This thesis investigates interactive scene reconstruction and understanding using RGB-D data only. Indeed, we believe that depth cameras will still be in the near future a cheap and low-power 3D sensing alternative suitable for mobile devices too. Therefore, our contributions build on top of state-of-the-art approaches to achieve advances in three main challenging scenarios, namely mobile mapping, large scale surface reconstruction and semantic modeling. First, we will describe an effective approach dealing with Simultaneous Localization And Mapping (SLAM) on platforms with limited resources, such as a tablet device. Unlike previous methods, dense reconstruction is achieved by reprojection of RGB-D frames, while local consistency is maintained by deploying relative bundle adjustment principles. We will show quantitative results comparing our technique to the state-of-the-art as well as detailed reconstruction of various environments ranging from rooms to small apartments. Then, we will address large scale surface modeling from depth maps exploiting parallel GPU computing. We will develop a real-time camera tracking method based on the popular KinectFusion system and an online surface alignment technique capable of counteracting drift errors and closing small loops. We will show very high quality meshes outperforming existing methods on publicly available datasets as well as on data recorded with our RGB-D camera even in complete darkness. Finally, we will move to our Semantic Bundle Adjustment framework to effectively combine object detection and SLAM in a unified system. Though the mathematical framework we will describe does not restrict to a particular sensing technology, in the experimental section we will refer, again, only to RGB-D sensing. We will discuss successful implementations of our algorithm showing the benefit of a joint object detection, camera tracking and environment mapping.

Self-organising pervasive ecosystems of devices are set to become a major vehicle for delivering infrastructure and end-user services. The inherent complexity of such systems poses new challenges to those who want to dominate it by applying the principles of engineering. The recent growth in number and distribution of devices with decent computational and communicational abilities, that suddenly accelerated with the massive diffusion of smartphones and tablets, is delivering a world with a much higher density of devices in space. Also, communication technologies seem to be focussing on short-range device-to-device (P2P) interactions, with technologies such as Bluetooth and Near-Field Communication gaining greater adoption. Locality and situatedness become key to providing the best possible experience to users, and the classic model of a centralised, enormously powerful server gathering and processing data becomes less and less efficient with device density. Accomplishing complex global tasks without a centralised controller responsible of aggregating data, however, is a challenging task. In particular, there is a local-to-global issue that makes the application of engineering principles challenging at least: designing device-local programs that, through interaction, guarantee a certain global service level. In this thesis, we first analyse the state of the art in coordination systems, then motivate the work by describing the main issues of pre-existing tools and practices and identifying the improvements that would benefit the design of such complex software ecosystems. The contribution can be divided in three main branches. First, we introduce a novel simulation toolchain for pervasive ecosystems, designed for allowing good expressiveness still retaining high performance. Second, we leverage existing coordination models and patterns in order to create new spatial structures. Third, we introduce a novel language, based on the existing ``Field Calculus'' and integrated with the aforementioned toolchain, designed to be usable for practical aggregate programming.

Information is nowadays a key resource: machine learning and data mining techniques have been developed to extract high-level information from great amounts of data. As most data comes in form of unstructured text in natural languages, research on text mining is currently very active and dealing with practical problems. Among these, text categorization deals with the automatic organization of large quantities of documents in priorly defined taxonomies of topic categories, possibly arranged in large hierarchies. In commonly proposed machine learning approaches, classifiers are automatically trained from pre-labeled documents: they can perform very accurate classification, but often require a consistent training set and notable computational effort. Methods for cross-domain text categorization have been proposed, allowing to leverage a set of labeled documents of one domain to classify those of another one. Most methods use advanced statistical techniques, usually involving tuning of parameters. A first contribution presented here is a method based on nearest centroid classification, where profiles of categories are generated from the known domain and then iteratively adapted to the unknown one. Despite being conceptually simple and having easily tuned parameters, this method achieves state-of-the-art accuracy in most benchmark datasets with fast running times. A second, deeper contribution involves the design of a domain-independent model to distinguish the degree and type of relatedness between arbitrary documents and topics, inferred from the different types of semantic relationships between respective representative words, identified by specific search algorithms. The application of this model is tested on both flat and hierarchical text categorization, where it potentially allows the efficient addition of new categories during classification. Results show that classification accuracy still requires improvements, but models generated from one domain are shown to be effectively able to be reused in a different one.

