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.

Desarrollo de técnicas de aprendizaje automático para la predicción de resultados de partidos en ligas futbolísiticas

En la actualidad, existen un gran número de investigaciones que usan técnicas de aprendizaje automático basadas en árboles de decisión. Como evolución de dichos trabajos, se han desarrollado métodos que usan Multiclasificadores (Random forest, Boosting, Bagging) que resuelven los mismos problemas abordados con árboles de decisión simples, aumentando el porcentaje de acierto. El ámbito de los problemas resueltos tradicionalmente por dichas técnicas es muy variado aunque destaca la bio-informática. En cualquier caso, la clasificación siempre puede ser consultada a un experto considerándose su respuesta como correcta. Existen problemas donde un experto en la materia no siempre acierta. Un ejemplo, pueden ser las quinielas (1X2). Donde podemos observar que un conocimiento del dominio del problema aumenta el porcentaje de aciertos, sin embargo, predecir un resultado erróneo es muy posible. El motivo es que el número de factores que influyen en un resultado es tan grande que, en muchas ocasiones, convierten la predicción en un acto de azar. En este trabajo pretendemos encontrar un multiclasificador basado en los clasificadores simples más estudiados como pueden ser el Perceptrón Multicapa o Árboles de Decisión con el porcentaje de aciertos más alto posible. Con tal fin, se van a estudiar e implementar una serie de configuraciones de clasificadores propios junto a multiclasificadores desarrollados por terceros. Otra línea de estudio son los propios datos, es decir, el conjunto de entrenamiento. Mediante un estudio del dominio del problema añadiremos nuevos atributos que enriquecen la información que disponemos de cada resultado intentando imitar el conocimiento en el que se basa un experto. Los desarrollos descritos se han realizado en R. Además, se ha realizado una aplicación que permite entrenar un multiclasificador (bien de los propios o bien de los desarrollados por terceros) y como resultado obtenemos la matriz de confusión junto al porcentaje de aciertos. En cuanto a resultados, obtenemos porcentajes de aciertos entre el 50% y el 55%. Por encima del azar y próximos a los resultados de los expertos.

VOX POPULI: THREE ESSAYS ON THE USE OF SOCIAL MEDIA FOR VALUE CREATION IN SERVICES

Prior research shows that electronic word of mouth (eWOM) wields considerable influence over consumer behavior. However, as the volume and variety of eWOM grows, firms are faced with challenges in analyzing and responding to this information. In this dissertation, I argue that to meet the new challenges and opportunities posed by the expansion of eWOM and to more accurately measure its impacts on firms and consumers, we need to revisit our methodologies for extracting insights from eWOM. This dissertation consists of three essays that further our understanding of the value of social media analytics, especially with respect to eWOM. In the first essay, I use machine learning techniques to extract semantic structure from online reviews. These semantic dimensions describe the experiences of consumers in the service industry more accurately than traditional numerical variables. To demonstrate the value of these dimensions, I show that they can be used to substantially improve the accuracy of econometric models of firm survival. In the second essay, I explore the effects on eWOM of online deals, such as those offered by Groupon, the value of which to both consumers and merchants is controversial. Through a combination of Bayesian econometric models and controlled lab experiments, I examine the conditions under which online deals affect online reviews and provide strategies to mitigate the potential negative eWOM effects resulting from online deals. In the third essay, I focus on how eWOM can be incorporated into efforts to reduce foodborne illness, a major public health concern. I demonstrate how machine learning techniques can be used to monitor hygiene in restaurants through crowd-sourced online reviews. I am able to identify instances of moral hazard within the hygiene inspection scheme used in New York City by leveraging a dictionary specifically crafted for this purpose. To the extent that online reviews provide some visibility into the hygiene practices of restaurants, I show how losses from information asymmetry may be partially mitigated in this context. Taken together, this dissertation contributes by revisiting and refining the use of eWOM in the service sector through a combination of machine learning and econometric methodologies.

