1731 research outputs found
Sort by
Efficient Reasoning with Constrained Goal Models
No full textGOAL models have been widely used in Computer Science to represent software requirements, business objectives, and design qualities. Existing goal modelling techniques, however, have shown limitations of expressiveness and/or tractability in coping with complex real-world problems.
In this work, we exploit advances in automated reasoning technologies, notably Satisfiability and Optimization Modulo Theories (SMT/OMT), and we propose and formalize:
(i) an extended modelling language for goals, namely the Constrained Goal Model (CGM), which makes explicit the notion of goal refinements and of domain as- sumptions, allows for expressing preferences between goals and refinements, and allows for associating numerical attributes to goals and refinements for defining constraints and optimization goals over multiple objective functions, refinements and their numerical attributes;
(i) a novel set of automated reasoning functionalities over CGMs, allowing for automatically generating suitable realizations of input CGMs, under user-specified assumptions and constraints, that also maximize preferences and optimize given objective functions. We are also interested in supporting software evolution caused by changing requirements and/or changes in the operational environment of a software system. For example, users of a system may want new functionalities or performance enhancements to cope with growing user population (requirements evolution). Alternatively, vendors of a system may want to minimize costs in implementing requirements changes (evolution requirements). We propose to use CGMs to represent the requirements of a system and capture requirements changes in terms of incremental operations on a goal model. Evolution requirements are then represented as optimization goals that minimize implementation costs or customer value. We can then exploit reasoning techniques to derive optimal new specifications for an evolving software system.
We have implemented these modelling and reasoning functionalities in a tool, named CGM-Tool, using the OMT solver OptiMathSAT as automated reasoning backend. More- over, we have conducted an experimental evaluation on large CGMs to support the claim that our proposal scales well for goal models with thousands of elements. To access our framework usability, we have employed a user-oriented evaluation using enquiry evaluation method
Video Scene Understanding: Semantic-based representation, Temporal Variation Modeling, Multi-Task Learning
No full textOne of the major research topics in computer vision is automatic video scene understanding where the ultimate goal is to build artificial intelligence systems
comparable with humans in understanding video contents. Automatic video scene understanding covers many applications including (i) semantic functional complex scene categorization, (ii) human body-pose estimation in videos, (iii) human fine-grained daily living action recognition, (vi) video retrieval, and genre recognition. In this thesis, we introduce computer vision and pattern analysis techniques that outperform the state of art of the above mentioned applications on some publicly available datasets.
Our major research contributions towards automatic video scene understanding are (i) introducing an efficient approach to combine low and high-level information content of videos, (ii) modeling temporal variation of frame-based descriptors in videos, and (iii) proposing a multitask learning framework to leverage the huge amount of unlabeled videos.
The first category covers a method for enriching visual words that contain local motion information but they lack information about the cause of the motion. Our proposed approach embeds the source of a generated motion in video descriptors and hence induces some semantic information in the employed visual words in the pattern analysis task. Our approach is validated on traffic scene analysis as well as human body pose estimation applications. When employing an already-trained off-the-shelves model over an unseen dataset, the accuracy of the model usually drops significantly. We present an approach that considers low-level cues such as the optical flow in the foreground of a video to make an already-trained, off-the-shelves, pictorial deformable model work well on a body pose estimation working well for an unseen dataset.
The second category covers methods that induce temporal variation information to video descriptors. Many video descriptors are based on global video representations, where, frame-based descriptors are combined to a unified video descriptor without preserving much of the temporal information content. To include the temporal information content in video descriptors, we introduce a descriptor, namely, the Hard and Soft Cluster Encoding. The descriptor includes how similar frames are distributed over a video timespan. We present that our approach yields significant improvements on the human fine-grained daily living action recognition task.
The third category includes a novel Multi-Task Clustering (MTC) approach to leverage the information of unlabeled videos. Our proposed method is on human fine-grained daily living action recognition application. People tend to perform similar activities in the similar environments. Therefore, a proper clustering approach could determine patterns of fine-grained activities during some learning process. Our proposed MTC approach rather than clustering the data of each individual separately, capture more generic patterns across users over the training data and hence leads to remarkable recognition rates.
