SEQUENTIAL DECISIONS AND PREDICTIONS IN NATURAL LANGUAGE PROCESSING

Natural language processing has achieved great success in a wide range of ap- plications, producing both commercial language services and open-source language tools. However, most methods take a static or batch approach, assuming that the model has all information it needs and makes a one-time prediction. In this disser- tation, we study dynamic problems where the input comes in a sequence instead of all at once, and the output must be produced while the input is arriving. In these problems, predictions are often made based only on partial information. We see this dynamic setting in many real-time, interactive applications. These problems usually involve a trade-off between the amount of input received (cost) and the quality of the output prediction (accuracy). Therefore, the evaluation considers both objectives (e.g., plotting a Pareto curve). Our goal is to develop a formal understanding of sequential prediction and decision-making problems in natural language processing and to propose efficient solutions. Toward this end, we present meta-algorithms that take an existent batch model and produce a dynamic model to handle sequential inputs and outputs. Webuild our framework upon theories of Markov Decision Process (MDP), which allows learning to trade off competing objectives in a principled way. The main machine learning techniques we use are from imitation learning and reinforcement learning, and we advance current techniques to tackle problems arising in our settings. We evaluate our algorithm on a variety of applications, including dependency parsing, machine translation, and question answering. We show that our approach achieves a better cost-accuracy trade-off than the batch approach and heuristic-based decision- making approaches. We first propose a general framework for cost-sensitive prediction, where dif- ferent parts of the input come at different costs. We formulate a decision-making process that selects pieces of the input sequentially, and the selection is adaptive to each instance. Our approach is evaluated on both standard classification tasks and a structured prediction task (dependency parsing). We show that it achieves similar prediction quality to methods that use all input, while inducing a much smaller cost. Next, we extend the framework to problems where the input is revealed incremen- tally in a fixed order. We study two applications: simultaneous machine translation and quiz bowl (incremental text classification). We discuss challenges in this set- ting and show that adding domain knowledge eases the decision-making problem. A central theme throughout the chapters is an MDP formulation of a challenging problem with sequential input/output and trade-off decisions, accompanied by a learning algorithm that solves the MDP.

A Visual Analytics Approach to Comparing Cohorts of Event Sequences

Sequences of timestamped events are currently being generated across nearly every domain of data analytics, from e-commerce web logging to electronic health records used by doctors and medical researchers. Every day, this data type is reviewed by humans who apply statistical tests, hoping to learn everything they can about how these processes work, why they break, and how they can be improved upon. To further uncover how these processes work the way they do, researchers often compare two groups, or cohorts, of event sequences to find the differences and similarities between outcomes and processes. With temporal event sequence data, this task is complex because of the variety of ways single events and sequences of events can differ between the two cohorts of records: the structure of the event sequences (e.g., event order, co-occurring events, or frequencies of events), the attributes about the events and records (e.g., gender of a patient), or metrics about the timestamps themselves (e.g., duration of an event). Running statistical tests to cover all these cases and determining which results are significant becomes cumbersome. Current visual analytics tools for comparing groups of event sequences emphasize a purely statistical or purely visual approach for comparison. Visual analytics tools leverage humans' ability to easily see patterns and anomalies that they were not expecting, but is limited by uncertainty in findings. Statistical tools emphasize finding significant differences in the data, but often requires researchers have a concrete question and doesn't facilitate more general exploration of the data. Combining visual analytics tools with statistical methods leverages the benefits of both approaches for quicker and easier insight discovery. Integrating statistics into a visualization tool presents many challenges on the frontend (e.g., displaying the results of many different metrics concisely) and in the backend (e.g., scalability challenges with running various metrics on multi-dimensional data at once). I begin by exploring the problem of comparing cohorts of event sequences and understanding the questions that analysts commonly ask in this task. From there, I demonstrate that combining automated statistics with an interactive user interface amplifies the benefits of both types of tools, thereby enabling analysts to conduct quicker and easier data exploration, hypothesis generation, and insight discovery. The direct contributions of this dissertation are: (1) a taxonomy of metrics for comparing cohorts of temporal event sequences, (2) a statistical framework for exploratory data analysis with a method I refer to as high-volume hypothesis testing (HVHT), (3) a family of visualizations and guidelines for interaction techniques that are useful for understanding and parsing the results, and (4) a user study, five long-term case studies, and five short-term case studies which demonstrate the utility and impact of these methods in various domains: four in the medical domain, one in web log analysis, two in education, and one each in social networks, sports analytics, and security. My dissertation contributes an understanding of how cohorts of temporal event sequences are commonly compared and the difficulties associated with applying and parsing the results of these metrics. It also contributes a set of visualizations, algorithms, and design guidelines for balancing automated statistics with user-driven analysis to guide users to significant, distinguishing features between cohorts. This work opens avenues for future research in comparing two or more groups of temporal event sequences, opening traditional machine learning and data mining techniques to user interaction, and extending the principles found in this dissertation to data types beyond temporal event sequences.

