1731 research outputs found
Sort by
Event-centric management of personal photos
No full textSince the last decade we have been observing a tremendous growth in the size of personal photo collections. For this reason, and due to the lack of proper automatic classification and annotation in standard album-centric photo software, users find it increasingly difficult to organise and make use of their photos. Although automatic annotation of media content can work to achieve more sophisticated multimedia classification and retrieval if its used in combination with rich knowledge representations, it still requires the availability of well-annotated training sets to produce the type of higher-level descriptions that would be of interest to casual users. Thus, the applicability of this approach is highly unlikely in the broad domain of personal photography. Recent developments in the media industry show an interest towards the organisation and structuring of media collections using an event-centric metaphor. This event-centric approach is inspired by strong research in psychology on how our autobiographical memory works to organise, recollect and share our life experiences. While this metaphor is backed by some early user studies, these were led before the large adoption of social media sharing services and there has been little recent research on how users actually use events digitally to organise and share their media. In this work we first present an updated study on what users are doing with their photos in current online platforms to support the suitability of an event-centric approach. Next, we introduce a simple framework for event-centric personal photo management focused on temporal and spatial aspects and through it we describe our techniques for automatic photo organisation and sharing. Finally, we propose a platform for personal photo management that makes use of these automatic techniques and present an evaluation of a prototypical implementation
Braided rivers: an exploratory study combining flume experiments and the analysis of remotely-sensed data
No full textBraided rivers exhibit extremely complex and dynamic morphologies as their multiple channels are constantly re-worked. The research reported in this thesis explored a number of properties of braided river form and dynamics and some controlling factors through three individual but complementary research elements.
The first research element was concerned with some of the controls on the transition between single thread and multi-thread channel patterns. Twenty-seven different flume experiments were conducted, supported by fourteen replicates. In these experiments, channel confinement (maximum possible channel width) and formative discharge were varied in a 25 x 2.9 m flume of constant slope (1%) and bed material (D50 = 1mm) with sediment supply constrained to match sediment output. As the maximum potential channel width increased, the channel pattern changed from a single channel with alternate bars, to the formation of mid-channel bars, and finally to a multi-thread braided pattern. Bed elevation frequency distributions showed distinct changes in their median, standard deviation, skewness and kurtosis as channel width and discharge increased, indicating the consequences of confining braided channels and regulating
discharge on their bed elevation and morphology.
The second and third parts of the research use remotely sensed data sets to explored
(i) the degree to which a real river shows similar characteristics to those generated in the
flume experiments and
(ii) the variety in braiding patterns that are found in association with different boundary conditions of slope, width, discharge, and riparian vegetation.
For the second research element, a Lidar survey of a 36 km reach of the lower Tagliamento river, Italy, was investigated. Within this reach, the river shows only small variations in slope and bed material size and is subject to the same flood flows. Analysis focused on thirty-six 1 km sub-reaches and demonstrated clear associations among the median, standard deviation, kurtosis and skewness of the bed and also clear downstream trends. Measures of vegetation cover showed statistically-significant associations with the median, standard deviation, kurtosis and skewness of the bed, particularly when only the 32 truly braided reaches were analysed. The measures of vegetation cover also showed downstream trends that corresponded with the trends in bed morphology.
Overall, variations in bed morphology showed similar characteristics to those observed in the laboratory flume, but also they showed correspondence with riparian vegetation cover, indicating a topographic signature of vegetation on the bed morphology. The downstream trends appear to be associated with the changing vigour of the riparian vegetation and possibly variations in river baseflow characteristics associated with varying groundwater levels in the alluvial aquifer. The most mature patches of vegetation within the braid plain of the most downstream part of the 36 km reach appear to occur on remnants of braid plain isolated by river bed incision.
The third and final research element considered the morphology of six European
braided rivers of different slope, width, discharge and riparian vegetation type.
Information extracted from Google Earth and other aerial imagery, and gauged river
flow data supported an analysis of changes in braided river characteristics through time,
and among the six European river sites. Four traditional planform indices were used to
characterise the braiding pattern (Bi – braiding index, Ai and Ai2 –anastomosing
indices; Si – main channel sinuosity) were combined with measures of stream power
and its component variables (width, Q10, and slope). Robust data for bed material calibre was not available. Statistical analysis of the entire data set revealed a potential influence of riparian vegetation type on the relationship between unit stream power and braid channel width; and a trend of increasing Bi, Ai, Ai2, and Si with decreasing unit stream power. However, a larger and more complete data set is needed to confirm these general trends and to fully explore transitional rivers.
