Analysis of COVID-related symptoms from a database of Italian tweets

Twitter is a highly popular social media which on one hand allows information transmission in real time and on the other hand represents a source of open access homogeneous text data. We propose an analysis of the most common self-reported COVID symptoms from a dataset of Italian tweets to investigate the evolution of the pandemic in Italy from the end of September 2020 to the end of January 2021. After manually filtering tweets actually describing COVID symptoms from the database - which contains words related to fever, cough and sore throat - we discuss usefulness of such filtering. We then compare our time series with the daily data of new hospitalisations in Italy, with the aim of building a simple linear regression model that accounts for the delay which is observed from the tweets mentioning individual symptoms to new hospitalisations. We discuss both the results and limitations of linear regression given that our data suggests that the relationship between time series of symptoms tweets and of new hospitalisations changes towards the end of the acquisition.

Descriptive Text Mining in ambito medico: applicazione della metodologia POIROT

Poiché la nostra conoscenza collettiva continua ad essere digitalizzata e memorizzata, diventa più difficile trovare e scoprire ciò che stiamo cercando. Abbiamo bisogno di nuovi strumenti computazionali per aiutare a organizzare, rintracciare e comprendere queste vaste quantità di informazioni. I modelli di linguaggio sono potenti strumenti che possono essere impiegati per estrarre conoscenza statisticamente significativa ed interpretabile tramite apprendimento non supervisionato, testuali o nel codice sorgente. L’obiettivo di questa tesi è impiegare una metodologia di descriptive text mining, denominata POIROT, per analizzare i rapporti medici del dataset Adverse Drug Reaction (ADE). Si vogliono stabilire delle correlazioni significative che permettano di comprendere le ragioni per cui un determinato rapporto medico fornisca o meno informazioni relative a effetti collaterali dovuti all’assunzione di determinati farmaci.

Big Code Applications and Approaches

The availability of a huge amount of source code from code archives and open-source projects opens up the possibility to merge machine learning, programming languages, and software engineering research fields. This area is often referred to as Big Code where programming languages are treated instead of natural languages while different features and patterns of code can be exploited to perform many useful tasks and build supportive tools. Among all the possible applications which can be developed within the area of Big Code, the work presented in this research thesis mainly focuses on two particular tasks: the Programming Language Identification (PLI) and the Software Defect Prediction (SDP) for source codes. Programming language identification is commonly needed in program comprehension and it is usually performed directly by developers. However, when it comes at big scales, such as in widely used archives (GitHub, Software Heritage), automation of this task is desirable. To accomplish this aim, the problem is analyzed from different points of view (text and image-based learning approaches) and different models are created paying particular attention to their scalability. Software defect prediction is a fundamental step in software development for improving quality and assuring the reliability of software products. In the past, defects were searched by manual inspection or using automatic static and dynamic analyzers. Now, the automation of this task can be tackled using learning approaches that can speed up and improve related procedures. Here, two models have been built and analyzed to detect some of the commonest bugs and errors at different code granularity levels (file and method levels). Exploited data and models’ architectures are analyzed and described in detail. Quantitative and qualitative results are reported for both PLI and SDP tasks while differences and similarities concerning other related works are discussed.

Continual learning for computer vision applications

One of the most visionary goals of Artificial Intelligence is to create a system able to mimic and eventually surpass the intelligence observed in biological systems including, ambitiously, the one observed in humans. The main distinctive strength of humans is their ability to build a deep understanding of the world by learning continuously and drawing from their experiences. This ability, which is found in various degrees in all intelligent biological beings, allows them to adapt and properly react to changes by incrementally expanding and refining their knowledge. Arguably, achieving this ability is one of the main goals of Artificial Intelligence and a cornerstone towards the creation of intelligent artificial agents. Modern Deep Learning approaches allowed researchers and industries to achieve great advancements towards the resolution of many long-standing problems in areas like Computer Vision and Natural Language Processing. However, while this current age of renewed interest in AI allowed for the creation of extremely useful applications, a concerningly limited effort is being directed towards the design of systems able to learn continuously. The biggest problem that hinders an AI system from learning incrementally is the catastrophic forgetting phenomenon. This phenomenon, which was discovered in the 90s, naturally occurs in Deep Learning architectures where classic learning paradigms are applied when learning incrementally from a stream of experiences. This dissertation revolves around the Continual Learning field, a sub-field of Machine Learning research that has recently made a comeback following the renewed interest in Deep Learning approaches. This work will focus on a comprehensive view of continual learning by considering algorithmic, benchmarking, and applicative aspects of this field. This dissertation will also touch on community aspects such as the design and creation of research tools aimed at supporting Continual Learning research, and the theoretical and practical aspects concerning public competitions in this field.