Finally, we discuss opportunities for future applications of our research and conclude with a summary of our contributions to video understanding
Modelling environmental changes in the Udzungwa Mountains of Tanzania through impact assessment on rainforest mammals
No full textThe management and conservation of threatened animal populations require accurate knowledge on their distribution and abundance. At the same time, knowledge on the factors driving changes and fluctuation in distribution and abundance is also critical. Nevertheless, gaining such insight is especially challenging for species living in patchy and fragmented landscapes, as is the case for most mammals in tropical forests. This thesis addressed these issues by developing and validating analytical frameworks that allow to make robust spatial inference on population abundance, and ultimately aimed at gaining knowledge on the conservation status of selected mammal species in the rainforest of the Udzungwa Mountains of Tanzania, with emphasis on arboreal primates. This area is an outstanding hotspot for biodiversity and endemism at continental level and it is especially important for primates. Results of the research project were used to provide management recommendations for the conservation of target species and of the environment these inhabit, which is undergoing rapid and critical modifications through habitat depletion and fragmentation
All-Silicon-Based Photonic Quantum Random Number Generators
Get PDFRandom numbers are fundamental elements in different fields of science and technology such as computer simulation like Monte Carlo-method simulation, statistical sampling, cryptography, games and gambling, and other areas where unpredictable results are necessary.
Random number generators (RNG) are generally classified as “pseudo”-random number generators (PRNG) and "truly" random number generators (TRNG). Pseudo random numbers are generated by computer algorithms with a (random) seed and a specific formula. The random numbers produced in this way (with a small degree of unpredictability) are good enough for some applications such as computer simulation. However, for some other applications like cryptography they are not completely reliable. When the seed is revealed, the entire sequence of numbers can be produced. The periodicity is also an undesirable property of PRNGs that can be disregarded for most practical purposes if the sequence recurs after a very long period. However, the predictability still remains a tremendous disadvantage of this type of generators.
Truly random numbers, on the other hand, can be generated through physical sources of randomness like flipping a coin. However, the approaches exploiting classical motion and classical physics to generate random numbers possess a deterministic nature that is transferred to the generated random numbers. The best solution is to benefit from the assets of indeterminacy and randomness in quantum physics.
Based on the quantum theory, the properties of a particle cannot be determined with arbitrary precision until a measurement is carried out. The result of a measurement, therefore, remains unpredictable and random. Optical phenomena including photons as the quanta of light have various random, non-deterministic properties. These properties include the polarization of the photons, the exact number of photons impinging a detector and the photon arrival times. Such intrinsically random properties can be exploited to generate truly random numbers.
Silicon (Si) is considered as an interesting material in integrated optics. Microelectronic chips made from Si are cheap and easy to mass-fabricate, and can be densely integrated. Si integrated optical chips, that can generate, modulate, process and detect light signals, exploit the benefits of Si while also being fully compatible with electronic. Since many electronic components can be integrated into a single chip, Si is an ideal candidate for the production of small, powerful devices. By complementary metal-oxide-semiconductor (CMOS) technology, the fabrication of compact and mass manufacturable devices with integrated components on the Si platform is achievable.
In this thesis we aim to model, study and fabricate a compact photonic quantum random number generator (QRNG) on the Si platform that is able to generate high quality, "truly" random numbers. The proposed QRNG is based on a Si light source (LED) coupled with a Si single photon avalanche diode (SPAD) or an array of SPADs which is called Si photomultiplier (SiPM). Various implementations of QRNG have been developed reaching an ultimate geometry where both the source and the SPAD are integrated on the same chip and fabricated by the same process.
