46,492 research outputs found

    Applied Machine Learning for Health Informatics: Human Motion Analysis and Affective Computing Application

    No full text
    Il monitoraggio della qualità della vita e del benessere della persona rappresenta una sfida aperta nello scenario sanitario. La necessità di risolvere questo task nella nuova era dell'Intelligenza Artificiale porta all’applicazione di metodi dal campo del machine learning. Gli obiettivi e i contributi di questa tesi riflettono le attività di ricerca svolte (i) nell’ambito dell’analisi del movimento: valutazione e monitoraggio automatico del movimento umano durante la riabilitazione fisica, e (ii) nell’ambito dell’affective computing: stima dello stato affettivo del soggetto. Nel primo tema il candidato presenta un algoritmo in grado di estrarre le caratteristiche di movimento clinicamente rilevanti dalle traiettorie dello skeleton acquisite da un sensore RGBD, e fornire un punteggio sulla prestazione del soggetto. L'approccio proposto si basa su regole derivate da indicazioni cliniche e su un algoritmo di machine learning (i.e., Hidden Semi-Markov Model). L'affidabilità dell'approccio proposto è studiata su un dataset collezionato dal candidato rispetto ad un algoritmo gold standard e alla valutazione clinica. I risultati sostengono l'uso della metodologia proposta per la valutazione quantitativa delle prestazioni motorie durante la riabilitazione fisica. Nel secondo topic il candidato propone l’applicazione del framework di Multiple Instance Learning per l'apprendimento della risposta emotiva in presenza di label continui ed ambigui. Questa varaibilità è spesso presente nella risposta affettiva ad uno stimolo esterno (e.g., interazione multimediale). L'affidabilità dell'approccio di Multiple Instance Learning è indagata su un database di benchmark e un dataset più vicino alle problematiche del mondo reale acquisito dal candidato. I risultati ottenuti evidenziano come la metodologia proposta è consistente per la stima dello stato affettivo.The monitoring of the quality of life and the subject's well-being represent an open challenge in the healthcare scenario. The emergence of solving this task in the new era of Artificial Intelligence leads to the application of methods in the machine learning field. The objectives and the contributions of this thesis reflect the research activities performed on the topics of (i) human motion analysis: the automatic monitoring and assessment of human movement during physical rehabilitation and (ii) affective computing: the inferring of the affective state of the subject. In the first topic, the author presents an algorithm able to extract clinically relevant motion features from the RGB-D visual skeleton joints input and provide a related score about subject’s performance. The proposed approach is respectively based on rules derived by clinician suggestions and machine learning algorithm (i.e., Hidden Semi Markov Model). The reliability of the proposed approach is tested over a dataset collected by the author and with respect to a gold standard algorithm and with respect to the clinical assessment. The results support the use of the proposed methodology for quantitatively assessing motor performance during a physical rehabilitation. In the second topic, the author proposes the application of a Multiple Instance Learning (MIL) framework for learning emotional response in presence of continuous and ambiguous labels. This is often the case with affective response to external stimuli (e.g., multimedia interaction). The reliability of the MIL approach is investigated over a benchmark database and one dataset closer to real-world problematic collected by the author. The obtained results point out how the applied methodology is consistent for predicting the human affective response

    Complexity certification and efficient implementation of model predictive control for embedded applications