Caracterização dos fatores contribuintes em acidentes de pequenas aeronaves da aviação geral brasileira utilizando inteligência artificial

Dissertação (mestrado)—Universidade de Brasília, Faculdade de Tecnologia, Departamento de Engenharia Civil e Ambiental, 2016.

Ensemble Stream Model for Data-Cleaning in Sensor Networks

Ensemble Stream Modeling and Data-cleaning are sensor information processing systems have different training and testing methods by which their goals are cross-validated. This research examines a mechanism, which seeks to extract novel patterns by generating ensembles from data. The main goal of label-less stream processing is to process the sensed events to eliminate the noises that are uncorrelated, and choose the most likely model without over fitting thus obtaining higher model confidence. Higher quality streams can be realized by combining many short streams into an ensemble which has the desired quality. The framework for the investigation is an existing data mining tool. First, to accommodate feature extraction such as a bush or natural forest-fire event we make an assumption of the burnt area (BA*), sensed ground truth as our target variable obtained from logs. Even though this is an obvious model choice the results are disappointing. The reasons for this are two: One, the histogram of fire activity is highly skewed. Two, the measured sensor parameters are highly correlated. Since using non descriptive features does not yield good results, we resort to temporal features. By doing so we carefully eliminate the averaging effects; the resulting histogram is more satisfactory and conceptual knowledge is learned from sensor streams. Second is the process of feature induction by cross-validating attributes with single or multi-target variables to minimize training error. We use F-measure score, which combines precision and accuracy to determine the false alarm rate of fire events. The multi-target data-cleaning trees use information purity of the target leaf-nodes to learn higher order features. A sensitive variance measure such as f-test is performed during each node’s split to select the best attribute. Ensemble stream model approach proved to improve when using complicated features with a simpler tree classifier. The ensemble framework for data-cleaning and the enhancements to quantify quality of fitness (30% spatial, 10% temporal, and 90% mobility reduction) of sensor led to the formation of streams for sensor-enabled applications. Which further motivates the novelty of stream quality labeling and its importance in solving vast amounts of real-time mobile streams generated today.

Aplicaciones de los modelos de predicción bio-inspirados en la administración

Las organizaciones y sus entornos son sistemas complejos. Tales sistemas son difíciles de comprender y predecir. Pese a ello, la predicción es una tarea fundamental para la gestión empresarial y para la toma de decisiones que implica siempre un riesgo. Los métodos clásicos de predicción (entre los cuales están: la regresión lineal, la Autoregresive Moving Average y el exponential smoothing) establecen supuestos como la linealidad, la estabilidad para ser matemática y computacionalmente tratables. Por diferentes medios, sin embargo, se han demostrado las limitaciones de tales métodos. Pues bien, en las últimas décadas nuevos métodos de predicción han surgido con el fin de abarcar la complejidad de los sistemas organizacionales y sus entornos, antes que evitarla. Entre ellos, los más promisorios son los métodos de predicción bio-inspirados (ej. redes neuronales, algoritmos genéticos /evolutivos y sistemas inmunes artificiales). Este artículo pretende establecer un estado situacional de las aplicaciones actuales y potenciales de los métodos bio-inspirados de predicción en la administración.

