best releasing point. The success of the solution leverages on a reliable interpretation of the DLO shape. For this reason, further developments are made on the visual segmentation.

Reinforcement learning over encrypted data

Reinforcement learning is a particular paradigm of machine learning that, recently, has proved times and times again to be a very effective and powerful approach. On the other hand, cryptography usually takes the opposite direction. While machine learning aims at analyzing data, cryptography aims at maintaining its privacy by hiding such data. However, the two techniques can be jointly used to create privacy preserving models, able to make inferences on the data without leaking sensitive information. Despite the numerous amount of studies performed on machine learning and cryptography, reinforcement learning in particular has never been applied to such cases before. Being able to successfully make use of reinforcement learning in an encrypted scenario would allow us to create an agent that efficiently controls a system without providing it with full knowledge of the environment it is operating in, leading the way to many possible use cases. Therefore, we have decided to apply the reinforcement learning paradigm to encrypted data. In this project we have applied one of the most well-known reinforcement learning algorithms, called Deep Q-Learning, to simple simulated environments and studied how the encryption affects the training performance of the agent, in order to see if it is still able to learn how to behave even when the input data is no longer readable by humans. The results of this work highlight that the agent is still able to learn with no issues whatsoever in small state spaces with non-secure encryptions, like AES in ECB mode. For fixed environments, it is also able to reach a suboptimal solution even in the presence of secure modes, like AES in CBC mode, showing a significant improvement with respect to a random agent; however, its ability to generalize in stochastic environments or big state spaces suffers greatly.

Classificazione di Radiografie Toraciche per la Diagnosi del COVID-19 con Reti Convoluzionali e Vision Transformer

Negli ultimi due anni, per via della pandemia generata dal virus Covid19, la vita in ogni angolo del nostro pianeta è drasticamente cambiata. Ad oggi, nel mondo, sono oltre duecentoventi milioni le persone che hanno contratto questo virus e sono quasi cinque milioni le persone decedute. In alcuni periodi si è arrivati ad avere anche un milione di nuovi contagiati al giorno e mediamente, negli ultimi sei mesi, questo dato è stato di più di mezzo milione al giorno. Gli ospedali, soprattutto nei paesi meno sviluppati, hanno subito un grande stress e molte volte hanno avuto una carenza di risorse per fronteggiare questa grave pandemia. Per questo motivo ogni ricerca in questo campo diventa estremamente importante, soprattutto quelle che, con l'ausilio dell'intelligenza artificiale, riescono a dare supporto ai medici. Queste tecnologie una volta sviluppate e approvate possono essere diffuse a costi molto bassi e accessibili a tutti. In questo elaborato sono stati sperimentati e valutati due diversi approcci alla diagnosi del Covid-19 a partire dalle radiografie toraciche dei pazienti: il primo metodo si basa sul transfer learning di una rete convoluzionale inizialmente pensata per la classificazione di immagini. Il secondo approccio utilizza i Vision Transformer (ViT), un'architettura ampiamente diffusa nel campo del Natural Language Processing adattata ai task di Visione Artificiale. La prima soluzione ha ottenuto un’accuratezza di 0.85 mentre la seconda di 0.92, questi risultati, soprattutto il secondo, sono molto incoraggianti soprattutto vista la minima quantità di dati di training necessaria.

Sperimentazione di Deep Metric Loss per Self-Supervised Information Retrieval Systems su CORD19.

Dopo lo sviluppo dei primi casi di Covid-19 in Cina nell’autunno del 2019, ad inizio 2020 l’intero pianeta è precipitato in una pandemia globale che ha stravolto le nostre vite con conseguenze che non si vivevano dall’influenza spagnola. La grandissima quantità di paper scientifici in continua pubblicazione sul coronavirus e virus ad esso affini ha portato alla creazione di un unico dataset dinamico chiamato CORD19 e distribuito gratuitamente. Poter reperire informazioni utili in questa mole di dati ha ulteriormente acceso i riflettori sugli information retrieval systems, capaci di recuperare in maniera rapida ed efficace informazioni preziose rispetto a una domanda dell'utente detta query. Di particolare rilievo è stata la TREC-COVID Challenge, competizione per lo sviluppo di un sistema di IR addestrato e testato sul dataset CORD19. Il problema principale è dato dal fatto che la grande mole di documenti è totalmente non etichettata e risulta dunque impossibile addestrare modelli di reti neurali direttamente su di essi. Per aggirare il problema abbiamo messo a punto nuove soluzioni self-supervised, a cui abbiamo applicato lo stato dell'arte del deep metric learning e dell'NLP. Il deep metric learning, che sta avendo un enorme successo soprattuto nella computer vision, addestra il modello ad "avvicinare" tra loro immagini simili e "allontanare" immagini differenti. Dato che sia le immagini che il testo vengono rappresentati attraverso vettori di numeri reali (embeddings) si possano utilizzare le stesse tecniche per "avvicinare" tra loro elementi testuali pertinenti (e.g. una query e un paragrafo) e "allontanare" elementi non pertinenti. Abbiamo dunque addestrato un modello SciBERT con varie loss, che ad oggi rappresentano lo stato dell'arte del deep metric learning, in maniera completamente self-supervised direttamente e unicamente sul dataset CORD19, valutandolo poi sul set formale TREC-COVID attraverso un sistema di IR e ottenendo risultati interessanti.