    No full text
    A causa delle alte frequenze di campionamento e delle ridotte risorse computazionali, la certificazione di complessità ha un ruolo chiave nella determinazione del successo del Model Predictive Control (MPC) nelle applicazioni embedded. Questa tesi propone un algoritmo di certificazione per metodi active-set duali, che permette di calcolare esattamente il tempo massimo di risoluzione di un problema di Quadratic Programming (QP) parametrico, risultante ad esempio da formulazioni MPC lineari. Dato un problema MPC e una piattaforma di calcolo è quindi possibile certificare se il problema di ottimizzazione sarà sempre risolto nel limite di tempo. La mancanza di una certificazione è anche una minaccia per la validità dei metodi di accelerazione, dato che il miglioramento del tempo massimo di soluzione è molto più importante di quello medio per embedded MPC. Due nuovi metodi sono presentati per i quali il miglioramento nel caso peggiore è certificabile esattamente. Il primo è un MPC semi-esplicito che combina un risolutore online con la legge multiparametrica delle partizioni poliedrali che incidono maggiormente sul caso peggiore. Il secondo consiste in una selezione alternativa dei vincoli violati per metodi active-set duali, la quale diminuisce sia il numero massimo di iterazioni, sia la complessità della singola iterazione. Infine, la tesi propone applicazioni sperimentali di embedded MPC a motori elettrici e convertitori di potenza. Il controllo di coppia di un motore brushless tramite MPC è validato su un’unità di controllo economica, risultando più veloce della corrispondente soluzione multiparametrica. Viene poi presentato un controllo MPC per convertitori DC-DC pre-compensati per aggirare il problema dei controllori primali non modificabili. Inoltre, è affrontato il problema della stima dello stato per diversi convertitori nella stessa unità di alimentazione, sviluppando un osservatore robusto e non lineare unificato per sei diverse tipologie di convertitori.Due to the fast sampling frequency and the scarce computational resources, the complexity certification of optimization algorithms plays a key role in determining the success of embedded Model Predictive Control (MPC). This thesis proposes a certification algorithm for dual active-set methods, able to compute exactly the worst-case number of iterations and the amount of time needed to solve a parametric Quadratic Programming (QP) problem, like those that arise in linear MPC. Therefore, given an MPC problem and a computational unit, it can be certified if the optimization problem will be always solved in the prescribed amount of time. The lack of a complexity certification is a threat for accelerating methods as well, as speeding up the worst-case time is much more important than improving the average case in embedded MPC. The thesis presents two novel accelerating methodologies, for which the worst-case improvement can be exactly certified. The first is a semi-explicit MPC, combining an online solver with the multiparametric solution of those polyhedral regions that most affect the worst-case time. The second method consists of an alternative selection for violated constraints in dual active-set solvers, which lowers the worst-case number of iterations and the complexity of the single iteration. Finally, embedded MPC for electrical drives and power converters is experimentally investigated. MPC for the torque control of a brushless motor is demonstrated to be feasible on a cheap control board, and even faster than the corresponding multiparametric solution. Embedded MPC for pre-compensated DC-DC converters is developed, in order to overcome the obstacle of a non-modifiable primal controller, very common in power converters. The issue of estimating the state for multiple DC-DC converters on the same power supply is also addressed, by presenting a unified nonlinear robust observer for six different converter topologies

    Thesis advisor Author

    No full text
    Standardisation of the protection and governance of multimedia content b

    混合金屬團簇化合物之金金屬鍵的斷裂與形成:Te2Fe3(CO)9或Se2Fe3(CO)9與[Co(CO)4]-、[Mn(CO)5]-及[Fe(CO)4]2-之反應研究

    No full text
    [[abstract]]Te2Fe3 (CO)9 reacts with [Co (CO)4]- in THF forms a tetrahedral complex anion [TeFe2Co (CO)9] and the treatment with [Mn (CO)5]- gives a Mn (CO)4 bridged butterfly [Te2Fe2Mn (CO)10]-. Further, the reaction of Te2Fe3 (CO)9 with collman's reagent [Fe (CO)4]2- produces the previously characterized [Te6Fe8 (CO)24]2-. Analogous reaction of Se2Fe3 (CO)9 with [Co (CO)4- affords a tetrahedral anion [SeFe2Co (CO)9. On the other hand, the treatment with Se2Fe3(CO)9 with [Mn(CO)5]- forms a square-pyramidal [Se2Fe2Mn(CO)9]-. Moreover, the reaction of Se2Fe3(CO)9 with [Fe(CO)4]2- forms the known complex [SeFe3(CO)9]2- and [HSeFe3(CO)6]-. [TeFe2Co(CO)9]- [Te2Fe2Mn(CO)10]-, [SeFe2Co (CO)9]- and [Se2Fe2Mn(CO)9]- are all characterized by IR, elemental analysis, or/and X-ray diffraction methods.