This research has illustrated the morphological consequences of confining braided rivers and the dependence of the braiding pattern on stream power. It has also illustrated the role of vegetation in contributing to the morphological complexity of braided rivers and the potential role of riparian vegetation in constraining the relationship between stream power and braided river width
Semantic Language models with deep neural Networks
No full textSpoken language systems (SLS) communicate with users in natural language through speech. There are two main problems related to processing the spoken input in SLS. The first one is automatic speech recognition (ASR) which recognizes what the user says. The second one is spoken language understanding (SLU) which understands what the user means. We focus on the language model (LM) component of SLS. LMs constrain the search space that is used in the search for the best hypothesis. Therefore, they play a crucial role in the performance of SLS. It has long been discussed that an improvement in the recognition performance does not necessarily yield a better understanding performance. Therefore, optimization of LMs for the understanding performance is crucial. In addition, long-range dependencies in languages are hard to handle with statistical language models. These two problems are addressed in this thesis.
We investigate two different LM structures. The first LM that we investigate enable SLS to understand better what they recognize by searching the ASR hypotheses for the best understanding performance. We refer to these models as joint LMs. They use lexical and semantic units jointly in the LM. The second LM structure uses the semantic context of an utterance, which can also be described as “what the system understands”, to search for a better hypothesis that improves the recognition and the understanding performance. We refer to these models as semantic LMs (SELMs). SELMs use features that are based on a well established theory of lexical semantics, namely the theory of frame semantics. They incorporate the semantic features which are extracted from the ASR hypothesis into the LM and handle long-range dependencies by using the semantic relationships between words and semantic context. ASR noise is propagated to the semantic features, to suppress this noise we introduce the use of deep semantic encodings for semantic feature extraction. In this way, SELMs optimize both the recognition and the understanding performance
Development of Gold-Magnetite Hybrid Nanoparticles for Advanced Radiotherapy
No full textThe term “theranostics” defines the effort to develop individualized therapies by the combination of diagnostic and therapeutic functions in the same agent. Gold-magnetite hybrid nanoparticles (H-NPs) are proposed as innovative theranostic nanotools for imaging-guided radiosensitization of cancers. H-NPs are designed to exert a dual function: (i) to provide contrast enhancement in magnetic resonance imaging (MRI).and (ii) to enhance radiation effects in the cancer. The imaging and the radiosensitization potentialities of H-NPs arise from the superparamagnetic behaviour of magnetite and the large x-ray extinction coefficient of gold, respectively. Hybrid nanoparticles allows cancer theranostics as the biodistibution of nanoparticles can be tracked by MR imaging, providing a real-time picture of the cancer radiosensitivity profile and allowing precise modulation of radiotherapy.
The purposes of this work are to synthetize properly designed gold-magnetite hybrid nanoparticles and to provide preliminary in vitro evaluations about the potentialities of nanoparticles as MRI-contrast agents and radiosensitizers.
A novel method for the synthesis of hydrophilic and superparamagnetic Tween20-stabilized dumbbell-like gold-magnetite hybrid nanoparticles was set up. Morphology and chemical composition of nanoparticles were assessed by transmission electron microscopy, x-ray diffraction analysis and ion-coupled plasma optical emission spectroscopy. Colloidal stability and magnetic properties of nanoparticles were determined by dynamic light scattering and alternating field magnetometer. The potentialities of H-NPs for MR imaging were studied using a human 4T-MRI scanner. Nanoparticles were proven to induce concentration-dependent contrast enhancement in T2*-weighted MR-images. The biosafety, the cellular uptake and the radiosensitization activity of H-NPs were investigated in human osteosarcoma MG63 cell cultures and murine 3T3 fibroblasts, using specific bioassays and laser scanning confocal microscopy. The results evidenced that nanoparticles were taken up by cells without inducing any cytotoxic effects, even at high nanoparticle concentration. In addition, nanoparticles were proven to induce osteosarcoma-specific reduction of cell viability in clonogenic cell cultures treated with radiotherapy. The experimental results confirmed the potentialities of H-NPs as theranostic tools for MRI-guided radiosensitization. Further studies are needed to confirm our findings and to identify other potential biological targets for MRI-guided radiosensitization
Maternal Relationship, Social Skills and Parental Behavior Through Neuroimaging Techniques and Behavioral Studies
Get PDFMother child relationship is the first and the most important social relationship as it has implications on psychological and neural development of the individual. Here we investigated mother child relationship focusing on different aspects and using a combination of behavioural and neuroimaging techniques. In the first study we addressed the association between brain connectivity and interpersonal competences which are at the basis of every social interaction including the ones involved in mother-child relationship. Several studies suggests that higher White Matter (WM) integrity - an index of increased brain connectivity - , is associated with better cognition and behavioural performance. To test the hypothesis that higher WM integrity is associated with higher interpersonal competence we used Diffusion Tensor Imaging (DTI), a neuroimaging technique which allows to study in vivo the anatomy of boundless of axons conveying information in