Cyber systems, including IoT (Internet of Things), are increasingly being used ubiquitously to vastly improve the efficiency and reduce the cost of critical application areas, such as finance, transportation, defense, and healthcare. Over the past two decades, computing efficiency and hardware cost have dramatically been improved. These improvements have made cyber systems omnipotent, and control many aspects of human lives. Emerging trends in successful cyber system breaches have shown increasing sophistication in attacks and that attackers are no longer limited by resources, including human and computing power. Most existing cyber defense systems for IoT systems have two major issues: (1) they do not incorporate human user behavior(s) and preferences in their approaches, and (2) they do not continuously learn from dynamic environment and effectively adapt to thwart sophisticated cyber-attacks. Consequently, the security solutions generated may not be usable or implementable by the user(s) thereby drastically reducing the effectiveness of these security solutions.

In order to address these major issues, a comprehensive approach to securing ubiquitous smart devices in IoT environment by incorporating probabilistic human user behavioral inputs is presented. The approach will include techniques to (1) protect the controller device(s) [smart phone or tablet] by continuously learning and authenticating the legitimate user based on the touch screen finger gestures in the background, without requiring users’ to provide their finger gesture inputs intentionally for training purposes, and (2) efficiently configure IoT devices through controller device(s), in conformance with the probabilistic human user behavior(s) and preferences, to effectively adapt IoT devices to the changing environment. The effectiveness of the approach will be demonstrated with experiments that are based on collected user behavioral data and simulations.

Keyword search provides a simple and user-friendly mechanism for information search, and has become increasingly popular for accessing structured or semi-structured data. However, there are two open issues of keyword search on semi/structured data which are not well addressed by existing work yet.

First, while an increasing amount of investigation has been done in this important area, most existing work concentrates on efficiency instead of search quality and may fail to deliver high quality results from semantic perspectives. Majority of the existing work generates minimal sub-graph results that are oblivious to the entity and relationship semantics embedded in the data and in the user query. There are also studies that define results to be subtrees or subgraphs that contain all query keywords but are not necessarily ``minimal''. However, such result construction method suffers from the same problem of semantic mis-alignment between data and user query. In this work the semantics of how to {\em define} results that can capture users' search intention and then the generation of search intention aware results is studied.

Second, most existing research is incapable of handling large-scale structured data. However, as data volume has seen rapid growth in recent years, the problem of how to efficiently process keyword queries on large-scale structured data becomes important. MapReduce is widely acknowledged as an effective programming model to process big data. For keyword query processing on data graph, first graph algorithms which can efficiently return query results that are consistent with users' search intention are proposed. Then these algorithms are migrated to MapReduce to support big data. For keyword query processing on schema graph, it first transforms a keyword query into multiple SQL queries, then all generated SQL queries are run on the structured data. Therefore it is crucial to find the optimal way to execute a SQL query using MapReduce, which can minimize the processing time. In this work, a system called SOSQL is developed which generates the optimal query execution plan using MapReduce for a SQL query $Q$ with time complexity $O(n^2)$, where $n$ is the number of input tables of $Q$.

Measuring node centrality is a critical common denominator behind many important graph mining tasks. While the existing literature offers a wealth of different node centrality measures, it remains a daunting task on how to intervene the node centrality in a desired way. In this thesis, we study the problem of minimizing the centrality of one or more target nodes by edge operation. The heart of the proposed method is an accurate and efficient algorithm to estimate the impact of edge deletion on the spectrum of the underlying network, based on the observation that the edge deletion is essentially a local, sparse perturbation to the original network. Extensive experiments are conducted on a diverse set of real networks to demonstrate the effectiveness, efficiency and scalability of our approach. In particular, it is average of 260.95%, in terms of minimizing eigen-centrality, better than the standard matrix-perturbation based algorithm, with lower time complexity.

The dawn of Internet of Things (IoT) has opened the opportunity for mainstream adoption of machine learning analytics. However, most research in machine learning has focused on discovery of new algorithms or fine-tuning the performance of existing algorithms. Little exists on the process of taking an algorithm from the lab-environment into the real-world, culminating in sustained value. Real-world applications are typically characterized by dynamic non-stationary systems with requirements around feasibility, stability and maintainability. Not much has been done to establish standards around the unique analytics demands of real-world scenarios.