    Occupational exposure to electromagnetic fields: risk assessment of operators performing Transcranial Magnetic Stimulation (TMS) treatments

    No full text
    The assessment of the risk from occupational exposure to electromagnetic fields (EMF) has attracted the attention of those involved in safety in the workplace, in particular after the updating of European legislation, with the publication for EMFs, of Directive 2013/35/EU1 of the European Parliament and of the Council, which made the risk assessment mandatory for this type of physical agents. The issue is made even more relevant by the proliferation of industrial and health applications using EMF even of considerable intensity. However, the rapid technological development has not always been accompanied by adequate growth in the culture of prevention and safety. Many devices expose both operators and persons of the general public to significant risks, but often, these risks are not adequately reported by the manufacturer, nor mentioned in the instruction manual, as would be expressly required by the harmonized standards. In this general framework is placed this Ph.D. research project, whose aim is to analyze possible conditions of risk in the workplace, considering only the environment where the EMF sources potentially expose the operator to risk. The research project involves a joint collaboration between two Institutions: the National Institute for Insurance against Accidents at Work - INAIL and of course Sapienza University of Rome. The project is developed in a multidisciplinary manner, providing experimental and numerical investigations to achieve the required goals, also considering the literature review and comparison for a more realistic analysis of the risk, in terms of human exposure to EMF. The work is based on a multiphysics approach to obtain a complete evaluation of the risk in the workplace, with the prospective to improve the current approach in the assessment of the risk and eventually suggest some indications to the operator for better use of the device under test. Therefore, the starting point has been a review of the workplaces to identify any gaps and critical issues in relation to the risk assessment and therefore for which it is considered necessary to deepen the protectionist issues. A literature analysis of the state of the art on the risk in the workplace is first carried out. This has been followed by numerical and accurate modeling of the device under test as well as the workers in a real reproduced work condition of exposure. Of paramount importance is the understanding of all the parameters that can affect the distribution of the induced EM quantities, which are essential for the risk assessment and the verification of compliance with the regulations system. To do this, it was necessary to study human exposure in-depth, also using different human body models available for dosimetric analysis on dedicated software. All the research has traveled on two parallel tracks, on the one hand, the need to fill the scientific gaps in the research area of exposure assessment of workers and on the other one to take into account the regulatory aspects, essential for a correct evaluation of professional exposure. Therefore, as a last step of the overall work, a possible new protocol of risk assessment analysis is proposed to move forward on the improvement of safety and security in the workplace

    Dialogical Skirmishes

    No full text
    Tan was guest editor for 'And Now China?', a special print edition of the Ctrl+P journal, which critically responded to the celebratory rhetoric’s of ‘China Now’ and other celebratory markers of China's global ascent in 2008. As well as the introductory article 'Dialogical Skirmishes', Tan also interviewed Hans Ulrich Obrist

    [[alternative]]Transformation of cluster shape for Co adatoms on Si(111)7×7 surfaces

    No full text
    [[abstract]]Transformation of cluster shape for Co adatoms on Si(111)7x7 surface after plating Co on basis with coverage of 0.29ML、0.41ML、0.52ML、0.71ML and 0.82ML was studied in UHV. The samples were treated at different temperature such as 600℃、700℃、800℃ and 900℃ by LEED (Low Energy Electron Diffraction) and STM (Scanning Tunneling Microscopy). The result shows that the reaction of Co-Si system is very complicated. Some special surface structures such as Pinhole、cloudy islands、triangular islands、clustered islands was observed. We find two factors, coverage of Co on basis and annealing temperature, take great influence on the transformation of surface structure and cluster shape. Sometimes, more than one kind of Co-Si alloy on the surface coexists. Furthermore, those Co-Si compounds have the tendency to spread toward the internal part of substrate after annealing to 700℃.

    [[alternative]]CREATE YOUR OWN FATE :TWO RUNNER’S LIFE STORY

    No full text
    [[abstract]]CREATE YOUR OWN FATE : TWO RUNNER’S LIFE STORY Master’s Thesis, June, 2006 Wu, Bae-Chyi Advisor: Liu, I-Min Ph.D. Abstract “Running is the Way for Runner to Achieve Self-Realization and Mental Fortitude”. This article explored two runners’ experiences of marathon running and presented through the telling of life’s stories. To experience personally what inspirations running could bring to life, the author chose running as a total novice. From the initial physical and mental anguish, struggle and through continual adaptation, training and retrospection, the author gradually discovered the fun of running and the wide spectrum of emotions it generated - tranquility, flowing, soul-searching, torment, bondage, contradiction…etc. These voices enriched and changed the life of the author in a subtle way. The author was also able to discern life’s philosophies in the process. In the lapse of time, the author reviewed his own life story back and forth, and shared the story of runner Ahdar. He re-discovered himself, re-acquainted with himself, and realized that he still held many dreams and that running was a goal he could achieve. The temple of marathon is a symbol. Stepping into this temple is as if breaking through the cocoon in a rebirth, during which pain is a part of the experience. The pain suffered during training for marathon helps accumulate the strength to face future challenges. A person who has completed the pilgrimage of marathon will be likely to gain greater strength to realize other dreams in life. Key words: Running, Runner, Life story