the brain. Then we correlated this information with a self-reported measure of interpersonal competences: the Adolescent Interpersonal Competence Questionnaire (AICQ). Results indicate that Interpersonal competence is associate with higher WM integrity in several major tracts of the right hemisphere, in specific the uncinate fasciculus, the cingulum, the forceps minor, the infero-fronto occipital fasciculus, the inferior longitudinal fasciculus, and the superior longitudinal fasciculus. These results provide the first direct analysis of the neuroanatomical basis of interpersonal competencies and young adult self-reported skills in social contexts. In the second work we used the same paradigm to test one of the main assumption of the attachment theory which states that social skills highly depends on the quality of attachment relationship. Results show higher integrity in four white matter association fibers in the left hemisphere: Uncinate Fasciculus, Cingulum, Superior Longitudinal Fasciculus and Inferior Fronto Occipital Fasciculus. This result supports the idea that the quality of the attachment relationship influences the emotional and social life of the individual from childhood to adulthood. Furthermore, the research represents an explorative approach to the study of mother-child relationship in healthy population, demonstrating the feasibility of using neuroimaging tools coupled with clinical investigations. Together those studies show that efficient structural connectivity is linked with secure attachment, improved social cognition and cognitive ability. Similarities and differences emerged in these studies will be discussed at the end of Chapter 3 in particular regarding left and right hemisphere specialization. In the second part of the thesis we switched the focus on parenting behaviour. Evidence from the literature suggest an association between Axonal Integrity measured with FA and functional connectivity measured with TMS in two region involved in preparing ad executing actions: premotor and motor cortex. Moreover neuroimaging reveals that infant cries activate parts of the premotor cortical system. In line with this evidence we linked parenting and brain functional connectivity conducting a study on motor cortex excitability in response of infant cries. We used event-related transcranial magnetic stimulation (TMS) to investigated the presence and the time course of modulation of motor cortex excitability in young adults who listened to infant cries. TMS was delivered from 0 to 250 ms from sound onset in six steps of 50 ms in 10 females and 10 males. Motor Evoked Potentials (MEPs) were recorded from the biceps brachii (BB) and interosseus dorsalis primus (ID1) muscles. Results indicate an excitatory modulation of MEPs at 100 ms from the onset of the infant cry specific to females and to the ID1 muscle. This modulation is considered as automatic response to natural cry as it was not present in response of control sounds and the effect is found at 100-ms latency which make this modulation not compatible with a voluntary reaction to the stimulus but suggests an automatic, bottom-up audiomotor association. These results indicate that the brains of adult females appear to be tuned to respond to infant cries with automatic motor excitation. This effect may reflect the greater and longstanding burden on females in caregiving infants. The second part of the thesis continue with a study addressing the natural condition in which baby cries arise when the parent is not attending for infant stimulation. In this study we investigated how infant crying, compared to control sounds, captures adults’ attentive resources. Participants were all nulliparous women and men, we investigated the effects of different sounds on cerebral activation of the default mode network (DMN) while listeners engaged in two different kind of tasks: one designed to activate the DMN ( self-referential decision task) and one designed to deactivate the DMN (syllabic counting tasks). We found a strong deactivation of DMN in woman during baby cry which suggest a shift of attention from self-referential thinking toward the baby cry stimuli. In men we found instead a weaker deactivation of DMN during woman cry while their attention was directed toward an external task and simultaneously a sudden woman crying arise. Gender differences found in our studies and in the literature will be discussed. In the third part of the thesis we investigated the ability to discriminate synchrony and asynchrony during interaction between mother and child with typical or atypical development. We tested two kind of population: in the first study we compared parents of children with Autism Spectrum Disorder (ASD) to parents of children with Typical Development (TD). In the second study we used the Autistic Quotient questionnaire to divide the sample in two groups according with their autistic traits. The relevance of this task is due to the extreme importance to promptly individuate cues of abnormal social behavior in those cases in which the child might shows deficit in the social development. We hypothesized that individual related with a child with ASD or an individual with high autistic traits, might show similar social difficulties as the individual with ASD finding more challenging to detect cue of appropriate or unappropriated social behavior. To test this hypothesis we asked parents of children with ASD and parents of children with TD to judge video of interactions between mothers and child with ASD and mothers and child with TD. Each video were 20s long and depicted either a synchrony or asynchrony interaction, as categorized by an expert clinician. Contrary to our initial hypothesis results indicate that parents of children with ASD are as accurate as parents of children with TD in discriminating synchrony and asynchrony interaction with ASD, however they are less accurate than parents of children with TD in judging interaction with TD. In the second study by testing individual with higher autistic traits (HAQ group) versus lower autistic traits (LAQ group) we confirmed this trend. Using the same paradigm we found that both groups were less accurate during asynchrony interaction. However HAQ was more accurate in judging synchrony interaction with ASD while LAQ was more accurate in judging synchrony interaction with TD. This result indicate a facilitation effects in understanding interaction which include people that share similar characteristic with the observer disconfirming the hypothesis that people with higher autistic traits would have more difficulties in understanding social interactions and pointing the attention on other factors which might contribute during this process. A discussion on the need of further investigation using neuroimaging techniques to understand similarities and differences on neural processing of social interactions is provided at the end of Chapter 4
