17 research outputs found
E il naufragar m'è dolce in questo mare
No full textThis edition is the result of a project run in collaboration with Cambridge Digital Humanities. All of the 41 manuscritps of Giacomo Leopardi are accessible online; they are preceded by the description of the codex gathering them. Each autograph is presented with a high-resolution image and a set of metadata giving information about the author, the date of creation, and the letter addressee (when available). In addition, every item is supplied with a diplomatic transcription accessible as an additional option. Every user can download a file containing the whole text of the autographs marked in XML-TEI
Running Away from the War in Ukraine: The Impact on Mental Health of Internally Displaced Persons (IDPs) and Refugees in Transit in Poland
No full textA growing body of research highlights how communities traumatized by conflict and displacement suffer from long-term mental and psychosocial illnesses. The Russian army’s attack on Ukraine has resulted in an estimated 10 million people being internally or externally displaced from Ukraine, of whom more than 3.8 million have left Ukraine to seek refuge elsewhere in Europe. Soleterre has decided to launch an intervention to provide psychological support to Ukrainian refugees and IDPs, aimed at containing war trauma, assessing the severity of symptoms, and enabling those affected to receive psychological support. The intervention model envisioned the administration of an intake form to provide a rapid collection of qualitative and quantitative information for those arriving in Poland or Lviv from Ukraine. Our results showed how most of the samples reported high or very high levels of anxiety, depression, and sleep disturbances. Moreover, results highlighted how being close to families or being able to keep in touch with them work as a protective factor in enhancing resilience, as well as a support network. These findings underscored the importance of re-thinking our perception of “family” in a broader sense, considering the new facets it can take on in post-conflict situations
Annotations with EARMARK in practice: a fairy tale
No full text• Context: the digitalization of literary texts opens new possibilities since multiple annotations can be added by multiple actors: – author(s); – contributors to documents production and diffusion (editors, proofreaders, reviewers, etc.); – users without write permissions on the documents • Problems: 1. overlapping annotations, i.e. annotations that cannot be expressed inside the strict tree-based hierarchical organization of XML documents; 9/10/2013 2 Motivations (2/2) 2. mechanisms to refer to external entities outside of a document, for example: • to associate a person to the role of an annotator on a specific text; • to link references to people, characters, places, events, etc. inside texts to their representation on the Web (e.g. DBPedia), opening interesting possibilities for data integration, automatic reasoning and semantic analysis in the Linked Open Data; • to state that different text fragments refer to the same univocally identified entity. • Goal: provide a flexible mechanism that supports the creation and sharing of semantic annotations on these documents. 9/10/2013 3 Adding annotations There are two ways to add annotations to existing digital documents: • embedding techniques: annotations are embedded in the document itself – pros: keep all the information in a single coherent file – cons: difficult to deal with overlapping markup. Moreover, it is not always possible (or not desirable) to modify the document • standoff techniques: annotations are stored in a separate document with references to the parts of the document they refer to – pros: greatly simplify the problems related to embedded annotations – cons: serous issues arise when the original document is modifie
Gizmo: The Theatre of Automation in the post-labour society of Hammerfest