In this thesis we made the first steps towards the systematic application of a methodology for automatically building formal models of complex biological systems. Such a methodology could be useful also to design artificial systems possessing desirable properties such as robustness and evolvability. The approach we follow in this thesis is to manipulate formal models by means of adaptive search methods called metaheuristics. In the first part of the thesis we develop state-of-the-art hybrid metaheuristic algorithms to tackle two important problems in genomics, namely, the Haplotype Inference by parsimony and the Founder Sequence Reconstruction Problem. We compare our algorithms with other effective techniques in the literature, we show strength and limitations of our approaches to various problem formulations and, finally, we propose further enhancements that could possibly improve the performance of our algorithms and widen their applicability. In the second part, we concentrate on Boolean network (BN) models of gene regulatory networks (GRNs). We detail our automatic design methodology and apply it to four use cases which correspond to different design criteria and address some limitations of GRN modeling by BNs. Finally, we tackle the Density Classification Problem with the aim of showing the learning capabilities of BNs. Experimental evaluation of this methodology shows its efficacy in producing network that meet our design criteria. Our results, coherently to what has been found in other works, also suggest that networks manipulated by a search process exhibit a mixture of characteristics typical of different dynamical regimes.

Many industries and academic institutions share the vision that an appropriate use of information originated from the environment may add value to services in multiple domains and may help humans in dealing with the growing information overload which often seems to jeopardize our life. It is also clear that information sharing and mutual understanding between software agents may impact complex processes where many actors (humans and machines) are involved, leading to relevant socioeconomic benefits. Starting from these two input, architectural and technological solutions to enable “environment-related cooperative digital services” are here explored. The proposed analysis starts from the consideration that our environment is physical space and here diversity is a major value. On the other side diversity is detrimental to common technological solutions, and it is an obstacle to mutual understanding. An appropriate environment abstraction and a shared information model are needed to provide the required levels of interoperability in our heterogeneous habitat. This thesis reviews several approaches to support environment related applications and intends to demonstrate that smart-space-based, ontology-driven, information-sharing platforms may become a flexible and powerful solution to support interoperable services in virtually any domain and even in cross-domain scenarios. It also shows that semantic technologies can be fruitfully applied not only to represent application domain knowledge. For example semantic modeling of Human-Computer Interaction may support interaction interoperability and transformation of interaction primitives into actions, and the thesis shows how smart-space-based platforms driven by an interaction ontology may enable natural ad flexible ways of accessing resources and services, e.g, with gestures. An ontology for computational flow execution has also been built to represent abstract computation, with the goal of exploring new ways of scheduling computation flows with smart-space-based semantic platforms.

The continuous advancements and enhancements of wireless systems are enabling new compelling scenarios where mobile services can adapt according to the current execution context, represented by the computational resources available at the local device, current physical location, people in physical proximity, and so forth. Such services called context-aware require the timely delivery of all relevant information describing the current context, and that introduces several unsolved complexities, spanning from low-level context data transmission up to context data storage and replication into the mobile system. In addition, to ensure correct and scalable context provisioning, it is crucial to integrate and interoperate with different wireless technologies (WiFi, Bluetooth, etc.) and modes (infrastructure-based and ad-hoc), and to use decentralized solutions to store and replicate context data on mobile devices. These challenges call for novel middleware solutions, here called Context Data Distribution Infrastructures (CDDIs), capable of delivering relevant context data to mobile devices, while hiding all the issues introduced by data distribution in heterogeneous and large-scale mobile settings. This dissertation thoroughly analyzes CDDIs for mobile systems, with the main goal of achieving a holistic approach to the design of such type of middleware solutions. We discuss the main functions needed by context data distribution in large mobile systems, and we claim the precise definition and clean respect of quality-based contracts between context consumers and CDDI to reconfigure main middleware components at runtime. We present the design and the implementation of our proposals, both in simulation-based and in real-world scenarios, along with an extensive evaluation that confirms the technical soundness of proposed CDDI solutions. Finally, we consider three highly heterogeneous scenarios, namely disaster areas, smart campuses, and smart cities, to better remark the wide technical validity of our analysis and solutions under different network deployments and quality constraints.

Automatically recognizing faces captured under uncontrolled environments has always been a challenging topic in the past decades. In this work, we investigate cohort score normalization that has been widely used in biometric verification as means to improve the robustness of face recognition under challenging environments. In particular, we introduce cohort score normalization into undersampled face recognition problem. Further, we develop an effective cohort normalization method specifically for the unconstrained face pair matching problem. Extensive experiments conducted on several well known face databases demonstrate the effectiveness of cohort normalization on these challenging scenarios. In addition, to give a proper understanding of cohort behavior, we study the impact of the number and quality of cohort samples on the normalization performance. The experimental results show that bigger cohort set size gives more stable and often better results to a point before the performance saturates. And cohort samples with different quality indeed produce different cohort normalization performance. Recognizing faces gone after alterations is another challenging problem for current face recognition algorithms. Face image alterations can be roughly classified into two categories: unintentional (e.g., geometrics transformations introduced by the acquisition devide) and intentional alterations (e.g., plastic surgery). We study the impact of these alterations on face recognition accuracy. Our results show that state-of-the-art algorithms are able to overcome limited digital alterations but are sensitive to more relevant modifications. Further, we develop two useful descriptors for detecting those alterations which can significantly affect the recognition performance. In the end, we propose to use the Structural Similarity (SSIM) quality map to detect and model variations due to plastic surgeries. Extensive experiments conducted on a plastic surgery face database demonstrate the potential of SSIM map for matching face images after surgeries.