Too Human To Be a Machine? Social robots, anthropomorphic appearance, and concerns on the negative impact of this technology on humans and their identity.
No full textIn this thesis I will talk about social robots, their appearance, and people’s concerns about potential negative impacts that social robotics technology could have on humans and their identity. The aim is to contribute to the understanding of why people fear social robots, and what the role of humanlike appearance is within this process. Social robots represent a new, fascinating technology. Research in social robotics not only develops new and better social robots but also tries to understand and prevent eventual problems that could arise when people and robots coexist. Moreover, the relations and reactions to social robots, especially those who highly resemble humans, is also an interesting topic from a social psychology point of view. Taking up professor Ishiguro’s words, developing androids opens up the question of “what is human?”. The study of psychological processes related to machines that imitate real persons allows us to know more about ourselves as human beings
Game Experiments with Communication
No full textEconomic experiments allow the investigator to achieve a controlled variation of some particular variable of interest, holding constant other background conditions. However, as it relates to communication, such a controlled variation may not be straightforward to achieve. We critically evaluate the design and implementation of game experiments in economics, and provide a review of the literature focusing mainly on bargaining and trust games (Chapter 1). Following some established results of anticipated communication in the dictator game, we investigate whether strategic considerations crowd out anticipatory effects of communication in the ultimatum game (Chapter 2). Finally, we identify and implement an experimental design that examines the role of first-mover anticipated communication on the inter-player strategic power dynamics that exist in a symmetric simultaneous move prisoners' dilemma and a sequential move investment game (Chapter 3)
Designing New Experiences of Music Making
No full textMusic making is among the activities that best fulfil a person’s full potential, but it is also one of the most complex and exclusive: successful music making requires study and dedication, combined with a natural aptitude that only gifted individuals possess. This thesis proposes new design solutions to reproduce the human ability to make music. It offers insights to provide the general public with novel experiences of music making by exploring a different interactive metaphor. Emotions are proposed as a mediator of musical meanings: an algorithmic composer is developed to generate new music, and the player can interact with the composition, controlling the desired levels of the composition’s emotional character. The adequacy of this metaphor is tested with the case study of The Music Room, an interactive installation that allows visitors to influence the emotional aspect of an original classical style musical composition by means of body movements. This thesis addresses research questions and performs exploratory studies that are grounded in and contribute to different fields of research, including musical interface design, algorithmic composition, and psychology of music. The thesis presents MINUET, a conceptual framework for the design of musical interfaces, and the Music Room, an example of interactive installation based on the emotional metaphor. The Music Room was the result of a two-year iteration of design and evaluation cycles that informed an operational definition of the concept of engagement with interactive art. New methods for evaluating visitors’ experience based on the integration of evidences from different user-research techniques are also presented. As regards the field of algorithmic composition, the thesis presents Robin, a rule-based algorithmic affective composer, and a study to test its validity in eliciting different emotions in listeners (N=33). Valence (positive vs. negative) and arousal (high vs. low) were manipulated in a 2*2 within-subjects design. Results showed that Robin correctly elicited valence and arousal in converging conditions (high valence, high arousal and low valence, low arousal). However, in cases of diverging conditions (high valence, low arousal and low valence, high arousal), valence received neutral values. As regards the psychology of music, this thesis contributes new evidence to the on-going debate about the innate or learned nature of musical competence, defined as the ability to recognise emotion in music. Results of an experimental study framed within Russell’s two-dimensional theory of emotion suggest that musical competence is not affected by training when listeners are required to evaluate arousal (dictated by variations of tempo). The evaluation of valence (dictated by the combination of tempo and mode), however, was found to be more complicated, highlighting a difference in the evaluation of musical excerpts when tempo and mode conveyed diverging emotional information. In this debate, Robin is proposed as a suitable tool for future experimental research as it allows the manipulation of individual musical factors
Towards structured representation of academic search results
No full textSearching for scientific publications is a tedious task, especially when exploring an unfamiliar domain.