Advanced Analytical Microscopy at the Nanoscale: Applications in Wide Bandgap and Solid Oxide Fuel Cell Materials

The atomic-level structure and chemistry of materials ultimately dictate their observed macroscopic properties and behavior. As such, an intimate understanding of these characteristics allows for better materials engineering and improvements in the resulting devices. In our work, two material systems were investigated using advanced electron and ion microscopy techniques, relating the measured nanoscale traits to overall device performance. First, transmission electron microscopy and electron energy loss spectroscopy (TEM-EELS) were used to analyze interfacial states at the semiconductor/oxide interface in wide bandgap SiC microelectronics. This interface contains defects that significantly diminish SiC device performance, and their fundamental nature remains generally unresolved. The impacts of various microfabrication techniques were explored, examining both current commercial and next-generation processing strategies. In further investigations, machine learning techniques were applied to the EELS data, revealing previously hidden Si, C, and O bonding states at the interface, which help explain the origins of mobility enhancement in SiC devices. Finally, the impacts of SiC bias temperature stressing on the interfacial region were explored. In the second system, focused ion beam/scanning electron microscopy (FIB/SEM) was used to reconstruct 3D models of solid oxide fuel cell (SOFC) cathodes. Since the specific degradation mechanisms of SOFC cathodes are poorly understood, FIB/SEM and TEM were used to analyze and quantify changes in the microstructure during performance degradation. Novel strategies for microstructure calculation from FIB-nanotomography data were developed and applied to LSM-YSZ and LSCF-GDC composite cathodes, aged with environmental contaminants to promote degradation. In LSM-YSZ, migration of both La and Mn cations to the grain boundaries of YSZ was observed using TEM-EELS. Few substantial changes however, were observed in the overall microstructure of the cells, correlating with a lack of performance degradation induced by the H2O. Using similar strategies, a series of LSCF-GDC cathodes were analyzed, aged in H2O, CO2, and Cr-vapor environments. FIB/SEM observation revealed considerable formation of secondary phases within these cathodes, and quantifiable modifications of the microstructure. In particular, Cr-poisoning was observed to cause substantial byproduct formation, which was correlated with drastic reductions in cell performance.

Search for Higgs Boson Production Beyond the Standard Model Using the Razor Kinematic Variables in pp Collisions at √s=8 TeV and Optimization of Higgs Boson Identification Using a Quantum Annealer

In the first part of this thesis we search for beyond the Standard Model physics through the search for anomalous production of the Higgs boson using the razor kinematic variables. We search for anomalous Higgs boson production using proton-proton collisions at center of mass energy √s=8 TeV collected by the Compact Muon Solenoid experiment at the Large Hadron Collider corresponding to an integrated luminosity of 19.8 fb^-1.

In the second part we present a novel method for using a quantum annealer to train a classifier to recognize events containing a Higgs boson decaying to two photons. We train that classifier using simulated proton-proton collisions at √s=8 TeV producing either a Standard Model Higgs boson decaying to two photons or a non-resonant Standard Model process that produces a two photon final state.

The production mechanisms of the Higgs boson are precisely predicted by the Standard Model based on its association with the mechanism of electroweak symmetry breaking. We measure the yield of Higgs bosons decaying to two photons in kinematic regions predicted to have very little contribution from a Standard Model Higgs boson and search for an excess of events, which would be evidence of either non-standard production or non-standard properties of the Higgs boson. We divide the events into disjoint categories based on kinematic properties and the presence of additional b-quarks produced in the collisions. In each of these disjoint categories, we use the razor kinematic variables to characterize events with topological configurations incompatible with typical configurations found from standard model production of the Higgs boson.

We observe an excess of events with di-photon invariant mass compatible with the Higgs boson mass and localized in a small region of the razor plane. We observe 5 events with a predicted background of 0.54 ± 0.28, which observation has a p-value of 10^-3 and a local significance of 3.35σ. This background prediction comes from 0.48 predicted non-resonant background events and 0.07 predicted SM higgs boson events. We proceed to investigate the properties of this excess, finding that it provides a very compelling peak in the di-photon invariant mass distribution and is physically separated in the razor plane from predicted background. Using another method of measuring the background and significance of the excess, we find a 2.5σ deviation from the Standard Model hypothesis over a broader range of the razor plane.

In the second part of the thesis we transform the problem of training a classifier to distinguish events with a Higgs boson decaying to two photons from events with other sources of photon pairs into the Hamiltonian of a spin system, the ground state of which is the best classifier. We then use a quantum annealer to find the ground state of this Hamiltonian and train the classifier. We find that we are able to do this successfully in less than 400 annealing runs for a problem of median difficulty at the largest problem size considered. The networks trained in this manner exhibit good classification performance, competitive with the more complicated machine learning techniques, and are highly resistant to overtraining. We also find that the nature of the training gives access to additional solutions that can be used to improve the classification performance by up to 1.2% in some regions.