Applications of metrological techniques for clinical implementation of dosimetry and radiobiology in molecular radiotherapy

Molecular radiotherapy (MRT) is a fast developing and promising treatment for metastasised neuroendocrine tumours. Efficacy of MRT is based on the capability to selectively "deliver" radiation to tumour cells, minimizing administered dose to normal tissues. Outcome of MRT depends on the individual patient characteristics. For that reason, personalized treatment planning is important to improve outcomes of therapy. Dosimetry plays a key role in this setting, as it is the main physical quantity related to radiation effects on cells. Dosimetry in MRT consists in a complex series of procedures ranging from imaging quantification to dose calculation. This doctoral thesis focused on several aspects concerning the clinical implementation of absorbed dose calculations in MRT. Accuracy of SPECT/CT quantification was assessed in order to determine the optimal reconstruction parameters. A model of PVE correction was developed in order to improve the activity quantification in small volume, such us lesions in clinical patterns. Advanced dosimetric methods were compared with the aim of defining the most accurate modality, applicable in clinical routine. Also, for the first time on a large number of clinical cases, the overall uncertainty of tumour dose calculation was assessed. As part of the MRTDosimetry project, protocols for calibration of SPECT/CT systems and implementation of dosimetry were drawn up in order to provide standard guidelines to the clinics offering MRT. To estimate the risk of experiencing radio-toxicity side effects and the chance of inducing damage on neoplastic cells is crucial for patient selection and treatment planning. In this thesis, the NTCP and TCP models were derived based on clinical data as help to clinicians to decide the pharmaceutical dosage in relation to the therapy control and the limitation of damage to healthy tissues. Moreover, a model for tumour response prediction based on Machine Learning analysis was developed.

Ethical Perspectives on Big Data in Agri-food: Ownership and Governance for Safety

Big data are reshaping the way we interact with technology, thus fostering new applications to increase the safety-assessment of foods. An extraordinary amount of information is analysed using machine learning approaches aimed at detecting the existence or predicting the likelihood of future risks. Food business operators have to share the results of these analyses when applying to place on the market regulated products, whereas agri-food safety agencies (including the European Food Safety Authority) are exploring new avenues to increase the accuracy of their evaluations by processing Big data. Such an informational endowment brings with it opportunities and risks correlated to the extraction of meaningful inferences from data. However, conflicting interests and tensions among the involved entities - the industry, food safety agencies, and consumers - hinder the finding of shared methods to steer the processing of Big data in a sound, transparent and trustworthy way. A recent reform in the EU sectoral legislation, the lack of trust and the presence of a considerable number of stakeholders highlight the need of ethical contributions aimed at steering the development and the deployment of Big data applications. Moreover, Artificial Intelligence guidelines and charters published by European Union institutions and Member States have to be discussed in light of applied contexts, including the one at stake. This thesis aims to contribute to these goals by discussing what principles should be put forward when processing Big data in the context of agri-food safety-risk assessment. The research focuses on two interviewed topics - data ownership and data governance - by evaluating how the regulatory framework addresses the challenges raised by Big data analysis in these domains. The outcome of the project is a tentative Roadmap aimed to identify the principles to be observed when processing Big data in this domain and their possible implementations.

New methodologies in CT perfusion and MRI analysis to develop cancer imaging biomarkers

Quantitative imaging in oncology aims at developing imaging biomarkers for diagnosis and prediction of cancer aggressiveness and therapy response before any morphological change become visible. This Thesis exploits Computed Tomography perfusion (CTp) and multiparametric Magnetic Resonance Imaging (mpMRI) for investigating diverse cancer features on different organs. I developed a voxel-based image analysis methodology in CTp and extended its use to mpMRI, for performing precise and accurate analyses at single-voxel level. This is expected to improve reproducibility of measurements and cancer mechanisms’ comprehension and clinical interpretability. CTp has not entered the clinical routine yet, although its usefulness in the monitoring of cancer angiogenesis, due to different perfusion computing methods yielding unreproducible results. Instead, machine learning applications in mpMRI, useful to detect imaging features representative of cancer heterogeneity, are mostly limited to clinical research, because of results’ variability and difficult interpretability, which make clinicians not confident in clinical applications. In hepatic CTp, I investigated whether, and under what conditions, two widely adopted perfusion methods, Maximum Slope (MS) and Deconvolution (DV), could yield reproducible parameters. To this end, I developed signal processing methods to model the first pass kinetics and remove any numerical cause hampering the reproducibility. In mpMRI, I proposed a new approach to extract local first-order features, aiming at preserving spatial reference and making their interpretation easier. In CTp, I found out the cause of MS and DV non-reproducibility: MS and DV represent two different states of the system. Transport delays invalidate MS assumptions and, by correcting MS formulation, I have obtained the voxel-based equivalence of the two methods. In mpMRI, the developed predictive models allowed (i) detecting rectal cancers responding to neoadjuvant chemoradiation showing, at pre-therapy, sparse coarse subregions with altered density, and (ii) predicting clinically significant prostate cancers stemming from the disproportion between high- and low- diffusivity gland components.