This activity was performed within the project SiQuro—on Si chip quantum optics for quantum computing and secure communications—which aims to bring the quantum world
into integrated photonics. By using the same successful paradigm of microelectronics—the study and design of very small electronic devices typically made from semiconductor materials—, the vision is to have low cost and mass manufacturable integrated quantum photonic circuits for a variety of different applications in quantum computing, measure, sensing, secure communications and services. The Si platform permits, in a natural way, the integration of quantum photonics with electronics. Two methodologies are presented to generate random numbers: one is based on photon counting measurements and another one is based on photon arrival time measurements. The latter is robust, masks all the drawbacks of afterpulsing, dead time and jitter of the Si SPAD and is effectively insensitive to ageing of the LED and to its emission drifts related to temperature variations. The raw data pass all the statistical tests in national institute of standards and technology (NIST) tests suite and TestU01 Alphabit battery without a post processing algorithm. The maximum demonstrated bit rate is 1.68 Mbps with the efficiency of 4-bits per detected photon.
In order to realize a small, portable QRNG, we have produced a compact configuration consisting of a Si nanocrystals (Si-NCs) LED and a SiPM. All the statistical test in the NIST tests suite pass for the raw data with the maximum bit rate of 0.5 Mbps. We also prepared and studied a compact chip consisting of a Si-NCs LED and an array of detectors. An integrated chip, composed of Si p+/n junction working in avalanche region and a Si SPAD, was produced as well. High quality random numbers are produced through our robust methodology at the highest speed of 100 kcps.
Integration of the source of entropy and the detector on a single chip is an efficient way to produce a compact RNG. A small RNG is an essential element to guarantee the security of our everyday life. It can be readily implemented into electronic devices for data encryption. The idea of "utmost security" would no longer be limited to particular organs owning sensitive information. It would be accessible to every one in everyday life
Essays on Productive Efficiency, Trade, and Market Power: Evidence from African Manufacturing Firms
No full textThis thesis examines three main themes, firms productive efficiency, internationalisation of African firms, and effect of liberalisation policies on market power and market imperfections. The thesis combines two main strands in economics literature in accessing the three main themes of the papers. The first strand regards methodological approaches to estimate a production function from which productive efficiency can be computed. Consistent estimation of productive efficiency is a necessary condition to analyse firm behaviour and their response to trade policies. The thesis critically examines methodologies to estimate productive efficiency. The second strand, international trade and industrial development, analyse firms behaviour in foreign market as well as firms responses to trade liberalisation policies and their overall impact on structural transformation. The two strands of literature examined in this thesis resulted in three independent papers, each of which addresses specific issues along the spectrum of productive efficiency estimation, internationalisation, and market power
Indigenous Rights and the Protection of Biodiversity: A Study of Conflict and Reconciliation in International Law
Get PDFIndigenous ways of living are typically described as being harmonious with—if not instrumental for—the protection of the environment. This dissertation moves from the quite different evidence that the protection of biodiversity may encroach on indigenous rights. More specifically, the legal regime of the Convention on Biological Diversity (CBD) establishes obligations for its Parties whose interpretation and/or implementation may lead to the violation of indigenous rights. In this context, this research identifies potential conflicts between the obligations incumbent on CBD Parties pursuant to the CBD and its Nagoya Protocol on access and benefit-sharing (ABS) and those stemming from human rights treaties and protecting indigenous rights. This thesis also develops an interpretative approach aiming to prevent or solve conflicts failing the applicability of hierarchy, lex specialis, or lex posterior rules to the relationship between indigenous rights and the protection of biodiversity. This dissertation argues that systemic interpretation offers a valuable interpretative tool to incorporate the rights of indigenous peoples into the CBD regime. Beyond substantive and procedural indigenous rights, another applicable rule between CBD Parties is the principle of self-determination, which this thesis derives from a teleological interpretation of indigenous rights. The dissertation concludes that conflicts between indigenous rights and obligations established in the CBD regime cannot be solved in the abstract but rather need a case-bycase approach. Evidence