Interpretable Convolutional Neural Networks for Decoding and Analyzing Neural Time Series Data

Machine learning is widely adopted to decode multi-variate neural time series, including electroencephalographic (EEG) and single-cell recordings. Recent solutions based on deep learning (DL) outperformed traditional decoders by automatically extracting relevant discriminative features from raw or minimally pre-processed signals. Convolutional Neural Networks (CNNs) have been successfully applied to EEG and are the most common DL-based EEG decoders in the state-of-the-art (SOA). However, the current research is affected by some limitations. SOA CNNs for EEG decoding usually exploit deep and heavy structures with the risk of overfitting small datasets, and architectures are often defined empirically. Furthermore, CNNs are mainly validated by designing within-subject decoders. Crucially, the automatically learned features mainly remain unexplored; conversely, interpreting these features may be of great value to use decoders also as analysis tools, highlighting neural signatures underlying the different decoded brain or behavioral states in a data-driven way. Lastly, SOA DL-based algorithms used to decode single-cell recordings rely on more complex, slower to train and less interpretable networks than CNNs, and the use of CNNs with these signals has not been investigated. This PhD research addresses the previous limitations, with reference to P300 and motor decoding from EEG, and motor decoding from single-neuron activity. CNNs were designed light, compact, and interpretable. Moreover, multiple training strategies were adopted, including transfer learning, which could reduce training times promoting the application of CNNs in practice. Furthermore, CNN-based EEG analyses were proposed to study neural features in the spatial, temporal and frequency domains, and proved to better highlight and enhance relevant neural features related to P300 and motor states than canonical EEG analyses. Remarkably, these analyses could be used, in perspective, to design novel EEG biomarkers for neurological or neurodevelopmental disorders. Lastly, CNNs were developed to decode single-neuron activity, providing a better compromise between performance and model complexity.

Supervised and weakly supervised counting-by-segmentation: the fluorescent microscopy use case

This thesis focuses on automating the time-consuming task of manually counting activated neurons in fluorescent microscopy images, which is used to study the mechanisms underlying torpor. The traditional method of manual annotation can introduce bias and delay the outcome of experiments, so the author investigates a deep-learning-based procedure to automatize this task. The author explores two of the main convolutional-neural-network (CNNs) state-of-the-art architectures: UNet and ResUnet family model, and uses a counting-by-segmentation strategy to provide a justification of the objects considered during the counting process. The author also explores a weakly-supervised learning strategy that exploits only dot annotations. The author quantifies the advantages in terms of data reduction and counting performance boost obtainable with a transfer-learning approach and, specifically, a fine-tuning procedure. The author released the dataset used for the supervised use case and all the pre-training models, and designed a web application to share both the counting process pipeline developed in this work and the models pre-trained on the dataset analyzed in this work.

Autonomous Pricing using Policy-Gradient Reinforcement Learning

Nella letteratura economica e di teoria dei giochi vi è un dibattito aperto sulla possibilità di emergenza di comportamenti anticompetitivi da parte di algoritmi di determinazione automatica dei prezzi di mercato. L'obiettivo di questa tesi è sviluppare un modello di reinforcement learning di tipo actor-critic con entropy regularization per impostare i prezzi in un gioco dinamico di competizione oligopolistica con prezzi continui. Il modello che propongo esibisce in modo coerente comportamenti cooperativi supportati da meccanismi di punizione che scoraggiano la deviazione dall'equilibrio raggiunto a convergenza. Il comportamento di questo modello durante l'apprendimento e a convergenza avvenuta aiuta inoltre a interpretare le azioni compiute da Q-learning tabellare e altri algoritmi di prezzo in condizioni simili. I risultati sono robusti alla variazione del numero di agenti in competizione e al tipo di deviazione dall'equilibrio ottenuto a convergenza, punendo anche deviazioni a prezzi più alti.