No full textWhile technological development and the rising Second Machine Age are already firmly affecting the notion of labour and the consequent disposition of our cities and territories, the remote island of Hammerfest, in the Arctic regions of Finnmark, prefigures the conditions for a radical rebirth. The full extents of automation, resulting in a post-labour organisation of economy, envisions the return of the homo ludens and enhances play as fundamental base of the new society. The explosive decentralisation of production and distribution suddenly wipes out the city, the core centre of our consumerist lives, and promotes the return to the landscape, where new pillars of society are established. On this island of utopian playground, individuals are now fully capable of engendering the shape of their existence, clustering around the ludic elements and interacting with each other in playful activities, while machines take care of their fundamental needs. Among the elements, the Theatre represents the highest form of play and, therefore, the most essential component of public life. Facing the growing automated port and the perished city, the machine of the theatre is a bridge between land and sea, on the particular conditions of a site that constantly changes under the cyclical variations of the tide. In front of the spectacle of automation, what humans stare at is the activity that has been taken away from them, the agency of their actions. A new kind of theatricality is therefore needed. Since the demise of labour invests anyone with the possibility of becoming actors of the play, the sense of spectatorship itself is questioned. Far from the optical relation of the theatron (sight), that has so far embodied its cornerstone, the theatre has now to be subjected to another relation, suggested by the word drama (action), thus endowing the becoming performative of production. The Theatre of Automation is therefore a social endeavour, aiming at bringing the theatrical performance to the level of a collective form of life, where the role of the spectator merges with the one of the actor. In other terms, the machine of the theatre takes active part in the transformation of the site, unfolding and accelerating the artificial manipulation of the territory at the hand of technology. As an act of formation, first, of physical, cultural and ultimately of political nature, the Theatre measures the pace of changes, interpolating the urgency of human artifex with the rhythm of natural systems. When climate change and sea level rise will set the eventual limit of the Machine, interrupting its floating dance on the tides, the Theatre will come to its final act. Rich with the collective memory of the site, it will close its stage and set sail towards new destinations, bearing the paradigm of a new society, on the threshold between human, machine and nature.Architecture, Urbanism and Building Sciences | Transitional Territorie
Learning Interpretable Reduced-order Models for Jumping Quadrupeds
No full textThis work introduces a novel methodology for the development of interpretable reduced-order dynamic models specifically tailored for jumping quadruped robots. Leveraging Symbolic Regression combined with autoencoder neural networks, the framework autonomously derives symbolic equations from data and fundamental physics principles capturing the complex dynamics of jumping actions with high fidelity. This approach significantly reduces model complexity while enhancing interpretability, facilitating deeper insights for legged robotic applications. The efficacy and accuracy of the proposed models are validated through comprehensive experimental studies, marking a substantial advancement in the design of agile and efficient legged robots. This research demonstrates the outperformance of a learned 2D model compared to existing template models such as the ASLIP. Also, an analysis of the dimensionality of the learned model is conducted showing the overarching tradeoff between accuracy and complexity. The method is validated on different simulated quadrupeds and an actual hardware robot.Mechanical Engineering | Vehicle Engineering | Cognitive Robotic
Joel i Amos w dialogu według Massimiliano Scandroglio
No full textArtykuł jest próbą zrecenzowania książki Massimiliano Scandroglio Gioele e Amos in dialogo. Inserzioni redazionali di collegamento e aperture interpretative, opublikowanej nakładem Gregorian&Biblical Press w 2011 r. Książka wpisuje się w dość popularny w dzisiejszej biblistyce nurt egzegezy intertekstualnej, która jest nową odsłoną starej metody porównawczej, badającej związki i zależności pomiędzy poszczególnymi księgami biblijnymi i tym samym proces ich powstawania. Scandroglio skupia swoją uwagę przede wszystkim na rozdziale 4. Księgi Joela i konfrontuje go z rozdziałami 1. i 9. Księgi Amosa. De facto, wnikliwej analizie intertekstualnej zostają poddane następujące perykopy: Jl 4, 1–3/Am 9, 13–15; Jl 4, 4–8/Am 1, 3–2,16; Jl 4, 9–17/Am 1, 2; Jl 4, 18–2/Am 9, 13–15. Ponadto wychodząc od literackich związków między dwoma prorokami, Autor stara się uchwycić podobieństwa i różnice w ich przesłaniu teologicznym, czemu poświęcona