A Reinforcement Learning Agent for Distributed Task Allocation

Al giorno d'oggi il reinforcement learning ha dimostrato di essere davvero molto efficace nel machine learning in svariati campi, come ad esempio i giochi, il riconoscimento vocale e molti altri. Perciò, abbiamo deciso di applicare il reinforcement learning ai problemi di allocazione, in quanto sono un campo di ricerca non ancora studiato con questa tecnica e perchè questi problemi racchiudono nella loro formulazione un vasto insieme di sotto-problemi con simili caratteristiche, per cui una soluzione per uno di essi si estende ad ognuno di questi sotto-problemi. In questo progetto abbiamo realizzato un applicativo chiamato Service Broker, il quale, attraverso il reinforcement learning, apprende come distribuire l'esecuzione di tasks su dei lavoratori asincroni e distribuiti. L'analogia è quella di un cloud data center, il quale possiede delle risorse interne - possibilmente distribuite nella server farm -, riceve dei tasks dai suoi clienti e li esegue su queste risorse. L'obiettivo dell'applicativo, e quindi del data center, è quello di allocare questi tasks in maniera da minimizzare il costo di esecuzione. Inoltre, al fine di testare gli agenti del reinforcement learning sviluppati è stato creato un environment, un simulatore, che permettesse di concentrarsi nello sviluppo dei componenti necessari agli agenti, invece che doversi anche occupare di eventuali aspetti implementativi necessari in un vero data center, come ad esempio la comunicazione con i vari nodi e i tempi di latenza di quest'ultima. I risultati ottenuti hanno dunque confermato la teoria studiata, riuscendo a ottenere prestazioni migliori di alcuni dei metodi classici per il task allocation.

Detecting social patterns within 20th century documentary photos: a deep learning based approach

The job of a historian is to understand what happened in the past, resorting in many cases to written documents as a firsthand source of information. Text, however, does not amount to the only source of knowledge. Pictorial representations, in fact, have also accompanied the main events of the historical timeline. In particular, the opportunity of visually representing circumstances has bloomed since the invention of photography, with the possibility of capturing in real-time the occurrence of a specific events. Thanks to the widespread use of digital technologies (e.g. smartphones and digital cameras), networking capabilities and consequent availability of multimedia content, the academic and industrial research communities have developed artificial intelligence (AI) paradigms with the aim of inferring, transferring and creating new layers of information from images, videos, etc. Now, while AI communities are devoting much of their attention to analyze digital images, from an historical research standpoint more interesting results may be obtained analyzing analog images representing the pre-digital era. Within the aforementioned scenario, the aim of this work is to analyze a collection of analog documentary photographs, building upon state-of-the-art deep learning techniques. In particular, the analysis carried out in this thesis aims at producing two following results: (a) produce the date of an image, and, (b) recognizing its background socio-cultural context,as defined by a group of historical-sociological researchers. Given these premises, the contribution of this work amounts to: (i) the introduction of an historical dataset including images of “Family Album” among all the twentieth century, (ii) the introduction of a new classification task regarding the identification of the socio-cultural context of an image, (iii) the exploitation of different deep learning architectures to perform the image dating and the image socio-cultural context classification.

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.

Model-Based validation of Driver Drowsiness Detection System for ADAS

The work described in this Master’s Degree thesis was born after the collaboration with the company Maserati S.p.a, an Italian luxury car maker with its headquarters located in Modena, in the heart of the Italian Motor Valley, where I worked as a stagiaire in the Virtual Engineering team between September 2021 and February 2022. This work proposes the validation using real-world ECUs of a Driver Drowsiness Detection (DDD) system prototype based on different detection methods with the goal to overcome input signal losses and system failures. Detection methods of different categories have been chosen from literature and merged with the goal of utilizing the benefits of each of them, overcoming their limitations and limiting as much as possible their degree of intrusiveness to prevent any kind of driving distraction: an image processing-based technique for human physical signals detection as well as methods based on driver-vehicle interaction are used. A Driver-In-the-Loop simulator is used to gather real data on which a Machine Learning-based algorithm will be trained and validated. These data come from the tests that the company conducts in its daily activities so confidential information about the simulator and the drivers will be omitted. Although the impact of the proposed system is not remarkable and there is still work to do in all its elements, the results indicate the main advantages of the system in terms of robustness against subsystem failures and signal losses.