Typed behavioural equivalences in the Pi-Calculus

In this thesis, I study the notion of program equivalences, i.e. proving that two programs can be used interchangeably without altering the overall observable behaviour. This definition is highly dependent on the contexts in which these programs can be used; does the context have exceptions, parallelism, etc... So proofs also need to be adapted according to the expressiveness of those contexts. This thesis presents on the pi-calculus – a concurrent programming language – under various typing constraints. Types allows us to impose different disciplines like forcing a sequential execution, or ensuring linearity, meaning an object can be used once. In each case, the bisimulation, a standard proof technique for the pi-calculus, needs to be adapted accordingly to obtain a suitable equivalence. We then test how using the modified bisimulations can be used to reason about a language with higher-order functions and references, which once translated into the pi-calculus satisfies the typing constraints.

Concepts and methods for efficient decentralized learning in federated settings

Deep Neural Networks (DNNs) have revolutionized a wide range of applications beyond traditional machine learning and artificial intelligence fields, e.g., computer vision, healthcare, natural language processing and others. At the same time, edge devices have become central in our society, generating an unprecedented amount of data which could be used to train data-hungry models such as DNNs. However, the potentially sensitive or confidential nature of gathered data poses privacy concerns when storing and processing them in centralized locations. To this purpose, decentralized learning decouples model training from the need of directly accessing raw data, by alternating on-device training and periodic communications. The ability of distilling knowledge from decentralized data, however, comes at the cost of facing more challenging learning settings, such as coping with heterogeneous hardware and network connectivity, statistical diversity of data, and ensuring verifiable privacy guarantees. This Thesis proposes an extensive overview of decentralized learning literature, including a novel taxonomy and a detailed description of the most relevant system-level contributions in the related literature for privacy, communication efficiency, data and system heterogeneity, and poisoning defense. Next, this Thesis presents the design of an original solution to tackle communication efficiency and system heterogeneity, and empirically evaluates it on federated settings. For communication efficiency, an original method, specifically designed for Convolutional Neural Networks, is also described and evaluated against the state-of-the-art. Furthermore, this Thesis provides an in-depth review of recently proposed methods to tackle the performance degradation introduced by data heterogeneity, followed by empirical evaluations on challenging data distributions, highlighting strengths and possible weaknesses of the considered solutions. Finally, this Thesis presents a novel perspective on the usage of Knowledge Distillation as a mean for optimizing decentralized learning systems in settings characterized by data heterogeneity or system heterogeneity. Our vision on relevant future research directions close the manuscript.

Towards a logical foundation of randomized computation

This dissertation investigates the relations between logic and TCS in the probabilistic setting. It is motivated by two main considerations. On the one hand, since their appearance in the 1960s-1970s, probabilistic models have become increasingly pervasive in several fast-growing areas of CS. On the other, the study and development of (deterministic) computational models has considerably benefitted from the mutual interchanges between logic and CS. Nevertheless, probabilistic computation was only marginally touched by such fruitful interactions. The goal of this thesis is precisely to (start) bring(ing) this gap, by developing logical systems corresponding to specific aspects of randomized computation and, therefore, by generalizing standard achievements to the probabilistic realm. To do so, our key ingredient is the introduction of new, measure-sensitive quantifiers associated with quantitative interpretations. The dissertation is tripartite. In the first part, we focus on the relation between logic and counting complexity classes. We show that, due to our classical counting propositional logic, it is possible to generalize to counting classes, the standard results by Cook and Meyer and Stockmeyer linking propositional logic and the polynomial hierarchy. Indeed, we show that the validity problem for counting-quantified formulae captures the corresponding level in Wagner's hierarchy. In the second part, we consider programming language theory. Type systems for randomized \lambda-calculi, also guaranteeing various forms of termination properties, were introduced in the last decades, but these are not "logically oriented" and no Curry-Howard correspondence is known for them. Following intuitions coming from counting logics, we define the first probabilistic version of the correspondence. Finally, we consider the relationship between arithmetic and computation. We present a quantitative extension of the language of arithmetic able to formalize basic results from probability theory. This language is also our starting point to define randomized bounded theories and, so, to generalize canonical results by Buss.