The wide diffusion of cheap, small, and portable sensors integrated in an unprecedented large variety of devices and the availability of almost ubiquitous Internet connectivity make it possible to collect an unprecedented amount of real time information about the environment we live in. These data streams, if properly and timely analyzed, can be exploited to build new intelligent and pervasive services that have the potential of improving people's quality of life in a variety of cross concerning domains such as entertainment, health-care, or energy management. The large heterogeneity of application domains, however, calls for a middleware-level infrastructure that can effectively support their different quality requirements. In this thesis we study the challenges related to the provisioning of differentiated quality-of-service (QoS) during the processing of data streams produced in pervasive environments. We analyze the trade-offs between guaranteed quality, cost, and scalability in streams distribution and processing by surveying existing state-of-the-art solutions and identifying and exploring their weaknesses. We propose an original model for QoS-centric distributed stream processing in data centers and we present Quasit, its prototype implementation offering a scalable and extensible platform that can be used by researchers to implement and validate novel QoS-enforcement mechanisms. To support our study, we also explore an original class of weaker quality guarantees that can reduce costs when application semantics do not require strict quality enforcement. We validate the effectiveness of this idea in a practical use-case scenario that investigates partial fault-tolerance policies in stream processing by performing a large experimental study on the prototype of our novel LAAR dynamic replication technique. Our modeling, prototyping, and experimental work demonstrates that, by providing data distribution and processing middleware with application-level knowledge of the different quality requirements associated to different pervasive data flows, it is possible to improve system scalability while reducing costs.

The study of causality has drawn the attention of researchers from many different fields for centuries. In particular, nowadays causal inference is a central question in neuroscience and an entire body of research, called brain effective connectivity, is devoted to detecting causal interactions between distinct brain areas. Brain effective connectivity is typically studied by the statistical analysis of direct measurements of the neural activity. The main purpose of this work is on methods for studying time series causality. More in details, we focus on a well-establish criterion of causality: the Granger criterion, which is based on the concepts of temporal precedence and predictability. Firstly, we consider the standard parametric implementation of the Granger criterion that is based on the multivariate autoregressive model, where we face the problem of model identification. For this purpose, we present a new Bayesian method for linear model identification and we explore its capability of modeling the sparsity structure of the signals. As a second contribution, we look at the causal inference through the lens of machine learning and we propose an approach based on the concept of learning from examples. Thus, given a set of signals, their causal interactions are estimated by a classifier that is trained on a synthetic dataset generated by a parametric model. This approach, that we call supervised parametric approach, is implemented by adopting the Granger criterion of causality and compared with the standard parametric measure of Granger causality. Moreover, the roles of the feature space and the generative model of the training set are investigated through a simulation study. Additionally, we show an example of application on rat neural recordings. Finally, we focus on the bias introduced by parametric methods when applied in a real context, i.e. the inability of having a fully realistic generative model. For this purpose, we analyze how the supervised parametric approach can help in making the inference more application-dependent, by exploiting a physiologically plausible generative model.

Networks of smart interconnected objects have allowed the integration of the artificial world into the physical one. The interaction over a wireless medium is simultaneously the technology enabler and the primary hindering factor. The complexity and variability of the behavior of low power wireless communication is one of the challenges making the design and deployment of a system based on this technology a unique and demanding experience. In this thesis, we describe the deployment of two operational systems for structural health monitoring and adaptive lighting, undertaken by our research group. Our major contribution, among others, covers the definition and implementation of the system services enabling the monitoring infrastructure to guarantee the required quality. The resulting unique design and reliability provide concrete support to the vision of wireless sensor networks as dependable monitoring infrastructure. Despite the success in meeting the user needs, the simple yet effective solutions exploited in the aforementioned deployments make apparent the limitations of the widely used approaches to coordinate access to the communication medium. This thesis also argues that the currently employed solutions at the MAC layer are insufficient to provide guarantees to the resource user. Therefore, we introduce Reins-MAC, a Time-Division Multiple-Access (TDMA) communication scheduler that coordinates access to the medium in a fully decentralized fashion. Limited flexibility, scalability, robustness, as well as increased overhead and complexity are commonly recognized shortcomings of TDMA approaches. Reins-MAC overcomes these limitations by adapting the scheduling to match local availability and natural connectivity variations. Moreover, each node is empowered with full control over its own communication resources. The ability to anarchically apply changes to the communication schedule allows the steering of the resource allocation towards individual needs, dictated by the higher layers in the network stack. The resulting quality and anarchy in accessing the communication resource affect the design and implementation of WSNs, opening new horizons where the application regains control of the primary resource: communication.