End-to-end Deep Metric Learning con Vision-Language Model per il Fashion Image Captioning

L'image captioning è un task di machine learning che consiste nella generazione di una didascalia, o caption, che descriva le caratteristiche di un'immagine data in input. Questo può essere applicato, ad esempio, per descrivere in dettaglio i prodotti in vendita su un sito di e-commerce, migliorando l'accessibilità del sito web e permettendo un acquisto più consapevole ai clienti con difficoltà visive. La generazione di descrizioni accurate per gli articoli di moda online è importante non solo per migliorare le esperienze di acquisto dei clienti, ma anche per aumentare le vendite online. Oltre alla necessità di presentare correttamente gli attributi degli articoli, infatti, descrivere i propri prodotti con il giusto linguaggio può contribuire a catturare l'attenzione dei clienti. In questa tesi, ci poniamo l'obiettivo di sviluppare un sistema in grado di generare una caption che descriva in modo dettagliato l'immagine di un prodotto dell'industria della moda dato in input, sia esso un capo di vestiario o un qualche tipo di accessorio. A questo proposito, negli ultimi anni molti studi hanno proposto soluzioni basate su reti convoluzionali e LSTM. In questo progetto proponiamo invece un'architettura encoder-decoder, che utilizza il modello Vision Transformer per la codifica delle immagini e GPT-2 per la generazione dei testi. Studiamo inoltre come tecniche di deep metric learning applicate in end-to-end durante l'addestramento influenzino le metriche e la qualità delle caption generate dal nostro modello.

Design and implementation of a Reinforcement Learning framework for iOS devices

Reinforcement Learning is an increasingly popular area of Artificial Intelligence. The applications of this learning paradigm are many, but its application in mobile computing is in its infancy. This study aims to provide an overview of current Reinforcement Learning applications on mobile devices, as well as to introduce a new framework for iOS devices: Swift-RL Lib. This new Swift package allows developers to easily support and integrate two of the most common RL algorithms, Q-Learning and Deep Q-Network, in a fully customizable environment. All processes are performed on the device, without any need for remote computation. The framework was tested in different settings and evaluated through several use cases. Through an in-depth performance analysis, we show that the platform provides effective and efficient support for Reinforcement Learning for mobile applications.

Hybrid Artificial Intelligence to Extract Patterns and Rules from Argumentative and Legal Texts

This Thesis is composed of a collection of works written in the period 2019-2022, whose aim is to find methodologies of Artificial Intelligence (AI) and Machine Learning to detect and classify patterns and rules in argumentative and legal texts. We define our approach “hybrid”, since we aimed at designing hybrid combinations of symbolic and sub-symbolic AI, involving both “top-down” structured knowledge and “bottom-up” data-driven knowledge. A first group of works is dedicated to the classification of argumentative patterns. Following the Waltonian model of argument and the related theory of Argumentation Schemes, these works focused on the detection of argumentative support and opposition, showing that argumentative evidences can be classified at fine-grained levels without resorting to highly engineered features. To show this, our methods involved not only traditional approaches such as TFIDF, but also some novel methods based on Tree Kernel algorithms. After the encouraging results of this first phase, we explored the use of a some emerging methodologies promoted by actors like Google, which have deeply changed NLP since 2018-19 — i.e., Transfer Learning and language models. These new methodologies markedly improved our previous results, providing us with best-performing NLP tools. Using Transfer Learning, we also performed a Sequence Labelling task to recognize the exact span of argumentative components (i.e., claims and premises), thus connecting portions of natural language to portions of arguments (i.e., to the logical-inferential dimension). The last part of our work was finally dedicated to the employment of Transfer Learning methods for the detection of rules and deontic modalities. In this case, we explored a hybrid approach which combines structured knowledge coming from two LegalXML formats (i.e., Akoma Ntoso and LegalRuleML) with sub-symbolic knowledge coming from pre-trained (and then fine-tuned) neural architectures.