STORM - a novel information fusion and cluster interpretation technique

Analysis of data without labels is commonly subject to scrutiny by unsupervised machine learning techniques. Such techniques provide more meaningful representations, useful for better understanding of a problem at hand, than by looking only at the data itself. Although abundant expert knowledge exists in many areas where unlabelled data is examined, such knowledge is rarely incorporated into automatic analysis. Incorporation of expert knowledge is frequently a matter of combining multiple data sources from disparate hypothetical spaces. In cases where such spaces belong to different data types, this task becomes even more challenging. In this paper we present a novel immune-inspired method that enables the fusion of such disparate types of data for a specific set of problems. We show that our method provides a better visual understanding of one hypothetical space with the help of data from another hypothetical space. We believe that our model has implications for the field of exploratory data analysis and knowledge discovery.

Functional, clustered, feedforward, and mesoscale brain networks

The brain is a network spanning multiple scales from subcellular to macroscopic. In this thesis I present four projects studying brain networks at different levels of abstraction. The first involves determining a functional connectivity network based on neural spike trains and using a graph theoretical method to cluster groups of neurons into putative cell assemblies. In the second project I model neural networks at a microscopic level. Using diferent clustered wiring schemes, I show that almost identical spatiotemporal activity patterns can be observed, demonstrating that there is a broad neuro-architectural basis to attain structured spatiotemporal dynamics. Remarkably, irrespective of the precise topological mechanism, this behavior can be predicted by examining the spectral properties of the synaptic weight matrix. The third project introduces, via two circuit architectures, a new paradigm for feedforward processing in which inhibitory neurons have the complex and pivotal role in governing information flow in cortical network models. Finally, I analyze axonal projections in sleep deprived mice using data collected as part of the Allen Institute's Mesoscopic Connectivity Atlas. After normalizing for experimental variability, the results indicate there is no single explanatory difference in the mesoscale network between control and sleep deprived mice. Using machine learning techniques, however, animal classification could be done at levels significantly above chance. This reveals that intricate changes in connectivity do occur due to chronic sleep deprivation.

Improving Resource Management in Virtualized Data Centers using Application Performance Models

The rapid growth of virtualized data centers and cloud hosting services is making the management of physical resources such as CPU, memory, and I/O bandwidth in data center servers increasingly important. Server management now involves dealing with multiple dissimilar applications with varying Service-Level-Agreements (SLAs) and multiple resource dimensions. The multiplicity and diversity of resources and applications are rendering administrative tasks more complex and challenging. This thesis aimed to develop a framework and techniques that would help substantially reduce data center management complexity. We specifically addressed two crucial data center operations. First, we precisely estimated capacity requirements of client virtual machines (VMs) while renting server space in cloud environment. Second, we proposed a systematic process to efficiently allocate physical resources to hosted VMs in a data center. To realize these dual objectives, accurately capturing the effects of resource allocations on application performance is vital. The benefits of accurate application performance modeling are multifold. Cloud users can size their VMs appropriately and pay only for the resources that they need; service providers can also offer a new charging model based on the VMs performance instead of their configured sizes. As a result, clients will pay exactly for the performance they are actually experiencing; on the other hand, administrators will be able to maximize their total revenue by utilizing application performance models and SLAs. This thesis made the following contributions. First, we identified resource control parameters crucial for distributing physical resources and characterizing contention for virtualized applications in a shared hosting environment. Second, we explored several modeling techniques and confirmed the suitability of two machine learning tools, Artificial Neural Network and Support Vector Machine, to accurately model the performance of virtualized applications. Moreover, we suggested and evaluated modeling optimizations necessary to improve prediction accuracy when using these modeling tools. Third, we presented an approach to optimal VM sizing by employing the performance models we created. Finally, we proposed a revenue-driven resource allocation algorithm which maximizes the SLA-generated revenue for a data center.

Classifying And Predicting Software Security Vulnerabilities based on Reproducibility