from two thematic case studies—one on ABS and the other on conservation—shows that indigenous rights and self-determination allow interpreters both to choose between competing interpretations of the CBD regime and to privilege those interpretations that do not threaten the cultural distinctiveness of indigenous peoples. Successful examples of applying this interpretative approach in the thesis concern issues such as the ownership of genetic resources, the notion of traditional knowledge, and the articulation of concrete forms of participation in the application of CBD-related obligations. These findings have a broader significance for the debate on human rights and the environment, the interplay between self-determination and permanent sovereignty over natural resources, as well as for the harmonization of specialized regimes with the rights of indigenous peoples
New Analytical Methodologies at the Frontier of Cellular Lipidomics
No full textLipids were once thought to only be the building blocks of cell membranes and to serve as energy reserves. With time however, it became increasingly clear that they are actually involved in many more roles. Not surprisingly, the comprehensive characterisation of lipids in cells and tissues has experienced a growing interest worldwide, to the point that the term "lipidomics" was coined. This field is a subset of metabolomics, and the interesting point about these two sciences is that they are closest to the phenotype as compared to their "omics" counterparts (genomics, trascriptomics, ...), because metabolites and lipids are the end products of the –omics cascade. We have investigated mass spectrometry-based lipidomics from different perspectives: first of all, we have devised a targeted approach in which we have focused on sphingolipids and their perturbations. We started by working on neuronal cell cultures where we inhibited GBA, a key enzyme of the sphingolipid metabolism known to be one of the risk factors for Parkinson's disease. We found a significant sphingolipid unbalance characterised by an accumulation of glycosyl-ceramides. We then moved on by investigating the effects that LRRK2, an important and complex protein known to be related to autosomal-dominant forms of the disease, has on sphingolipids. We worked on mouse models, and we compared the sphingolipid profiles of wild-type (Lrrk2+/+) and knock-out (Lrrk2–/–) mice, finding a marked increase in ceramide levels and, more in general, in all lipids downstream of GBA. Such results hint to a possible interaction between LRRK2 and GBA, with LRRK2 playing a role in GBA regulation.
In a second lipidomics investigation, we tried to understand whether or not anti-cancer treatments affect the lipid composition of tumours. Specifically, we concentrated on a common anti-angiogenic drug, whose aim is to starve cancer cells by inhibiting angiogenesis, a process required by the tumours to grow. We considered four different adenocarcinoma cell lines, which were subcutaneously inoculated into mice; the "control" animals received no treatment, whereas the "treated" ones were periodically given the drug. Interestingly, we found the treatment to have significant effects on the cancer lipidome, although the different lines responded unequally to the drug. Such results may reflect the huge heterogeneity of cancers and of individual responses to the treatment. Finally, we developed an informatics algorithm that deals with labelling experiments. The key point is that mass spectrometry measures isotopic patterns of analytes, which depend on the isotopic distribution of the elements; consequently, if an analyte incorporates the stable isotope employed in a labelling experiment, it will show a modified isotopic pattern. Our algorithm analyses such pattern, estimating the abundance of the incorporated label; we first tested it over carefully planned samples, and then we used it in a biochemical application where we wished to establish whether the rate of de novo lipogenesis is influenced by diet. This was accomplished by designing an experiment where mice were given partially deuterated water, while being fed different diets; we were able to ascertain that diet does indeed affect de novo lipogenesis, with the lowest rates occurring on fat-rich diets. We are confident that our tool may find useful applications, considering that stable isotope-based labelling experiments are becoming more and more popular
Development of an Assay to study the Kinetics of HIV-1 Capsid Uncoating
No full textThe acquired immune deficiency syndrome (AIDS) has caused over 60 million deaths since the etiological agent, human immunodeficiency virus type 1 (HIV-1), was first discovered in 1981. Over 6000 new HIV-1 infections are reported every day, predominantly in economically deprived regions of sub-Saharan Africa. Despite impressive developments in antiretroviral therapy, current medical intervention is unable to prevent or cure HIV-1 infection, necessitating expensive life-long treatment. Difficulties in establishing a vaccine or cure, arise from its capacity to cause life-long latent infection, and its extraordinary ability to evolve resistance to therapeutic intervention.
