The Rich Communication Service (RCS) aims to be the default messaging protocol in mobile devices. The native integration of RCS opens up the possibility of RCS chatbots replacing single-purpose apps. This report analyzes this possibility through in-depth experimentation of the chatbot functionality, followed by user testing of the chatbot features. The report found that the RCS chatbot could replace mobile applications since many solve a specific task and follow a closed-loop system. The report also identified RCS as a solution to the accumulated unused apps users have on their mobile devices. This report also attempts to fill the research gap on the current situation of RCS and the reasons behind the different rollout rates of RCS globally. The examination of interviews with RCS stakeholders was the basis of a stakeholder analysis. This analysis found that the Mobile Network Operators have different standings regarding RCS. Some see potential, and some see issues with RCS hindering the global rollout. Furthermore, the reluctance from Apple and the high involvement of Google are reviewed and contrasted with the answers from the interviews. / Rich Communication Service (RCS) siktar på att bli det nya standardprotokollet för att skicka meddelanden mellan mobiler. RCS potentiella integration i mobilens egna meddelande applikation öppnar upp för möjligheten att RCS- chatbotar även kan ersätta enklare appar med begränsade användningsområden. Denna rapport undersöker den här möjligheten genom experimenterande med chatbotens funktionalitet följt av användartester på chatbotens olika funktioner. Rapporten fann att chatboten i RCS har goda möjligheter att ersätta enklare mobila applikationer då många löser specifika uppgifter som följer ett system med en sluten slinga. Rapporten identifierade även RCS som en lösning till den mängd oanvända appar som användare har i sina mobiler. Denna rapport försöker även fylla det forskningsgap som finns angående RCS nuvarande situation och anledningarna till de olika adoptionshastigheterna i världen. Granskning av de intervjuer som hölls med olika intressenter på marknaden blev grunden för den intressentanalys som presenteras. Denna analys visade att teleoperatörer har olika ställningar gentemot RCS. Vissa ser potential medan andra främst ser problem vilket hindrar spridningen av RCS. Dessutom analyserades Apples ovilja att anamma RCS samt Googles

No description available.

Les tâches de vision artificielle telles que la reconnaissance d’objets demeurent irrésolues à ce jour. Les algorithmes d’apprentissage tels que les Réseaux de Neurones Artificiels (RNA), représentent une approche prometteuse permettant d’apprendre des caractéristiques utiles pour ces tâches. Ce processus d’optimisation est néanmoins difficile. Les réseaux profonds à base de Machine de Boltzmann Restreintes (RBM) ont récemment été proposés afin de guider l’extraction de représentations intermédiaires, grâce à un algorithme d’apprentissage non-supervisé. Ce mémoire présente, par l’entremise de trois articles, des contributions à ce domaine de recherche. Le premier article traite de la RBM convolutionelle. L’usage de champs réceptifs locaux ainsi que le regroupement d’unités cachées en couches partageant les même paramètres, réduit considérablement le nombre de paramètres à apprendre et engendre des détecteurs de caractéristiques locaux et équivariant aux translations. Ceci mène à des modèles ayant une meilleure vraisemblance, comparativement aux RBMs entraînées sur des segments d’images. Le deuxième article est motivé par des découvertes récentes en neurosciences. Il analyse l’impact d’unités quadratiques sur des tâches de classification visuelles, ainsi que celui d’une nouvelle fonction d’activation. Nous observons que les RNAs à base d’unités quadratiques utilisant la fonction softsign, donnent de meilleures performances de généralisation. Le dernière article quand à lui, offre une vision critique des algorithmes populaires d’entraînement de RBMs. Nous montrons que l’algorithme de Divergence Contrastive (CD) et la CD Persistente ne sont pas robustes : tous deux nécessitent une surface d’énergie relativement plate afin que leur chaîne négative puisse mixer. La PCD à "poids rapides" contourne ce problème en perturbant légèrement le modèle, cependant, ceci génère des échantillons bruités. L’usage de chaînes tempérées dans la phase négative est une façon robuste d’adresser ces problèmes et mène à de meilleurs modèles génératifs. / High-level vision tasks such as generic object recognition remain out of reach for modern Artificial Intelligence systems. A promising approach involves learning algorithms, such as the Arficial Neural Network (ANN), which automatically learn to extract useful features for the task at hand. For ANNs, this represents a difficult optimization problem however. Deep Belief Networks have thus been proposed as a way to guide the discovery of intermediate representations, through a greedy unsupervised training of stacked Restricted Boltzmann Machines (RBM). The articles presented here-in represent contributions to this field of research. The first article introduces the convolutional RBM. By mimicking local receptive fields and tying the parameters of hidden units within the same feature map, we considerably reduce the number of parameters to learn and enforce local, shift-equivariant feature detectors. This translates to better likelihood scores, compared to RBMs trained on small image patches. In the second article, recent discoveries in neuroscience motivate an investigation into the impact of higher-order units on visual classification, along with the evaluation of a novel activation function. We show that ANNs with quadratic units using the softsign activation function offer better generalization error across several tasks. Finally, the third article gives a critical look at recently proposed RBM training algorithms. We show that Contrastive Divergence (CD) and Persistent CD are brittle in that they require the energy landscape to be smooth in order for their negative chain to mix well. PCD with fast-weights addresses the issue by performing small model perturbations, but may result in spurious samples. We propose using simulated tempering to draw negative samples. This leads to better generative models and increased robustness to various hyperparameters.