Advancing digital healthcare research and machine learning-aided knowledge extraction amidst the COVID-19 pandemic

The rapid progression of biomedical research coupled with the explosion of scientific literature has generated an exigent need for efficient and reliable systems of knowledge extraction. This dissertation contends with this challenge through a concentrated investigation of digital health, Artificial Intelligence, and specifically Machine Learning and Natural Language Processing's (NLP) potential to expedite systematic literature reviews and refine the knowledge extraction process. The surge of COVID-19 complicated the efforts of scientists, policymakers, and medical professionals in identifying pertinent articles and assessing their scientific validity. This thesis presents a substantial solution in the form of the COKE Project, an initiative that interlaces machine reading with the rigorous protocols of Evidence-Based Medicine to streamline knowledge extraction. In the framework of the COKE (“COVID-19 Knowledge Extraction framework for next-generation discovery science”) Project, this thesis aims to underscore the capacity of machine reading to create knowledge graphs from scientific texts. The project is remarkable for its innovative use of NLP techniques such as a BERT + bi-LSTM language model. This combination is employed to detect and categorize elements within medical abstracts, thereby enhancing the systematic literature review process. The COKE project's outcomes show that NLP, when used in a judiciously structured manner, can significantly reduce the time and effort required to produce medical guidelines. These findings are particularly salient during times of medical emergency, like the COVID-19 pandemic, when quick and accurate research results are critical.

studio e sperimentazione di tecniche di valutazione della significatività statistica in modelli nlp

Negli ultimi anni, il natural language processing ha subito una forte evoluzione, principalmente dettata dai paralleli avanzamenti nell’area del deep-learning. Con dimensioni architetturali in crescita esponenziale e corpora di addestramento sempre più comprensivi, i modelli neurali sono attualmente in grado di generare testo in maniera indistinguibile da quello umano. Tuttavia, a predizioni accurate su task complessi, si contrappongono metriche frequentemente arretrate, non capaci di cogliere le sfumature semantiche o le dimensioni di valutazione richieste. Tale divario motiva ancora oggi l’adozione di una valutazione umana come metodologia standard, ma la natura pervasiva del testo sul Web rende evidente il bisogno di sistemi automatici, scalabili, ed efficienti sia sul piano dei tempi che dei costi. In questa tesi si propone un’analisi delle principali metriche allo stato dell’arte per la valutazione di modelli pre-addestrati, partendo da quelle più popolari come Rouge fino ad arrivare a quelle che a loro volta sfruttano modelli per valutare il testo. Inoltre, si introduce una nuova libreria – denominata Blanche– finalizzata a raccogliere in un unico ambiente le implementazioni dei principali contributi oggi disponibili, agevolando il loro utilizzo da parte di sviluppatori e ricercatori. Infine, si applica Blanche per una valutazione ad ampio spettro dei risultati generativi ottenuti all’interno di un reale caso di studio, incentrato sulla verbalizzazione di eventi biomedici espressi nella letteratura scientifica. Una particolare attenzione è rivolta alla gestione dell’astrattività, un aspetto sempre più cruciale e sfidante sul piano valutativo.

Reti Neurali Ricorrenti per la generazione di musica simbolica

Il Deep Learning ha radicalmente trasformato il mondo del Machine Learning migliorando lo stato dell'arte in diversi campi che spaziano dalla computer vision al natural language processing. Non fermandosi a problemi di classificazione, negli ultimi anni, applicazioni di tipo generativo hanno portato alla creazione di immagini realistiche e documenti letterali. Il mondo della musica non è esente da una moltitudine di esperimenti nello stesso campo, con risultati ancora acerbi ma comunque potenzialmente interessanti. In questa tesi verrà discussa l'applicazione di un di modello appartenente alla famiglia del Deep Learning per la generazione di musica simbolica.

Generazione automatica di una knowledge base con applicazione ai Sustainable Development Goals

Questa tesi di laurea compie uno studio sull’ utilizzo di tecniche di web crawling, web scraping e Natural Language Processing per costruire automaticamente un dataset di documenti e una knowledge base di coppie verbo-oggetto utilizzabile per la classificazione di testi. Dopo una breve introduzione sulle tecniche utilizzate verrà presentato il metodo di generazione, prima in forma teorica e generalizzabile a qualunque classificazione basata su un insieme di argomenti, e poi in modo specifico attraverso un caso di studio: il software SDG Detector. In particolare quest ultimo riguarda l’applicazione pratica del metodo esposto per costruire una raccolta di informazioni utili alla classificazione di documenti in base alla presenza di uno o più Sustainable Development Goals. La parte relativa alla classificazione è curata dal co-autore di questa applicazione, la presente invece si concentra su un’analisi di correttezza e performance basata sull’espansione del dataset e della derivante base di conoscenza.