Intermediate linguistic task fine-tuning on Multi-lingual models

State-of-the-art NLP systems are generally based on the assumption that the underlying models are provided with vast datasets to train on. However, especially when working in multi-lingual contexts, datasets are often scarce, thus more research should be carried out in this field. This thesis investigates the benefits of introducing an additional training step when fine-tuning NLP models, named Intermediate Training, which could be exploited to augment the data used for the training phase. The Intermediate Training step is applied by training models on NLP tasks that are not strictly related to the target task, aiming to verify if the models are able to leverage the learned knowledge of such tasks. Furthermore, in order to better analyze the synergies between different categories of NLP tasks, experimentations have been extended also to Multi-Task Training, in which the model is trained on multiple tasks at the same time.

L’alè del clarinet : la respiració en la interpretació musical amb el clarinet

El projecte és una proposta d’exploració i reflexió personal sobre la respiració en la interpretació musical amb el clarinet i els aspectes corporals i pedagògics relacionats que fan possible l’estat psicofísic necessari per a una interpretació fluïda, creativa i amb veu pròpia. Per al seu desenvolupament, s’ha aprofundit en el coneixement de l’anatomia i fisiologia de la respiració, així com en la consciència corporal per aplicar-ho a la interpretació amb el clarinet i desenvolupar unes eines per a la seva didàctica. El mètode de treball consta de tres fases: la documentació, les entrevistes i les sessions didàctiques. Tot plegat ha significat un profund aprenentatge a nivell personal.

Responsabilité sociale et gouvernance d’entreprise : études de cas de firmes en situation de crise

Les cas d’entreprises touchées par des scandales financiers, environnementaux ou concernant des conditions de travail abusives imposées à leur main-d’œuvre, n’ont cessé de jalonner l’actualité ces vingt dernières années. La multiplication des comportements à l’origine de ces scandales s’explique par l’environnement moins contraignant, que leur ont offert les politiques de privatisation, dérégulation et libéralisation, amorcées à partir des années 1980. Le développement de la notion de responsabilité sociale des entreprises à partir des années 1980, en réaction à ces excès, incarne l'idée que si une entreprise doit certes faire des profits et les pérenniser elle se doit de les réaliser en favorisant les comportements responsables, éthiques et transparents avec toutes ses parties prenantes. Nous analysons dans cette thèse le processus par lequel, face à des dysfonctionnements et abus, touchant les conditions de travail de leur main d’œuvre ou leur gouvernance, des entreprises peuvent être amenées, ou non, à questionner et modifier leurs pratiques. Nous avons axé notre étude de cas sur deux entreprises aux trajectoires diamétralement opposées. La première entreprise, issue du secteur de la fabrication de vêtements et dont la crise concernait des atteintes aux droits des travailleurs, a surmonté la crise en réformant son modèle de production. La seconde entreprise, située dans le secteur des technologies de l'information et de la communication, a fait face à une crise liée à sa gouvernance d’entreprise, multiplié les dysfonctionnements pendant dix années de crises et finalement déclaré faillite en janvier 2009. Les évolutions théoriques du courant néo-institutionnel ces dernières années, permettent d’éclairer le processus par lequel de nouvelles normes émergent et se diffusent, en soulignant le rôle de différents acteurs, qui pour les uns, définissent de nouvelles normes et pour d’autres se mobilisent en vue de les diffuser. Afin d’augmenter leur efficacité à l’échelle mondiale, il apparaît que ces acteurs agissent le plus souvent en réseaux, parfois concurrents. L’étude du cas de cette compagnie du secteur de la confection de vêtement nous a permis d’aborder le domaine lié aux conditions de travail de travailleurs œuvrant au sein de chaînes de production délocalisées dans des pays aux lois sociales absentes ou inefficaces. Nous avons analysé le cheminement par lequel cette entreprise fut amenée à considérer, avec plus de rigueur, la dimension éthique dans sa chaîne de production. L’entreprise, en passant par différentes étapes prenant la forme d’un processus d’apprentissage organisationnel, a réussi à surmonter la crise en réformant ses pratiques. Il est apparu que ce processus ne fut pas spontané et qu’il fut réalisé suite aux rôles joués par deux types d’acteurs. Premièrement, par la mobilisation incessante des mouvements de justice globale afin que l’entreprise réforme ses pratiques. Et deuxièmement, par le cadre normatif et le lieu de dialogue entre les différentes parties prenantes, fournis par un organisme privé source de normes. C’est fondamentalement le risque de perdre son accréditation à la cet organisme qui a poussé l’entreprise à engager des réformes. L’entreprise est parvenue à surmonter la crise, certes en adoptant et en respectant les normes définies par cette organisation mais fondamentalement en modifiant sa culture d'entreprise. Le leadership du CEO et du CFO a en effet permis la création d'une culture d'entreprise favorisant la remise en question, le dialogue et une plus grande prise en considération des parties prenantes, même si la gestion locale ne va pas sans poser parfois des difficultés de mise en œuvre. Concernant le domaine de la gouvernance d’entreprise, nous mettons en évidence, à travers l’étude des facteurs ayant mené au déclin et à la faillite d’une entreprise phare du secteur des technologies de l’information et de la communication, les limites des normes en la matière comme outil de bonne gouvernance. La légalité de la gestion comptable et la conformité de l’entreprise aux normes de gouvernance n'ont pas empêché l’apparition et la multiplication de dysfonctionnements et abus stratégiques et éthiques. Incapable de se servir des multiples crises auxquelles elle a fait face pour se remettre en question et engager un apprentissage organisationnel profond, l'entreprise s'est focalisée de manière obsessionnelle sur la rentabilité à court terme et la recherche d'un titre boursier élevé. La direction et le conseil d'administration ont manqué de leadership afin de créer une culture d'entreprise alliant innovation technologique et communication honnête et transparente avec les parties prenantes. Alors que l'étude consacrée à l’entreprise du secteur de la confection de vêtement illustre le cas d'une entreprise qui a su, par le biais d'un changement stratégique, relever les défis que lui imposait son environnement, l'étude des quinze dernières années de la compagnie issue du secteur des technologies de l’information et de la communication témoigne de la situation inverse. Il apparaît sur base de ces deux cas que si une gouvernance favorisant l'éthique et la transparence envers les parties prenantes nécessite la création d'une culture d'entreprise valorisant ces éléments, elle doit impérativement soutenir et être associée à une stratégie adéquate afin que l'entreprise puisse pérenniser ses activités.