The purpose of this study is to investigate and analyze stops that occur in the painting process of trucks; specifically related to the buffer system after the top coat process. Based on this investigation, improvements are presented in order to eliminate stops and improve overall assembly time. With this scope in mind, the following issues have been raised:   1.     What are the root causes for stops in the buffer system? 2.     How much can corrective maintenance be reduced? 3.     What potential savings are achieved by reducing or eliminating stops in the buffer system? 4.     How would reducing stops increase the availability for the top coat paint shop?   A case study has been performed where historical data has been analyzed and staff from the production, maintenance and IT departments have been interviewed. The analyzed data along with a theoretical framework are the basis for the suggested actions that will improve flows through the top coat production. The result indicates that a great part of the delays related to the buffer system are caused by soft losses. The stops are the results of: Lack of a distinct set of rules for how data and software may be used Lack of communication between the workshops The implications of these delays are: increased costs and reduced efficiency of the production line. The purposed actions presented in the study gives the company opportunity to achieve full capacity, eliminate delays and increase overall efficiency. The study is limited to the top coat paint shop and does not include the primary paint shop or final assembly. The results, analysis and suggestions presented in the study includes the soft losses that causes stop in the buffer system. / Syftet med följande studie är att undersöka och analysera stopp och dess rotorsaker som uppkommer i buffertsystemet på ett täcklacksmåleri. Utifrån denna utredning presenteras sedan åtgärdsförslag i syfte att eliminera undersökta stopp. För att kunna uppfylla syftet med studien har följande frågeställningar formulerats:   Vad finns det för stopp relaterade till buffertsystemet och vad är rotorsakerna till dessa? Hur mycket kan det akuta underhållet reduceras? Hur stora besparingar skulle företaget kunna göra genom reducering eller eliminering av stoppen? På vilket sätt kan en minskning av stoppen öka tillgängligheten för täcklacksmåleriet?   En fallstudie har genomförts där historisk data har analyserats och personal från avdelningar som produktion, underhåll och IT har intervjuats. Den empiri som samlats in tillsammans med det teoretiska ramverket som presenteras i rapporten ligger till grund för resultatet, analysen och åtgärdsförslagen.   Resultatet från studien visar att en stor andel av de stopp som sker i buffertsystemet orsakas av mjuka förluster. Dessa stopp är en konsekvens av bristfällig hantering av data och mjukvara samt bristfällig kommunikation mellan verkstäderna. Konsekvenserna av dessa stopp resulterar i ökade kostnader samt en nedsatt effektivitet för täcklacksmåleriet. Åtgärdsförslagen som presenteras ger företaget goda möjligheter att uppnå sin fulla kapacitet, eliminera stoppen och därmed öka effektiviteten.   Studien är avgränsad till måleriprocessen och omfattas inte av grundmåleriet eller monteringen. De resultat, analyser och åtgärdsförslag som presenteras är avgränsade till de mjuka förlusterna som orsakar stopp i buffertsystemet.

