Edge computing is the idea of moving computations away from the cloud andinstead perform them at the edge of the network. The benefits of edge computing arereduced latency, increased integrity, and less strain on networks. Edge AI is the practiceof deploying machine learning algorithms to perform computations on the edge.In this project, a pre-trained model YAMNet is retrained and used to perform audioclassification in real-time to detect gunshots, glass shattering, and speech. The modelis deployed onto the edge device both as a full TensorFlow model and as TensorFlowLite models. Comparing results of accuracy, inference time, and memory allocationfor full TensorFlow and TensorFlow Lite models with and without optimization. Resultsfrom this research were that it was a valid option to use both TensorFlow andTensorFlow Lite but there was a lot of performance to gain by using TensorFlow Litewith little downside.

V tomto texte sa zameriame na populárnu koncepciu kolaboratívnej tvorby dokumentov. Predstavíme si myšlienku využitia tohto mechanizmu v rôznych oblastiach rozhodovania, popíšeme si koncept a princíp fungovania. Následne si predstavíme a rozoberieme portály a portletovú technológiu, ich výhody a využitie. Cieľom práce je implementácia kolaboratívneho editora s využitím knižnice pre prácu so zmenami v dokumentoch s perzistentnou a aplikačnou logikou na platforme JEE a vytvorenie jednoduchého portletu pre túto službu.

The demand and collection of hydrometeorological data is growing to support hydrologic and hydraulic analyses, and other studies. These data can amount to extensive information that requires sound data management to enable efficient storage, access, and use. While much of the globe is using technology to efficiently collect and store hydrometeorological data, other parts, such as developing countries, are unable to do so. This thesis presents an assessment of available river observations data in Latin American countries in Central America and the Caribbean. The assessment analyzes 1) access to available data, 2) spatial density of data, and 3) the temporal extents of data. This assessment determines that there are sections of the study area that constitute a drought of data or have limited data available.Furthermore, the development of an internationalized HydroServer Lite, a lite-weight web-based application for database and data management, is undertaken. A pilot program of the translated system in Spanish is established with an agency in each of the following countries: Guatemala, Honduras, and Nicaragua. The internationalized version of HydroServer Lite promises to be a useful tool for these groups. While full implementation is currently underway, benefits include improved database management, access to data, and connectivity to global groups seeking to aid developing countries with hydrometeorological data.

The Association of Research Libraries (ARL) provides an option for librarians to administer a combination of LibQUAL+® and LibQUAL+® Lite to measure users' perceptions of library service quality. LibQUAL+® Lite is a shorter version of LibQUAL+® that uses planned missing data in its design. The present study investigates the loss of information in commonly administered proportions of LibQUAL+® and LibQUAL+® Lite when compared to administering LibQUAL+® alone. Data from previous administrations of LibQUAL+® protocol (2005, N = 525; 2007, N = 3,261; and 2009, N = 2,103) were used to create simulated datasets representing various proportions of LibQUAL+® versus LibQUAL+® Lite administration (0.2:0.8, 0.4:0.6. 0.5:0.5, 0.6:0.4, and 0.8:0.2). Statistics (i.e., means, adequacy and superiority gaps, standard deviations, Pearson product-moment correlation coefficients, and polychoric correlation coefficients) from simulated and real data were compared. Confidence intervals captured the original values. Root mean square errors and absolute and relative biases of correlations showed that accuracy in the estimates decreased with increase in percentage of planned missing data. The recommendation is to avoid using combinations with more than 20% planned missing data.