Security defects are common in large software systems because of their size and complexity. Although efficient development processes, testing, and maintenance policies are applied to software systems, there are still a large number of vulnerabilities that can remain, despite these measures. Some vulnerabilities stay in a system from one release to the next one because they cannot be easily reproduced through testing. These vulnerabilities endanger the security of the systems. We propose vulnerability classification and prediction frameworks based on vulnerability reproducibility. The frameworks are effective to identify the types and locations of vulnerabilities in the earlier stage, and improve the security of software in the next versions (referred to as releases). We expand an existing concept of software bug classification to vulnerability classification (easily reproducible and hard to reproduce) to develop a classification framework for differentiating between these vulnerabilities based on code fixes and textual reports. We then investigate the potential correlations between the vulnerability categories and the classical software metrics and some other runtime environmental factors of reproducibility to develop a vulnerability prediction framework. The classification and prediction frameworks help developers adopt corresponding mitigation or elimination actions and develop appropriate test cases. Also, the vulnerability prediction framework is of great help for security experts focus their effort on the top-ranked vulnerability-prone files. As a result, the frameworks decrease the number of attacks that exploit security vulnerabilities in the next versions of the software. To build the classification and prediction frameworks, different machine learning techniques (C4.5 Decision Tree, Random Forest, Logistic Regression, and Naive Bayes) are employed. The effectiveness of the proposed frameworks is assessed based on collected software security defects of Mozilla Firefox.

Criação de uma ontologia e respectiva povoação a partir do processamento de relatórios médicos

A evolução tecnológica tem provocado uma evolução na medicina, através de sistemas computacionais voltados para o armazenamento, captura e disponibilização de informações médicas. Os relatórios médicos são, na maior parte das vezes, guardados num texto livre não estruturado e escritos com vocabulário proprietário, podendo ocasionar falhas de interpretação. Através das linguagens da Web Semântica, é possível utilizar antologias como modo de estruturar e padronizar a informação dos relatórios médicos, adicionando¬ lhe anotações semânticas. A informação contida nos relatórios pode desta forma ser publicada na Web, permitindo às máquinas o processamento automático da informação. No entanto, o processo de criação de antologias é bastante complexo, pois existe o problema de criar uma ontologia que não cubra todo o domínio pretendido. Este trabalho incide na criação de uma ontologia e respectiva povoação, através de técnicas de PLN e Aprendizagem Automática que permitem extrair a informação dos relatórios médicos. Foi desenvolvida uma aplicação, que permite ao utilizador converter relatórios do formato digital para o formato OWL. ABSTRACT: Technological evolution has caused a medicine evolution through computer systems which allow storage, gathering and availability of medical information. Medical reports are, most of the times, stored in a non-structured free text and written in a personal way so that misunderstandings may occur. Through Semantic Web languages, it’s possible to use ontology as a way to structure and standardize medical reports information by adding semantic notes. The information in those reports can, by these means, be displayed on the web, allowing machines automatic information processing. However, the process of creating ontology is very complex, as there is a risk creating of an ontology that not covering the whole desired domain. This work is about creation of an ontology and its population through NLP and Machine Learning techniques to extract information from medical reports. An application was developed which allows the user to convert reports from digital for¬ mat to OWL format.

Improving understandability in consumer health information search: Uevora @ 2016 fire chis

This paper presents our work at 2016 FIRE CHIS. Given a CHIS query and a document associated with that query, the task is to classify the sentences in the document as relevant to the query or not; and further classify the relevant sentences to be supporting, neutral or opposing to the claim made in the query. In this paper, we present two different approaches to do the classification. With the first approach, we implement two models to satisfy the task. We first implement an information retrieval model to retrieve the sentences that are relevant to the query; and then we use supervised learning method to train a classification model to classify the relevant sentences into support, oppose or neutral. With the second approach, we only use machine learning techniques to learn a model and classify the sentences into four classes (relevant & support, relevant & neutral, relevant & oppose, irrelevant & neutral). Our submission for CHIS uses the first approach.

Wireless solutions for the industrial internet of things

The fourth industrial revolution is paving the way for Industrial Internet of Things applications where industrial assets (e.g., robotic arms, valves, pistons) are equipped with a large number of wireless devices (i.e., microcontroller boards that embed sensors and actuators) to enable a plethora of new applications, such as analytics, diagnostics, monitoring, as well as supervisory, and safety control use-cases. Nevertheless, current wireless technologies, such as Wi-Fi, Bluetooth, and even private 5G networks, cannot fulfill all the requirements set up by the Industry 4.0 paradigm, thus opening up new 6G-oriented research trends, such as the use of THz frequencies. In light of the above, this thesis provides (i) a broad overview of the main use-cases, requirements, and key enabling wireless technologies foreseen by the fourth industrial revolution, and (ii) proposes innovative contributions, both theoretical and empirical, to enhance the performance of current and future wireless technologies at different levels of the protocol stack. In particular, at the physical layer, signal processing techniques are being exploited to analyze two multiplexing schemes, namely Affine Frequency Division Multiplexing and Orthogonal Chirp Division Multiplexing, which seem promising for high-frequency wireless communications. At the medium access layer, three protocols for intra-machine communications are proposed, where one is based on LoRa at 2.4 GHz and the others work in the THz band. Different scheduling algorithms for private industrial 5G networks are compared, and two main proposals are described, i.e., a decentralized scheme that leverages machine learning techniques to better address aperiodic traffic patterns, and a centralized contention-based design that serves a federated learning industrial application. Results are provided in terms of numerical evaluations, simulation results, and real-world experiments. Several improvements over the state-of-the-art were obtained, and the description of up-and-running testbeds demonstrates the feasibility of some of the theoretical concepts when considering a real industry plant.