jest druga część książki. Na uwagę czytelnika zasługuje przede wszystkim obszerna analiza dwóch teologicznych tematów, a mianowicie motyw „Dnia Pańskiego” i relacja Izraela z obcymi narodami. W Zakończeniu Autor stara się ukazać metodę intertekstualną w jej historycznym rozwoju i wpływ, jaki ma ona obecnie na egzegezę Księgi Joela i Amosa.La tesi dottorale di Massimiliano Scandroglio Gioele e Amos in dialogo. Inserzioni redazionali di collegamento e aperture interpretative, difesa al Pontificio Istituto Biblico di Roma il 4 giugno 2010, è uno studio articolato in due parti: nella prima parte il capitolo 4 di Gioele viene messo a confronto con alcuni versetti dei capitoli 1 e 9 del libro di Amos. I titoli dei rispettivi capitoli si riferiscono alle seguenti pericopi: Gl 4, 1–3 → Am 9, 13–15; Gl 4, 4–8 → Am 1, 3–2, 16; Gl 4, 9–17 → Am 1, 2; Gl 4, 18–21 → Am 9, 13–15, sebbene de facto ne vengano studiati soltanto alcuni versetti o espressioni. Dopo l’analisi dei versetti scelti, fatta nella prima parte del libro, l’Autore – nella seconda – propone una serie di riflessioni su due temi (il Giorno di JHWH in Gl 1, 15–20; 2, 1–11; 3, 1–5; 4, 1–3.9–17 e in Am 2, 6–16; 6, 1–7; 9, 1–4; 5, 18–20; e il rapporto tra Israele e le nazioni in Gl 4 e in Am 1–2; 3, 1–2; 6, 1–7; 9, 7–10), che costituiscono, a suo dire, “la base tematica”, su cui Gioele entra in dialogo con Amos. Nella Conclusione (pp. 339–367) viene offerta la panoramica generale sul cosiddetto “approccio intertestuale” e sulle conseguenze che esso ha per l’esegesi di Gioele ed Amos. Alla fine si trovano Bibliografia, Indice biblico e Indice degli autori.This article attempts to review a book by Massimiliano Scandroglio Gioele e Amos in dialogo. Inserzioni redazionali di collegamento e aperture interpretative which was published by Gregorian&Biblical Press w 2011. This book forms a part of the trend of intertextual exegesis which is popular in contemporary Bible studies. It is a new version of the old comparative method which researches the relations and interdependencies between individual books of the Bible and consequently the process of their formation. Scandroglio focuses mostly on the chapters 4 of the book of Joel and 1 and 9 of Amos. He thoroughly analyses the following excerpts: Jl 4, 1–3/Am 9, 13–15; Jl 4, 4–8/Am 1, 3–2, 16; Jl 4, 9–17/Am 1, 2; Jl 4, 18–2/Am 9, 13–15. Starting from literary connections between the two prophets, the author tries to find similarities and differences in their theological message and he devotes the second part of his book to this subject. Two theological subjects are particularly noteworthy: the topic of the Day of the Lord and the relation between Israel and gentile nations. In conclusion the author tries to present the intersexual method, its historic development and the influence which it currently has on the exegesis of the books of Joel and Amos
The Fate of Patients Treated in Emergency for True-Ruptured Abdominal Aortic Aneurysms After 30 Days and 6 Months: Is There a Benefit from Endovascular Repair?
No full textBackground. Endovascular aneurysm repair (EVAR) has been claimed to decrease periprocedural mortality for treatment of abdominal aortic aneurysms with suspected rupture (rAAA). However the benefit of EVAR in true clinical emergencies (plain ruptured AAA) is still uncertain. This study aimed to analyze immediate and mid term outcomes of true rAAA treated in emergency.
Methods. All patients treated for rAAA from 2006 to 2010 were reviewed. Our policy is to perform an immediate computed tomography (CT) in emergency room (ER) for all patients with suspected rAAA. Only patients with evident aneurysm rupture on imaging were selected for this study. EVAR feasibility was left to the discretion of the operative specialized team available 24hours/day. Stability of EVAR repair was assessed with pre-discharge CT-scan.
Results. A total of 66 patients with plain rAAA were treated in emergency: 23 underwent EVAR and 43 open surgery. At baseline there was an equal distribution of physiologic characteristics and comorbidities with the exception of age, older in the EVAR group. Intraoperative mortality was 0% in EVAR and 9.3% (4/43) in the open group (p=0.29) while 30-day mortality rates were 21.7%(5/23) vs 34.8%(15/43), respectively (p=0.39). Life table estimates at 6 and 15 months showed survival rates of 54% and 50% after open surgery and of 65% and 58% after EVAR (p=0.19). Endoleak presence was detected on CT in 6 EVAR patients who survived until discharge (none of these required immediate treatment due to small size).