When an item is submitted to a crime laboratory, the first step in analysis is to examine the item under white light. The next step in examination often involves the use of alternate light sources (ALS) to search for potential biological evidence. Dark adaptation has been suggested as a method to maximize benefits from fluorescent evidence examinations. Dark adaptation refers to the ways the human eye adapts to low light conditions through pupil dilation and rod cell activation. This study evaluates the Crime-lite Eye®, a device designed to help forensic scientists determine their level of dark adaptation before commencing an ALS examination. When tested, the use of the Crime-lite Eye® device resulted in 14% more visualization of fluorescence and 11% more accuracy in pattern recognition, though results were not statistically significant. Since results were more accurate overall, however, and because allowing for a period of dark adaptation would not significantly delay investigation efforts, laboratories should consider implementing a dark adaptation period into their workflow during fluorescent examinations. Additionally, some ALS manufacturers have begun to include a strobing feature in their devices, such as the GlowTorch FN Forensic Light™. It is known from studies on human attention that flickering stimuli are very effective at capturing attention. Any material fluorescing in response to a strobing ALS would look to the examiner to be suddenly appearing and disappearing as a series of sudden onset stimuli, potentially making weak fluorescence easier to see. Participants in this study were asked to examine four plaid shirts in ambient light conditions and dark conditions, while using either a solid ALS or a strobing ALS. Participants were slightly more accurate while using the strobing ALS device (90% vs 82% in ambient light conditions and 93% vs 92% in dark lighting conditions). However, many of them reported feeling eye strain and annoyance while using the strobing feature. As the change in accuracy was not statistically significant, there is no evidence at present that suggests forensic laboratories need to implement a strobing ALS device into their protocols.

No description available.

En este trabajo se presenta el estudio de arquitecturas de bus y la implementaci on de AMBA AHB-Lite en el marco de un sistema que incluye un microprocesador basado en la arquitectura MIPS32, cuya interfaz es adaptada para ser compatible con las especi caciones del protocolo mencionado. Se realiza una comparaci on del sistema implementado con respecto a otras arquitecturas est andar que emplean procesadores como Cortex-M0, OpenRISC1200, ZPU y LEON3, contrastando los requisitos que cada uno insume en sus implementaciones en FPGA y el desempe~no alcanzado por cada uno. / This work presents the study of bus architectures and the implementation of AMBA AHB-Lite in the context of a system that includes a microprocessor based on the MIPS32 architecture, whose interface is modi ed to comply with the protocol's speci cations. The implemented system is compared with other standard architectures that employ microprocessors such as Cortex-M0, OpenRISC1200, ZPU y LEON3, examining the requirements needed for each case in their FPGA implementation and the performance yielded by each of them.

Die grundlegenden neurobiologischen Mechanismen für das Zusammenwirken eines pathologisch erhöhten Körpergewichts und der gewichts-assoziierten emotionalen Belastung sind bisher noch wenig erforscht. Die vorliegende Arbeit untersucht gezielt Abweichungen der regionalen Transporter-Verfügbarkeiten mittels Positronen-Emissions-Tomographie (PET) mit hochselektiven Marker für den Noradrenalin- (NET) sowie den Serotonin-(5-Hydroxytryptamin-)transporter (5-HTT) und funktioneller Magnetresonanztomographie (fMRT) sowie damit verbundene Alterationen der neuronalen Ruhe-(resting-state-)Aktivität konnektierter Hirnregionen. Die Ergebnisse der kombinierten PET/fMRT Analyse wurden mit zwei neuropsychologischen Scores zur Erfassung allgemeiner und gewichtsabhängiger emotionaler Veränderungen korreliert. Insgesamt 48 Teilnehmer (Body-Mass-Index [BMI]: 19 - 50 kg/m2) erhielten eine fMRT und eine PET mittels NET-selektivem [11C]MRB (n = 20) beziehungsweise 5-HTT-selektivem [11C]DASB (n = 28). Die PET ergab im Hypothalamus eine tendentielle, BMI-abhängig verminderte Verfügbarkeit des NET, nicht jedoch des 5-HTT. Zusätzlich fand sich bei steigendem BMI innerhalb beider Neurotransmitternetzwerke in Abhängigkeit zur jeweiligen Transporter-Verfügbarkeit eine teils geschlechtsspezifisch verstärkte funktionelle Konnektivität zwischen dem Hypothalamus und Hirnregionen der Verarbeitung und Bewertung von Nahrungsreizen. Korrelationen der lokalen resting-state Aktivitäten mit den neuropsychologischen Scores lassen vermuten, dass diese Regionen zudem auch mit langfristigen, negativen Veränderungen des gewichtsabhängigen emotionalen Wohlbefindens assoziiert sind. Insgesamt spielen diese beiden zentralen Neurotransmitter-Systeme eine wichtige Rolle in der Modulation von Netzwerken zur Regulation des gewichtsabhängigen emotionalen Wohlbefindens und könnten somit wichtige Anhaltspunkte für neue pharmakologische Ansätze bereitstellen.