Capsid uncoating is the process by which p24CA proteins (CA) disassemble from the viral ribonucleoprotein during the early phase of the HIV-1 lifecycle. Despite intensive investigation, much is yet unknown about the spatial and temporal occurrence of this process within the cell, and the viral or host cellular factors involved. However, studies investigating p24CA mutations which alter the stability, and consequently the kinetics of capsid uncoating, have shown that timely capsid uncoating is crucial for efficient HIV-1 infection. Recent advancements in microscopic techniques have enabled high resolution analysis of cellular protein interactions, including the in situ localisation and dynamics of these events. We have developed three imaging techniques for the analysis of different aspects of HIV-1 capsid uncoating: 1) a dual-fluorescently labelled virus 2) a fluorescently labelled antibody targeting a capsid internalised repeat peptide array, and 3) a split-luciferase system tagging an internalised component of the capsid core.
Fluorescent labelling of both the CA and IN proteins enabled the sensitive and specific analysis of uncoating in response to both restriction factors and CA mutations, and the visualisation of CA colocalised pre-integration complexes (PICs) within the nucleus. This system is ideally suited for studying the longer-term kinetics of uncoating, and the in-situ visualisation of protein interactions. The use of the repeat peptide array in conjunction with fluorescently labelled antibodies, reinforced reports of an initial uncoating event early after viral fusion. This system enabled the rapid and reproducible imaging of uncoating events in real-time, within the same cell sample population. Finally, the split-luciferase system added further weight to a primary early uncoating stage, and showed capsid disassembly responses specific to mutations within the p24CA that affect the stability of the viral core. Put together, these three assays support a model of uncoating involving an initial early phase of uncoating, followed by a more gradual disassembly of CA from the PICs during cytoplasmic trafficking towards the nucleus. The colocalisation of these components within the nucleus suggests incomplete uncoating at the time of nuclear docking. The user-friendliness of the split-luciferase system, along with its capacity for high-throughput, real-time analysis, support great potential for its use as a screening assay for testing antiviral compounds targeting capsid uncoating events
Decoration of graphene sheets with metal and metal oxide nanostructures by low-pressure plasma deposition
Get PDFThis thesis was dedicated to decorate graphene sheets with metal and metal oxide nanostructures by RF sputtering technique. Two main objectives were focused in this thesis. 1) To decorate graphene sheets uniformly with metal and metal oxide nanostructure without agglomeration. 2) To explore different kinds of application of decorated graphene sheets with metal and metal oxide nanostructures
In the first step, we presented the experimental study results about Nb2O5 deposition onto graphite nanoplatelets (GNPs) by the variation of the deposition process parameters. The structural, chemical and electronic properties of the decorated GNPs with Nb2O5 layers were studied. It was found that with deposition of Nb2O5 layers onto GNPs, tensile strain was developed into the planes of the GNPs. The induced tensile strain in and between the planes of GNPs increased with raising the amount of the Nb2O5 concentration. TEM images shows that GNPs decorated with around 5 to10 nm uniform layer of Nb2O5 at 100 W on their surface were successfully fabricated. From the XPS analysis it was confirmed that, by increasing Nb2O5 layer thickness on the GNPs surface with rising RF power values binding energy downshift in C 1s peak suggests a p-type doping of GNPs due to charge transfer at the interface as a consequence of the higher work function difference between the Nb2O5 (4.70 eV) and GNPs (4.33 eV).
In the second step, the interface between the graphene sheets and Nb2O5 nanoparticles were studied. It was established that the structural defects were pronounced with increasing amounts of the Nb2O5 concentration. XPS measurement on graphene/Nb2O5 suggests p-type doping of graphene due to charge transfer at the interface as a consequence of the high work function of Nb2O5. The strong p-doping effect was also confirmed by Raman analysis where the positions of the G and 2D peaks of graphene gradually upshifted upon increasing the Nb2O5 concentration. The uniform distribution of decorated Nb2O5 nanoparticles onto graphene was confirmed from TEM analysis. The ferromagnetic behavior was observed for the undecorated graphene and decorated graphene with Nb2O5 nanoparticles. The ferromagnetic behavior of graphene was enhanced with decoration of the Nb2O5 nanoparticles.