Conclusions. Despite the use of EVAR can decrease intraoperative mortality in emergencies for plain rAAA, mortality can involve more than one fifth of these patients at 30-day and one third at six months. The frequent detection of endoleak (more than one fourth of rAAA) at discharge can raise the question whether EVAR might be used as an effective bridge solution for hemodynamic stabilization to be re-evaluated in a later stage.
Author Disclosures: P. De Rango: None. M. Lenti: None. E. Cieri: None. F. Verzini: None. G. Simonte: None. P. Bonanno: None. A. Casalino: None. P. Cao: None.
Key Words: Abdominal aortic aneurysm • Stent • Emergency care • Vascular surgery • Tea
A new machine learning approach to support asset management in water distribution networks
No full textOne of the main causes of the widespread problem of freshwater scarcity lies in unfruitful maintenance of distribution infrastructure, leading to failures with consequent waste of precious resources. It is estimated that more than 25% of the annual loss of water is due to poor conditions of the distribution networks and, in a scenario of continuously increasing demand for water, effects of such inefficiency might be even more dramatic, beyond the merely economic aspect. However, with the rise of data analysis, the awareness of the power of predictive technologies and machine learning techniques, the opportunity to make use of these tools to support decision making has become more than a hope. With this study, the author attempts to address the problem of usage of historical data of pipes and their failures in the Spanish city of Manresa to deduce conclusions on how to conduct maintenance interventions. After conducting an explorative study on how pipes intrinsic factors may have reflections on breakages, machine learning algorithms (Logistic Regression and Random Forest have been chosen in this thesis) are used to predict pipe failures over time. Lately, results from predictions will be used to take out conclusions from two different assessment models. The first method, given the structure of cost of a general distribution company, tries to establish the optimal ratio between sensitivity and sensibility of a predictive model to return the best economic benefit from the predictive maintenance. The second approach wants to assess how the uptime of the service level can be improved whether relying on prediction to replace pipes, given a certain agreed investment budget. In an old industry such as water distribution, difficulties come up not only during the development of predictive models but also during the reconstruction of the data on which training and testing models, since they can suffer from inconsistencies. Indeed, data gathering has not unique and standardized methodologies and time and people take-over have changed procedures during the data collection, making the whole work harde
A new machine learning approach to support asset management in water distribution networks
No full textOne of the main causes of the widespread problem of freshwater scarcity lies in unfruitful maintenance of distribution infrastructure, leading to failures with consequent waste of precious resources. It is estimated that more than 25% of the annual loss of water is due to poor conditions of the distribution networks and, in a scenario of continuously increasing demand for water, effects of such inefficiency might be even more dramatic, beyond the merely economic aspect. However, with the rise of data analysis, the awareness of the power of predictive technologies and machine learning techniques, the opportunity to make use of these tools to support decision making has become more than a hope. With this study, the author attempts to address the problem of usage of historical data of pipes and their failures in the Spanish city of Manresa to deduce conclusions on how to conduct maintenance interventions. After conducting an explorative study on how pipes intrinsic factors may have reflections on breakages, machine learning algorithms (Logistic Regression and Random Forest have been chosen in this thesis) are used to predict pipe failures over time. Lately, results from predictions will be used to take out conclusions from two different assessment models. The first method, given the structure of cost of a general distribution company, tries to establish the optimal ratio between sensitivity and sensibility of a predictive model to return the best economic benefit from the predictive maintenance. The second approach wants to assess how the uptime of the service level can be improved whether relying on prediction to replace pipes, given a certain agreed investment budget. In an old industry such as water distribution, difficulties come up not only during the development of predictive models but also during the reconstruction of the data on which training and testing models, since they can suffer from inconsistencies. Indeed, data gathering has not unique and standardized methodologies and time and people take-over have changed procedures during the data collection, making the whole work harde
Multiple Object Tracking and Face-based Video Retrieval: Applications of Deep Learning to Video Analysis
No full text2019 - 2020In recent years, deep learning (DL) has obtained numerous successes in
analyzing complex data, such as images or audio. A particularly recent area
of application is the analysis of videos.