This research explores the problem of the why so few of the published algorithms enter production and furthermore, fewer end up generating sustained value. The dissertation proposes a ‘Design for Deployment’ (DFD) framework to successfully build machine learning analytics so they can be deployed to generate sustained value. The framework emphasizes and elaborates the often neglected but immensely important latter steps of an analytics process: ‘Evaluation’ and ‘Deployment’. A representative evaluation framework is proposed that incorporates the temporal-shifts and dynamism of real-world scenarios. Additionally, the recommended infrastructure allows analytics projects to pivot rapidly when a particular venture does not materialize. Deployment needs and apprehensions of the industry are identified and gaps addressed through a 4-step process for sustainable deployment. Lastly, the need for analytics as a functional area (like finance and IT) is identified to maximize the return on machine-learning deployment.

The framework and process is demonstrated in semiconductor manufacturing – it is highly complex process involving hundreds of optical, electrical, chemical, mechanical, thermal, electrochemical and software processes which makes it a highly dynamic non-stationary system. Due to the 24/7 uptime requirements in manufacturing, high-reliability and fail-safe are a must. Moreover, the ever growing volumes mean that the system must be highly scalable. Lastly, due to the high cost of change, sustained value proposition is a must for any proposed changes. Hence the context is ideal to explore the issues involved. The enterprise use-cases are used to demonstrate the robustness of the framework in addressing challenges encountered in the end-to-end process of productizing machine learning analytics in dynamic read-world scenarios.

Nowadays, Computing is so pervasive that it has become indeed the 5th utility (after water, electricity, gas, telephony) as Leonard Kleinrock once envisioned. Evolved from utility computing, cloud computing has emerged as a computing infrastructure that enables rapid delivery of computing resources as a utility in a dynamically scalable, virtualized manner. However, the current industrial cloud computing implementations promote segregation among different cloud providers, which leads to user lockdown because of prohibitive migration cost. On the other hand, Service-Orented Computing (SOC) including service-oriented architecture (SOA) and Web Services (WS) promote standardization and openness with its enabling standards and communication protocols. This thesis proposes a Service-Oriented Cloud Computing Architecture by combining the best attributes of the two paradigms to promote an open, interoperable environment for cloud computing development. Mutil-tenancy SaaS applicantions built on top of SOCCA have more flexibility and are not locked down by a certain platform. Tenants residing on a multi-tenant application appear to be the sole owner of the application and not aware of the existence of others. A multi-tenant SaaS application accommodates each tenant’s unique requirements by allowing tenant-level customization. A complex SaaS application that supports hundreds, even thousands of tenants could have hundreds of customization points with each of them providing multiple options, and this could result in a huge number of ways to customize the application. This dissertation also proposes innovative customization approaches, which studies similar tenants’ customization choices and each individual users behaviors, then provides guided semi-automated customization process for the future tenants. A semi-automated customization process could enable tenants to quickly implement the customization that best suits their business needs.

A story is defined as "an actor(s) taking action(s) that culminates in a resolution(s)''. I present novel sets of features to facilitate story detection among text via supervised classification and further reveal different forms within stories via unsupervised clustering. First, I investigate the utility of a new set of semantic features compared to standard keyword features combined with statistical features, such as density of part-of-speech (POS) tags and named entities, to develop a story classifier. The proposed semantic features are based on triplets that can be extracted using a shallow parser. Experimental results show that a model of memory-based semantic linguistic features alongside statistical features achieves better accuracy. Next, I further improve the performance of story detection with a novel algorithm which aggregates the triplets producing generalized concepts and relations. A major challenge in automated text analysis is that different words are used for related concepts. Analyzing text at the surface level would treat related concepts (i.e. actors, actions, targets, and victims) as different objects, potentially missing common narrative patterns. The algorithm clusters triplets into generalized concepts by utilizing syntactic criteria based on common contexts and semantic corpus-based statistical criteria based on "contextual synonyms''. Generalized concepts representation of text (1) overcomes surface level differences (which arise when different keywords are used for related concepts) without drift, (2) leads to a higher-level semantic network representation of related stories, and (3) when used as features, they yield a significant (36%) boost in performance for the story detection task. Finally, I implement co-clustering based on generalized concepts/relations to automatically detect story forms. Overlapping generalized concepts and relationships correspond to archetypes/targets and actions that characterize story forms. I perform co-clustering of stories using standard unigrams/bigrams and generalized concepts. I show that the residual error of factorization with concept-based features is significantly lower than the error with standard keyword-based features. I also present qualitative evaluations by a subject matter expert, which suggest that concept-based features yield more coherent, distinctive and interesting story forms compared to those produced by using standard keyword-based features.

Computational visual aesthetics has recently become an active research area. Existing state-of-art methods formulate this as a binary classification task where a given image is predicted to be beautiful or not. In many applications such as image retrieval and enhancement, it is more important to rank images based on their aesthetic quality instead of binary-categorizing them. Furthermore, in such applications, it may be possible that all images belong to the same category. Hence determining the aesthetic ranking of the images is more appropriate. To this end, a novel problem of ranking images with respect to their aesthetic quality is formulated in this work. A new data-set of image pairs with relative labels is constructed by carefully selecting images from the popular AVA data-set. Unlike in aesthetics classification, there is no single threshold which would determine the ranking order of the images across the entire data-set.