Data-driven and data-oriented methods for materials science and technologies

The discovery of new materials and their functions has always been a fundamental component of technological progress. Nowadays, the quest for new materials is stronger than ever: sustainability, medicine, robotics and electronics are all key assets which depend on the ability to create specifically tailored materials. However, designing materials with desired properties is a difficult task, and the complexity of the discipline makes it difficult to identify general criteria. While scientists developed a set of best practices (often based on experience and expertise), this is still a trial-and-error process. This becomes even more complex when dealing with advanced functional materials. Their properties depend on structural and morphological features, which in turn depend on fabrication procedures and environment, and subtle alterations leads to dramatically different results. Because of this, materials modeling and design is one of the most prolific research fields. Many techniques and instruments are continuously developed to enable new possibilities, both in the experimental and computational realms. Scientists strive to enforce cutting-edge technologies in order to make progress. However, the field is strongly affected by unorganized file management, proliferation of custom data formats and storage procedures, both in experimental and computational research. Results are difficult to find, interpret and re-use, and a huge amount of time is spent interpreting and re-organizing data. This also strongly limit the application of data-driven and machine learning techniques. This work introduces possible solutions to the problems described above. Specifically, it talks about developing features for specific classes of advanced materials and use them to train machine learning models and accelerate computational predictions for molecular compounds; developing method for organizing non homogeneous materials data; automate the process of using devices simulations to train machine learning models; dealing with scattered experimental data and use them to discover new patterns.

Anticancer drug discovery using artificial intelligence: an application in pharmacological activity prediction

Hematological cancers are a heterogeneous family of diseases that can be divided into leukemias, lymphomas, and myelomas, often called “liquid tumors”. Since they cannot be surgically removable, chemotherapy represents the mainstay of their treatment. However, it still faces several challenges like drug resistance and low response rate, and the need for new anticancer agents is compelling. The drug discovery process is long-term, costly, and prone to high failure rates. With the rapid expansion of biological and chemical "big data", some computational techniques such as machine learning tools have been increasingly employed to speed up and economize the whole process. Machine learning algorithms can create complex models with the aim to determine the biological activity of compounds against several targets, based on their chemical properties. These models are defined as multi-target Quantitative Structure-Activity Relationship (mt-QSAR) and can be used to virtually screen small and large chemical libraries for the identification of new molecules with anticancer activity. The aim of my Ph.D. project was to employ machine learning techniques to build an mt-QSAR classification model for the prediction of cytotoxic drugs simultaneously active against 43 hematological cancer cell lines. For this purpose, first, I constructed a large and diversified dataset of molecules extracted from the ChEMBL database. Then, I compared the performance of different ML classification algorithms, until Random Forest was identified as the one returning the best predictions. Finally, I used different approaches to maximize the performance of the model, which achieved an accuracy of 88% by correctly classifying 93% of inactive molecules and 72% of active molecules in a validation set. This model was further applied to the virtual screening of a small dataset of molecules tested in our laboratory, where it showed 100% accuracy in correctly classifying all molecules. This result is confirmed by our previous in vitro experiments.

Introducing CHIMeRA: from the development of a comprehensive biomedical database to the analysis of its sub-components.

In this thesis we discuss the expansion of an existing project, called CHIMeRA, which is a comprehensive biomedical network, and the analysis of its sub-components by using graph theory. We describe how it is structured internally, what are the existing databases from which it retrieves information and what machine learning techniques are used in order to produce new knowledge. We also introduce a new technique for graph exploration that is aimed to speed-up the network cover time under the condition that the analyzed graph is stellar; if this condition is satisfied, the improvement in the performance compared to the conventional exploration technique is extremely appealing. We show that the stellar structure is highly recurrent for sub-networks in CHIMeRA generated by queries, which made this technique even more interesting. Finally, we describe the convenience in using the CHIMeRA network for research purposes and what it could become in a very near future.