Les tâches de vision artificielle telles que la reconnaissance d’objets demeurent irrésolues à ce jour. Les algorithmes d’apprentissage tels que les Réseaux de Neurones Artificiels (RNA), représentent une approche prometteuse permettant d’apprendre des caractéristiques utiles pour ces tâches. Ce processus d’optimisation est néanmoins difficile. Les réseaux profonds à base de Machine de Boltzmann Restreintes (RBM) ont récemment été proposés afin de guider l’extraction de représentations intermédiaires, grâce à un algorithme d’apprentissage non-supervisé. Ce mémoire présente, par l’entremise de trois articles, des contributions à ce domaine de recherche. Le premier article traite de la RBM convolutionelle. L’usage de champs réceptifs locaux ainsi que le regroupement d’unités cachées en couches partageant les même paramètres, réduit considérablement le nombre de paramètres à apprendre et engendre des détecteurs de caractéristiques locaux et équivariant aux translations. Ceci mène à des modèles ayant une meilleure vraisemblance, comparativement aux RBMs entraînées sur des segments d’images. Le deuxième article est motivé par des découvertes récentes en neurosciences. Il analyse l’impact d’unités quadratiques sur des tâches de classification visuelles, ainsi que celui d’une nouvelle fonction d’activation. Nous observons que les RNAs à base d’unités quadratiques utilisant la fonction softsign, donnent de meilleures performances de généralisation. Le dernière article quand à lui, offre une vision critique des algorithmes populaires d’entraînement de RBMs. Nous montrons que l’algorithme de Divergence Contrastive (CD) et la CD Persistente ne sont pas robustes : tous deux nécessitent une surface d’énergie relativement plate afin que leur chaîne négative puisse mixer. La PCD à "poids rapides" contourne ce problème en perturbant légèrement le modèle, cependant, ceci génère des échantillons bruités. L’usage de chaînes tempérées dans la phase négative est une façon robuste d’adresser ces problèmes et mène à de meilleurs modèles génératifs. / High-level vision tasks such as generic object recognition remain out of reach for modern Artificial Intelligence systems. A promising approach involves learning algorithms, such as the Arficial Neural Network (ANN), which automatically learn to extract useful features for the task at hand. For ANNs, this represents a difficult optimization problem however. Deep Belief Networks have thus been proposed as a way to guide the discovery of intermediate representations, through a greedy unsupervised training of stacked Restricted Boltzmann Machines (RBM). The articles presented here-in represent contributions to this field of research. The first article introduces the convolutional RBM. By mimicking local receptive fields and tying the parameters of hidden units within the same feature map, we considerably reduce the number of parameters to learn and enforce local, shift-equivariant feature detectors. This translates to better likelihood scores, compared to RBMs trained on small image patches. In the second article, recent discoveries in neuroscience motivate an investigation into the impact of higher-order units on visual classification, along with the evaluation of a novel activation function. We show that ANNs with quadratic units using the softsign activation function offer better generalization error across several tasks. Finally, the third article gives a critical look at recently proposed RBM training algorithms. We show that Contrastive Divergence (CD) and Persistent CD are brittle in that they require the energy landscape to be smooth in order for their negative chain to mix well. PCD with fast-weights addresses the issue by performing small model perturbations, but may result in spurious samples. We propose using simulated tempering to draw negative samples. This leads to better generative models and increased robustness to various hyperparameters.