This problem is attempted using a deep neural network based approach that is trained on image pairs by incorporating principles from relative learning. Results show that such relative training procedure allows the network to rank the images with a higher accuracy than a state-of-art network trained on the same set of images using binary labels. Further analyzing the results show that training a model using the image pairs learnt better aesthetic features than training on same number of individual binary labelled images.

Additionally, an attempt is made at enhancing the performance of the system by incorporating saliency related information. Given an image, humans might fixate their vision on particular parts of the image, which they might be subconsciously intrigued to. I therefore tried to utilize the saliency information both stand-alone as well as in combination with the global and local aesthetic features by performing two separate sets of experiments. In both the cases, a standard saliency model is chosen and the generated saliency maps are convoluted with the images prior to passing them to the network, thus giving higher importance to the salient regions as compared to the remaining. Thus generated saliency-images are either used independently or along with the global and the local features to train the network. Empirical results show that the saliency related aesthetic features might already be learnt by the network as a sub-set of the global features from automatic feature extraction, thus proving the redundancy of the additional saliency module.

In supervised learning, machine learning techniques can be applied to learn a model on

a small set of labeled documents which can be used to classify a larger set of unknown

documents. Machine learning techniques can be used to analyze a political scenario

in a given society. A lot of research has been going on in this field to understand

the interactions of various people in the society in response to actions taken by their

organizations.

This paper talks about understanding the Russian influence on people in Latvia.

This is done by building an eeffective model learnt on initial set of documents

containing a combination of official party web-pages, important political leaders' social

networking sites. Since twitter is a micro-blogging site which allows people to post

their opinions on any topic, the model built is used for estimating the tweets sup-

porting the Russian and Latvian political organizations in Latvia. All the documents

collected for analysis are in Latvian and Russian languages which are rich in vocabulary resulting into huge number of features. Hence, feature selection techniques can

be used to reduce the vocabulary set relevant to the classification model. This thesis

provides a comparative analysis of traditional feature selection techniques and implementation of a new iterative feature selection method using EM and cross-domain

training along with supportive visualization tool. This method out performed other

feature selection methods by reducing the number of features up-to 50% along with

good model accuracy. The results from the classification are used to interpret user

behavior and their political influence patterns across organizations in Latvia using

interactive dashboard with combination of powerful widgets.

In this thesis multiple approaches are explored to enhance sentiment analysis of tweets. A standard sentiment analysis model with customized features is first trained and tested to establish a baseline. This is compared to an existing topic based mixture model and a new proposed topic based vector model both of which use Latent Dirichlet Allocation (LDA) for topic modeling. The proposed topic based vector model has higher accuracies in terms of averaged F scores than the other two models.

Online health forums provide a convenient channel for patients, caregivers, and medical professionals to share their experience, support and encourage each other, and form health communities. The fast growing content in health forums provides a large repository for people to seek valuable information. A forum user can issue a keyword query to search health forums regarding to some specific questions, e.g., what treatments are effective for a disease symptom? A medical researcher can discover medical knowledge in a timely and large-scale fashion by automatically aggregating the latest evidences emerging in health forums.

This dissertation studies how to effectively discover information in health forums. Several challenges have been identified. First, the existing work relies on the syntactic information unit, such as a sentence, a post, or a thread, to bind different pieces of information in a forum. However, most of information discovery tasks should be based on the semantic information unit, a patient. For instance, given a keyword query that involves the relationship between a treatment and side effects, it is expected that the matched keywords refer to the same patient. In this work, patient-centered mining is proposed to mine patient semantic information units. In a patient information unit, the health information, such as diseases, symptoms, treatments, effects, and etc., is connected by the corresponding patient.

Second, the information published in health forums has varying degree of quality. Some information includes patient-reported personal health experience, while others can be hearsay. In this work, a context-aware experience extraction framework is proposed to mine patient-reported personal health experience, which can be used for evidence-based knowledge discovery or finding patients with similar experience.

At last, the proposed patient-centered and experience-aware mining framework is used to build a patient health information database for effectively discovering adverse drug reactions (ADRs) from health forums. ADRs have become a serious health problem and even a leading cause of death in the United States. Health forums provide valuable evidences in a large scale and in a timely fashion through the active participation of patients, caregivers, and doctors. Empirical evaluation shows the effectiveness of the proposed approach.