Human activity data is typically represented as time series data, and RNNs, often with LSTM cells, are commonly used for recognition in this field. However, RNNs and LSTM-RNNs are often too resource-intensive for real-time applications on resource constrained devices, making them unsuitable. This thesis project is carried out at Wrlds AB, Stockholm. At Wrlds, all machine learning is run in the cloud, but they have been attempting to run their AI algorithms on their embedded devices. The main task of this project was to investigate alternative network structures to minimize the size of the networks to be used on human activity data. This thesis investigates the use of Fast GRNN, a deep learning algorithm developed by Microsoft researchers, to classify human activity on resource-constrained devices. The FastGRNN algorithm was compared to state-of-the-art RNNs, LSTM, GRU, and Simple RNN in terms of accuracy, classification time, memory usage, and energy consumption. This research is limited to implementing the FastRNN algorithm on Nordic SoCs using their SDK and TensorFlow Lite Micro. The result of this thesis shows that the proposed network has similar performance as LSTM networks in terms of accuracy while being both considerably smaller and faster, making it a promising solution for human activity recognition on embedded devices with limited computational resources and merits further investigation. / Rörelse igenkännings analys är oftast representerat av tidsseriedata där ett RNN modell meden LSTM arkitektur är oftast den självklara vägen att ta. Dock så är denna arkitektur väldigt resurskrävande för applikationer i realtid och gör att det uppstår problem med resursbegränsad hårdvara. Detta examensarbete är utfört i samarbete med Wrlds Technologies AB. På Wrlds så körs deras maskin inlärningsmodeller på molnet och lokalt på mobiltelefoner. Wrlds har nu påbörjat en resa för att kunna köra modeller direkt på små inbyggda system. Examensarbete kommer att utvärdera en FastGRNN som är en NN-arkitektur utvecklad av Microsoft i syfte att användas på resurs begränsad hårdvara. FastGRNN algoritmen jämfördes med andra högkvalitativa arkitekturer som RNNs, LSTM, GRU och en simpel RNN. Träffsäkerhet, klassifikationstid, minnesanvändning samt energikonsumtion användes för att jämföra dom olika varianterna. Detta arbete kommer bara att utvärdera en FastGRNN algoritm på en Nordic SoCs och kommer att användas deras SDK samt Tensorflow Lite Micro. Resultatet från detta examensarbete visar att det utvärderade nätverket har liknande prestanda som ett LSTM nätverk men också att nätverket är betydligt mindre i storlek och därmed snabbare. Detta betyder att ett FastGRNN visar lovande resultat för användningen av rörelseigenkänning på inbyggda system med begränsad prestanda kapacitet.

E-mail is an important tool for faculty and staff at the university, college, department, and instructor levels. E-mail is a useful medium in the academic setting for corresponding at all levels. Instructors e-mail students about assignments, lectures, and urgent information: for example, postponed classes and changes in the schedule. In addition e-mail is used to let potential students know about job opportunities. Other routine uses for e-mail include group communications within academic committees and groups of students collaborating on projects. Most users of e-mail who send messages to multiple recipients enter each recipients e-mail address into the TO: field individually or enter the name and e-mail address into a distribution list individually. Both methods are time consuming. This thesis describes a tool for facilitating the use of e-mail for classroom management. This tool, List Maker converts class roll listings to e-mail distribution lists for five common e-mail clients: Pegasus Mail, Eudora Lite, Netscape Messenger, Microsoft Outlook Express, and Microsoft Outlook 97/98. List Maker also converts address books and distribution lists from one e-mail client to another. The List Maker program has been adopted for use in selected departments within ETSU. A survey of the programÆs users indicates that List Maker made it easier for users to create distribution list from class rolls. Efforts to distribute List Maker to other Tennessee Board of Regents (TBR) colleges, unfortunately, have not yet succeeded due to a lack of uniform computing platforms and e-mail policies in the TBR.