Thesis (MScEng)--Stellenbosch University, 2014. / ENGLISH ABSTRACT: Process monitoring and fault diagnosis are used to detect abnormal events in processes. The early detection of such events or faults is crucial to continuous process improvement. Although principal component analysis and partial least squares are widely used for process monitoring and fault diagnosis in the metallurgical industries, these models are linear in principle; nonlinear approaches should provide more compact and informative models. The use of auto associative neural networks or auto encoders provide a principled approach for process monitoring. However, until very recently, these multiple layer neural networks have been difficult to train and have therefore not been used to any significant extent in process monitoring. With newly proposed algorithms based on the pre-training of the layers of the neural networks, it is now possible to train neural networks with very complex structures, i.e. deep neural networks. These neural networks can be used as auto encoders to extract features from high dimensional data. In this study, the application of deep auto encoders in the form of Restricted Boltzmann machines (RBM) to the extraction of features from process data is considered. These networks have mostly been used for data visualization to date and have not been applied in the context of fault diagnosis or process monitoring as yet. The objective of this investigation is therefore to assess the feasibility of using Restricted Boltzmann machines in various fault detection schemes. The use of RBM in process monitoring schemes will be discussed, together with the application of these models in automated control frameworks. / AFRIKAANSE OPSOMMING: Prosesmonitering en fout diagnose word gebruik om abnormale gebeure in prosesse op te spoor. Die vroeë opsporing van sulke gebeure of foute is noodsaaklik vir deurlopende verbetering van prosesse. Alhoewel hoofkomponent-analise en parsiële kleinste kwadrate wyd gebruik word vir prosesmonitering en fout diagnose in die metallurgiese industrieë, is hierdie modelle lineêr in beginsel; nie-lineêre benaderings behoort meer kompakte en insiggewende modelle te voorsien. Die gebruik van outo-assosiatiewe neurale netwerke of outokodeerders bied 'n beginsel gebaseerder benadering om dit te bereik. Hierdie veelvoudige laag neurale netwerke was egter tot onlangs moeilik om op te lei en is dus nie tot ʼn beduidende mate in die prosesmonitering gebruik nie. Nuwe, voorgestelde algoritmes, gebaseer op voorafopleiding van die lae van die neurale netwerke, maak dit nou moontlik om neurale netwerke met baie ingewikkelde strukture, d.w.s. diep neurale netwerke, op te lei. Hierdie neurale netwerke kan gebruik word as outokodeerders om kenmerke van hoë-dimensionele data te onttrek. In hierdie studie word die toepassing van diep outokodeerders in die vorm van Beperkte Boltzmann Masjiene vir die onttrekking van kenmerke van proses data oorweeg. Tot dusver is hierdie netwerke meestal vir data visualisering gebruik en dit is nog nie toegepas in die konteks van fout diagnose of prosesmonitering nie. Die doel van hierdie ondersoek is dus om die haalbaarheid van die gebruik van Beperkte Boltzmann Masjiene in verskeie foutopsporingskemas te assesseer. Die gebruik van Beperkte Boltzmann Masjiene se eienskappe in prosesmoniteringskemas sal bespreek word, tesame met die toepassing van hierdie modelle in outomatiese beheer raamwerke.

In this study, the goal is to optimize the process and to reduce the clogging tendency during the continuous casting process. The focus is on clogging when the refractory base material (RBM) in the SEN is in contact with the liquid steel. It is difficult or impossible to avoid non-metallic inclusions in the liquid steel, but by a selection of a good RBM in the SEN clogging can be reduced.   Different process steps were evaluated during the casting process in order to reduce the clogging tendency. First, the preheating of the SEN was studied. The results showed that the SEN can be decarburized during the preheating process. In addition, decarburization of SEN causes a larger risk for clogging. Two types of plasma coatings were implemented to protect the RBM, to prevent reactions with the RBM, and to reduce the clogging tendency. Calcium titanate (CaTiO3) mixed with yttria stabilized zirconia (YSZ) plasma coatings were tested in laboratory and pilot plant trials, for casting of aluminium-killed low-carbon steels. For casting of cerium alloyed stainless steels, YSZ plasma coatings were tested in laboratory, pilot plant and industrial trials. The results showed that the clogging tendency was reduced when implementing both coating materials.   It is also of importance to produce clean steel in order to reduce clogging. Therefore, the steel cleanliness in the tundish was studied experimentally. The result showed that inclusions originated from the slag, deoxidation products and tundish refractory and that they were present in the tundish as well as in the final steel product. / <p>QC 20170227</p> / VINNOVA

No description available.

No description available.

No description available.