Developing international business managers through international study visits to China

Globalization is a key factor in the success of business organizations today, impacting many aspects of management performance. Understanding the global business environment has therefore become a key objective in the teaching of international business on Executive MBA programs. Drawing on the theory of experiential learning, this study examines the relationship between program structure and learning activities of an international study visit (ISV) to China and the learning experience for Executive MBA students. The findings indicate that learning experience may be most effective where the structure of an ISV incorporates certain activities that promote experiential and deep-level learning. Educational implications are discussed.

Evaluating the robustness of EmotiBlog for sentiment analysis and opinion mining

Preliminary research demonstrated the EmotiBlog annotated corpus relevance as a Machine Learning resource to detect subjective data. In this paper we compare EmotiBlog with the JRC Quotes corpus in order to check the robustness of its annotation. We concentrate on its coarse-grained labels and carry out a deep Machine Learning experimentation also with the inclusion of lexical resources. The results obtained show a similarity with the ones obtained with the JRC Quotes corpus demonstrating the EmotiBlog validity as a resource for the SA task.

Ability assessment and job structure measures in occupational mobility

Recent and potential changes in technology have resulted in the anticipation of increases in the frequency of job changes. This has led manpower policy makers to investigate the feasibility of incorporating the employment skills of job groups in the general prediction of future job learning and performance with a view to the establishment of "job families" within which transfer might be considered reciprocally high. A structured job analysis instrument (the Position Analysis Questionnaire) is evaluated in terms of two distinct sets of scores; job dimensions and synthetically established attribute/trait profiles. Studies demonstrate that estimates of a job's structure/dimensions and requisite human attributes can be reliably established. Three alternative techniques of statistically assembling profiles of the requisite human attributes for jobs are found to have differential levels of reliability and differential degrees of validity in their estimation of the "actual" ability requirements of jobs. The utility of these two sets of job descriptors to serve as representations of the cognitive structure similarity of job groups is investigated in a study which simulates a job transfer situation. The central role of the index of similarity used to assess the relationship between "target" and "present" job is demonstrated. The relative extents to which job structure similarity and job attribute similariity are associated with positive transfer are investigated. The studies demonstrate that the dimensions of jobs, and more fruitfully their requisite human attributes can serve as bases to predict job transfer learning and performance. The nature of the index of similarity used to optimally formulate predictions of transfer is such that networks of jobs might be establishable to which current job incumbents could be expected to transfer positively. The derivation of "job families" with anticipated reciprocal transfer consequences is considered to be less appropriate.