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.

LAWSU-IT: Un Nuovo Dataset Giudiziario italiano per Long Document Summarization con Baseline Estrattive e Astrattive

L'avanzamento nel campo della long document summarization dipende interamente dalla disponibilità di dataset pubblici di alta qualità e con testi di lunghezza considerevole. Risulta pertanto problematico il fatto che tali dataset risultino spesso solo in lingua inglese, comportandone una limitazione notevole se ci si rivolge a linguaggi le cui risorse sono limitate. A tal scopo, si propone LAWSU-IT, un nuovo dataset giudiziario per long document summarization italiana. LAWSU-IT è il primo dataset italiano di summarization ad avere documenti di grandi dimensioni e a trattare il dominio giudiziario, ed è stato costruito attuando procedure di cleaning dei dati e selezione mirata delle istanze, con lo scopo di ottenere un dataset di long document summarization di alta qualità. Inoltre, sono proposte molteplici baseline sperimentali di natura estrattiva e astrattiva con modelli stato dell'arte e approcci di segmentazione del testo. Si spera che tale risultato possa portare a ulteriori ricerche e sviluppi nell'ambito della long document summarization italiana.

Integrazione di reti neurali con approcci simbolici per problemi di ragionamento deduttivo in linguaggio naturale

Uno degli obiettivi più ambizioni e interessanti dell'informatica, specialmente nel campo dell'intelligenza artificiale, consiste nel raggiungere la capacità di far ragionare un computer in modo simile a come farebbe un essere umano. I più recenti successi nell'ambito delle reti neurali profonde, specialmente nel campo dell'elaborazione del testo in linguaggio naturale, hanno incentivato lo studio di nuove tecniche per affrontare tale problema, a cominciare dal ragionamento deduttivo, la forma più semplice e lineare di ragionamento logico. La domanda fondamentale alla base di questa tesi è infatti la seguente: in che modo una rete neurale basata sull'architettura Transformer può essere impiegata per avanzare lo stato dell'arte nell'ambito del ragionamento deduttivo in linguaggio naturale? Nella prima parte di questo lavoro presento uno studio approfondito di alcune tecnologie recenti che hanno affrontato questo problema con intuizioni vincenti. Da questa analisi emerge come particolarmente efficace l'integrazione delle reti neurali con tecniche simboliche più tradizionali. Nella seconda parte propongo un focus sull'architettura ProofWriter, che ha il pregio di essere relativamente semplice e intuitiva pur presentando prestazioni in linea con quelle dei concorrenti. Questo approfondimento mette in luce la capacità dei modelli T5, con il supporto del framework HuggingFace, di produrre più risposte alternative, tra cui è poi possibile cercare esternamente quella corretta. Nella terza e ultima parte fornisco un prototipo che mostra come si può impiegare tale tecnica per arricchire i sistemi tipo ProofWriter con approcci simbolici basati su nozioni linguistiche, conoscenze specifiche sul dominio applicativo o semplice buonsenso. Ciò che ne risulta è un significativo miglioramento dell'accuratezza rispetto al ProofWriter originale, ma soprattutto la dimostrazione che è possibile sfruttare tale capacità dei modelli T5 per migliorarne le prestazioni.

Quality and Aspect based Argument Generation

Natural Language Processing has always been one of the most popular topics in Artificial Intelligence. Argument-related research in NLP, such as argument detection, argument mining and argument generation, has been popular, especially in recent years. In our daily lives, we use arguments to express ourselves. The quality of arguments heavily impacts the effectiveness of our communications with others. In professional fields, such as legislation and academic areas, arguments of good quality play an even more critical role. Therefore, argument generation with good quality is a challenging research task that is also of great importance in NLP. The aim of this work is to investigate the automatic generation of arguments with good quality, according to the given topic, stance and aspect (control codes). To achieve this goal, a module based on BERT [17] which could judge an argument's quality is constructed. This module is used to assess the quality of the generated arguments. Another module based on GPT-2 [19] is implemented to generate arguments. Stances and aspects are also used as guidance when generating arguments. After combining all these models and techniques, the ranks of the generated arguments could be acquired to evaluate the final performance. This dissertation describes the architecture and experimental setup, analyzes the results of our experimentation, and discusses future directions.