This thesis focuses on the application of deep learning algorithm to two
video analysis tasks: Multiple Object Tracking (MOT) and Face-based Video
Retrieval (FBVR).
The first main part of the thesis presents an in-depth survey of the state
of the art of DL-based MOT algorithms. This is the first comprehensive survey
specifically on the use of DL for MOT, focusing on 2D frames extracted from
single-camera videos. I identify the four main steps of a MOT algorithm and
describe the various DL techniques used in the literature in each of those four
steps. I also collect and compare results obtained by existing algorithms on
the most common MOT datasets and I analyze the most successful techniques
employed. Finally, I present a discussion about the open issues of current
MOT algorithms and the possible solutions and future directions of research.
The second part of the thesis focuses instead on the task of FBVR. I
present a novel pipeline for the retrieval of unconstrained multi-shot videos
using faces, specifically in the context of television-like videos. Since no
existing dataset in the literature is appropriate for an end-to-end evaluation
of the proposed pipeline, I build a large-scale video dataset by adapting the
VoxCeleb2 dataset to the task of FBVR. I compare and evaluate numerous
DL-based approaches for the various steps in pipeline, such as shot detection,
face detection and face recognition, and I describe the advantages and
disadvantages of each employed technique. The best-performing configuration of the pipeline obtains 97.25% Mean Average Precision on the independent
test set, while performing each query on thousands of videos in less than 0.5
seconds. Finally, I describe the integration of the presented pipeline into the
commercial software TVBridge, developed by CEDEO. [edited by Author]Negli ultimi anni il deep learning (DL) ha avuto molto successo
nell’analisi di dati complessi, quali immagini o audio. Un’area di applicazione
particolarmente recente `e l’analisi di video.
Questa tesi tratta dell’applicazione di algoritmi di deep learning a due
task di analisi video: il Multiple Object Tracking (MOT) e il Face-based Video
Retrieval (FBVR).
La prima parte della tesi presenta un’approfondita revisione della
letteratura sullo stato dell’arte di algoritmi MOT basati su DL. Questa `e la
prima revisione della letteratura a concentrarsi specificamente sull’utilizzo del
DL per il MOT, in particolare per frame 2D estratti da video registrati con una
singola videocamera. Ho identificato i quattro principali passi di un algoritmo
MOT e descritto le varie tecniche di DL utilizzate in letteratura per ciascuno di
questi quattro passi. Ho raccolto e confrontato i risultati ottenuti da algoritmi
in letteratura sui pi`u comuni dataset MOT e ho analizzato le migliori tecniche
utilizzate. Presento infine una discussione riguardo ai problemi aperti degli
algoritmi MOT esistenti, insieme alle possibili soluzioni e alle direzioni future
di ricerca.
La seconda parte della tesi si concentra invece sul task del FBVR. Ho
presentato una pipeline innovativa per la ricerca di video multi-shot senza
restrizioni (unconstrained) tramite l’utilizzo di facce, nel contesto specifico
di video di tipo televisivo. Poich´e nessun dataset esistente in letteratura era
appropriato per una valutazione esaustiva della pipeline proposta, ho costruito
un dataset di video di grandi dimensioni riadattando il dataset VoxCeleb2 al
task del FBVR. Ho confrontato e valutato diversi approcci basati su DL per i vari passi della pipeline, tra cui identificazione degli shot, identificazione
delle facce e riconoscimento facciale. Ho inoltre descritto vantaggi e svantaggi
di ciascuna tecnica utilizzata. La migliore configurazione della pipeline ha
ottenuto una Mean Average Precision pari al 97.25% sul test set indipendente,
il tutto eseguendo ciascuna query su migliaia di video in meno di 0.5 secondi.
Ho infine descritto il processo di integrazione della pipeline nel software
commerciale TVBridge, sviluppato da CEDEO. [a cura dell'Autore]XXXIII cicl