    Co/Ni/Pt(111)的磁性研究

    No full text
    [[abstract]]Experiments were performed in situ in a stainless ultrahigh vacuum (UHV) chamber. In situ magneto-optical Kerr effect (MOKE) was used to study the magnetic properties of the system. I study the magnetic properties of ultrathin Co films on Pt(111) with Ni buffer layers. We prepare dCo Co/1 ML Ni/Pt(111), and the Kerr signals measured at room temperature. The polar and longitudinal Kerr signal was not exist for 0 ~ 3 ML Co/1 ML Ni/Pt(111). The easy axis of the magnetization for 4&5 ML Co/ Pt(111) is in-plane. We find the same result, when dCo Co/1 ML Ni/Pt(111) was prepare at 450K, and the Kerr signals measured at 450K or at room temperature. If we prepare dCo Co/1 ML Ni/Pt(111) at RT, and the Kerr signals measured at 450K. The easy axis of the magnetization for 1 ML Co/ Pt(111) is out-of-plane. The polar and longitudinal Kerr signal was not exist for 2&3 ML Co/1 ML Ni/Pt(111). The easy axis of the magnetization for 4&5 ML Co/ Pt(111) is in-plane. The variation of AES of 1 ML Co/1 ML Ni/Pt(111) as a function of sample temperature. We find the mixing of Ni and Co layers occurs at 420K. The formation of Co-Pt alloy causes the perpendicular magnetic anisotropy. I also study the magnetic properties of ultrathin Ni films on Pt(111) with Co buffer layers. The easy axis of the magnetization for 1 ML Co/ Pt(111) is out of plane. We were surprised that only the polar Kerr signals were observed when dNi 24 ML. Studies of the magnetic properties showed that the easy axis of the magnetization changed from the cant to the out-of-plane direction when the Co buffer layer was inserted on Ni/Pt(111). After different temperature annealing, cause the different competition alloy. The competition in the alloy formations between Co-Pt and Ni-Pt in the Co/Ni/Pt(111) and Ni/Co/ Pt(111) system is also interesting, cause the different magnetic properties. I also try to understand that.

    Evaluating the Effectiveness of a Proactive Community Outreach and Support Team

    No full text
    To better support individuals with mental health related challenges that frequently come into contact with the police, four partner organizations in London, Ontario co-developed the Community Outreach and Support Team (COAST). This program was designed to be health care led and police supported and involves a police officer and a mental health professional responding to calls together. We conducted three studies aimed at evaluating the COAST. The first study is an implementation analysis that examines whether the COAST program is operating in the way that was intended. Based on a range of data sources, we found that while many aspects of the program do align with the original implementation plan, there are also numerous discrepancies, typically due to the fact that the program was adapting operations to emerging issues (e.g., COVID-19). In the second study, we conducted pre- and post-implementation surveys with staff members working within the partner organizations, as well as satisfaction surveys with clients. The results collectively suggest that while the COAST is perceived as a valuable resource that is perceived as beneficial for clients and staff, it may not be having a broader impact on staff knowledge, attitudes, or workload due to its limited size, scope, and availability, as well as communication challenges. The final study examined the impact of the COAST from multiple perspectives based on in-depth interviews with those who had direct contact with the team. A grounded theory was developed to explain how the COAST takes individuals from suffering to enhanced connection to support. The theory also outlines the supportive conditions that facilitate this process and the key barriers that detract from it. The current dissertation contributes to the growing body of literature that examines the impact of co-response teams. While this research generally supports the value of COAST, it also highlighted specific challenges with the program and identifies areas requiring improvement. The findings are discussed in terms of their implications for the London COAST, as well as for other communities that are considering implementing similar initiatives or modifying existing ones to enhance outcomes for clients. Keywords: police crisis response, co-response teams, persons in crisi
    corecore