Typical scholarly search engines produce lengthy unstructured result lists, which are difficult to comprehend, interpret and browse.An informative visual summary could convey useful information about the returned results as a whole, without the need to sift through individual publications.The first contribution of this thesis is a novel method of representing academic search results with concise and informative topic maps.
The method consists of two steps: i) extracting interrelated topics from the publication titles and abstracts, and ii) summarizing the resulting topic graph. In the first step we map the returned publications to articles and categories of Wikipedia, constructing a graph of relevant topics with hierarchical relations. In the second step we sequentially build a summary of the topic graph that represents the search results in the most informative way. We rely on sequential prediction to automatically learn to build informative summaries from examples. The summarized topic maps share the most of the benefits and avoid most of the drawbacks of the current methods for grouping documents, such as clustering, topic models, and predefined taxonomies.
Specifically, the topic maps are dynamic, fine-grained, of flexible granularity, with up-to-date topics connected with informative relations and having meaningful concise labels.
The second contribution of this thesis is a method for bootstrapping domain-specific ontologies from the categories of Wikipedia. The method performs three steps: i) selecting the set of categories relevant to the domain, ii) classifying the categories into classes and individuals, and iii) classifying the sub-category relations into ``subclass-of'', ``instance-of'', ``part-of'' and ``related-to''.
In each step we rely on binary classification, which makes the method flexible and easily extensible with new features.
For the purpose of academic search, the proposed method advances the creation of semantically rich topic maps.
In general, the method semi-automates the construction of large-scale domain ontologies, benefiting multiple potential applications. Providing ground truth data for structured prediction of large objects, such as topic map summaries or domain ontologies, is tedious. The last contribution of this thesis is an initial investigation into reducing the labeling effort in structured prediction tasks.
First, we present a labeling interface that suggests topics to be added to the ground truth topic map summary.
We modify a state of the art sequential prediction method to iteratively learn from the summaries one topic at a time, while retaining the convergence guarantees. Second, we present an interactive learning method for selecting the categories of Wikipedia relevant to a given domain.
The method reduces the number of required labels by actively selecting the queries to the annotator and learning one label at a time
Cellular mimics that sense and respond to external stimuli
No full textTo date little effort has been expended on the construction of cellular mimics from a minimum number of component parts. Such systems are desirable, because the cellular mimics could serve as useful tools to more deeply delve into the systems level reactions that sustain life and as a platform from which new types of technologies could be generated. Herein the building of cellular mimics that can sense and respond to external stimuli is presented. The majority of our efforts in building cellular mimics are directed towards the sensory element. Initially, previously characterized natural and artificial RNA sensors, i.e. a riboswitches, are exploited. Subsequently, the cellular mimics are implemented as chemical translators for natural bacterial cells.
To expand the capabilities of the engineered cellular mimics, we sought to develop a methodology for the selection of new RNA-based sensors capable of detecting new analytes. The tested methodologies were based on mRNA display and strand displacement reactions. The mRNA display selection did not lead to the identification of a sensor responsive to malachite green after eight cycles of selection. Conversely, via ligand induced triggering of a strand displacement reaction, new RNA sensors for thiamine pyrophosphate were selected from a small library. The sensors displayed translational control ability as is typical of certain classes of riboswitches. The strand displacement-based selection method represents a first step towards the in vitro evolution of sensing elements than can be exploited for new cellular mimics with programmable sensing capability