In the third step, the effect of the Mg concentration on the structural, chemical and morphological properties of the graphene was described. Well dispersed Mg nanoparticles were decorated onto graphene sheets. It was found that from the XRD results, different sizes of the crystalline Mg nanoparticles were obtained onto graphene sheets with variation of the process parameters.. Raman spectra indicated that G and 2D bands of the graphene were shifted to higher wavenumber with deposition of Mg nanoparticles. The well dispersed and small size of Mg nanoparticles in the range of (8-12 nm) onto graphene sheets was decorated by using a high powder vibration frequency. No agglomeration of the sputtered particles was observed with high powder vibration frequency. This observation was confirmed by TEM micrographs. XPS analysis revealed that the decorated Mg nanoparticles onto graphene were oxidized due to exposure to the atmosphere. The well dispersed decorated Mg nanoparticles onto graphene sheets were studied for the hydrogen absorption and desorption at two different temperatures 330 oC and 360 oC at 2 and 8 bars pressure. The hydrogen up taking capacity for the decorated graphene sheets with Mg nanoparticles was 3 wt. % in whole composite. However, the up taking hydrogen storage capacity of the only Mg nanoparticles was 6.6 wt. %.
In the last step, the interaction of the graphene sheets with TiO2 nanoparticles was studied. The XRD results indicated that the lattice of the graphene sheets was distorted with increasing amount of the TiO2 concentration. The particle nature of the deposited TiO2 was confirmed by TEM examination and also the TEM analysis shows that TiO2 nanoparticles were uniformly distributed onto graphene sheets. The Raman analysis showed that the G and 2D bands of graphene were shifted to higher wavenumber with increasing TiO2 concentration onto graphene sheet confirming the p doped graphene with TiO2 nanoparticles. The XPS analysis further confirmed the p doping of graphene upon the deposition of the TiO2 nanoparticles. The binding energy downshift the C 1s core level of was observed after charge transfer from graphene to TiO2 nanoparticles due to the larger work function of TiO2 relatively to that of graphene. It was observed that decorated graphene sheets with TiO2 nanoparticles shows reasonably catalytic activity
Development of multilayer for protection from intense electric fields
Get PDFThe experimental work presented in this thesis is done to develop an innovative procedure to create a protective nanostructured coating inside the X-band radio frequency cavity, a key component in future particle accelerator.
The scope of the multilayer coating is to prevent the breakdown due to high electric and magnetic field. In fact the electrical discharges damage, in irreversible way, the internal surface of the cavity and compromise the final operation of the device. The keen interest on the topic is due to decrease the length and the cost of the next generation linear accelerator. To do this it is essential to enhance the performance of X-band Linacs up to 100MV/m accelerating gradient and to maintain, high as possible, the electrical breakdown reliability. Several studies are made on different materials in order to develop these cavities [1] [2], but the use of physical vapor deposition technique (PVD), to obtain nanostructured coating directly on internal wall of these small sized cavities is not reported in literature. The size of the cavities is of order of few millimeters and the iris aperture ranges from 2 to 6mm: for this reason the direct PVD coating is not possible. Hence a mandrel, that is the negative shape of the cavity, is first coated using PVD technique and finally chemically dissolved after copper electroforming[3]. The novel nanostructured coating is a multilayer composed by two high purity and immiscible metals. One is Copper to guarantee electrical conductivity of the cavity and the second is Molybdenum because it is a refractory metal. Moreover the choice of immiscible materials is important, because these materials do not form alloy during the deposition phase.
Keeping a well-defined interface is important to guarantee a barrier effect to the motion of the defects inside the cavity’s material[4][5]. The experimental part of the thesis is divided in three different parts: design and setup of the PVD deposition system, plasma discharge analysis and, finally, the characterization of the coatings.
This work is a collaboration between Industrial Engineering Department (University of Trento) and the National Laboratory of Legnaro (National Institute of Nuclear Physics LNL-INFN), but this research involves several institutes in different countries: SLAC (USA), KEK (Japan) and UCLA (Los Angeles USA)