<p>Synthesis of terrains with complex features has been a challenging problem in computer graphics since most of the existing methods are based on the height field representation. Complex features in terrains adds to the overall realism of the terrain. Hence, there is a need to synthesize terrains in real-time with complex features that adhere to user input. The methodology described in this thesis describes a novel way to synthesize terrains with complex features based on user drawn sketches. Layered stack data structure is used to ensure that the resulting terrain has complex features. Since, Neural Networks are used to generate the terrains, the process is real-time.</p>

Internet of Things (IoT) is a revolutionary paradigm approaching both industries and consumers everyday life. It refers to a network of addressable physical objects that contain embedded sensing, communication and actuating technologies, to sense and interact with the environment where being deployed. It can be considered as a modern expression of Mark Weiser's vision of ubiquitous computing where tiny networked computers become part of everyday objects, fusing together the virtual world and the physical word. Recent advances in hardware solutions have led to the emergence of powerful wireless IoT systems that are entirely energy-autonomous. These systems extract energy from their environment and operate intermittently, only as power is available. Battery-less sensors present an opportunity for the pervasive wide-spread of remote sensor deployments that require little maintenance and have low cost. As the number of IoT endpoint grows -- industry forecast trillions of connected smart devices in the next few years -- new challenges to program, manage and maintain such a huge number of connected devices are emerging. Web technologies can significantly ease this process by providing well-known patterns and tools - like cloud computing - for developers and users. However, the existing solutions are often too heavyweight or unfeasible for highly resource-constrained IoT devices. This dissertation presents a comprehensive analysis of two of the biggest problems that the IoT is currently facing: R1) How are we going to provide connectivity to all these devices? R2) How can we improve the quality of service provided by these tiny autonomous motes that rely only on limited energy scavenged from the environment? The first contribution is the study and deployment of a Low-Power Wide-Area-Network as a feasible solution to provide connectivity to all the expected IoT devices to be deployed in the following years. The proposed technology offers a novel communication paradigm to address discrete IoT applications, like long-range (i.e., kilometers) at low-power (i.e., tens of mW). Moreover, results highlight the effectiveness of the technology also in the industrial environment thanks to the high immunity to external noises. In the second contribution, we focus on smart metering presenting the design of three smart energy meters targeted to different scenarios. The first design presents an innovative, cost-effective smart meter with embedded non-intrusive load monitoring capabilities intended for the domestic sector. This system shows an innovative approach to provide useful feedback to reduce and optimize household energy consumption. We then present a battery-free non-intrusive power meter targeted for low-cost energy monitoring applications that lower both installation cost due to the non-intrusive approach and maintenance costs associated to battery replacement. Finally, we present an energy autonomous smart sensor with load recognition capability that dynamically adapts and reconfigures its processing pipeline to the sensed energy consumption. This enables the sensor to be energy neutral, while still providing power consumption information every 5 minutes. In the third contribution, we focus on the study of low-power visual edge processing and edge machine learning for the IoT. Two different implementations are presented. The first one discusses an energy-neutral IoT device for precision agriculture, while the second one presents a battery-less long-range visual IoT system, both leveraging on deep learning algorithms to avoid unnecessary wireless data communication. We show that there is a clear benefit from implementing a first layer of data processing directly in-situ where the data is acquired, providing a higher quality of service to the implemented application.

På uppdrag av Tidningsutgivarna har en studie utförts angående olika möjligheter att införa tryck av variabeldata med tryckmetoden inkjet i svensk dagspress. Målet var att undersöka tekniska möjligheter och begränsningar samt att utreda om det fanns något intresse på marknaden som skulle kunna ge någon avkastning. Studien utfördes med hjälp av noggranna källstudier och ett fortlöpande samarbete med olika företag med intresse för dagspress. Rapporten beskriver också företag som arbetar med den här typen av innovationer idag och diverse tidigare projekt med variabeltryckta tidningar. Nya teknologier som eventuellt kan vara av intresse för framtida utveckling har också beskrivits och hur tidningens framtid kommer att se ut ur ett kortare och ett längre perspektiv. Studien visar att inkjetpressar inte klarar av den hastighet som de moderna tidningspressarna håller idag. Men i takt med minskade upplagor och utvecklad teknologi, samt ett stort marknadsmässigt intresse så tyder det på att det kommer att finnas möjlighet att tillverka anpassade digitaltryckta tidningsupplagor inom en överskådlig framtid. Inkjet kan fungera som ett bra komplement till offset. Inlinepressar ger möjligheten att införa variabeltryck i stora tidningsupplagor medan separata inkjetpressar passar bra för tryck av mindre lokala upplagor. / By an assignment from the Swedish Media Publishers’ Association a study has been carried out about various opportunities to introduce printing of variable data with inkjet in Swedish newspapers. The aim was to investigate the technical possibilities and limitations, and if there was any interest in the market that would give any return of investment. The study was conducted by accurate studying of sources and a continuous cooperation with different companies with interest in the newspapers. The study shows that inkjet presses can’t handle the speed of the modern newspaper presses yet today. But it could work as a good complement to offset printing and the technique will continue to develop. It will be launched more printing presses with hybrid solutions in the near future.

Market prices in the liberalized European electricity system play a crucial role in promoting competition, ensuring grid stability, and maximizing profits for market participants. Accurate electricity price forecasting algorithms have, therefore, become increasingly important in this competitive market. However, existing evaluations of forecasting models primarily focus on overall accuracy, overlooking the underlying causality of the predictions. The thesis explores two state-of-the-art forecasters, the deep neural network (DNN) and the Lasso Estimated AutoRegressive (LEAR) models, in the EPEX-BE and Nord Pool markets. The aim is to understand if their predictions can be trusted in more general settings than the limited context they are trained in. If the models produce poor predictions in extreme conditions or if their predictions are inconsistent with reality, they cannot be relied upon in the real world where these forecasts are used in downstream decision-making activities. The results show that for the EPEX-BE market, the DNN model outperforms the LEAR model in terms of overall accuracy. However, the LEAR model performs better in predicting negative prices, while the DNN model performs better in predicting price spikes. For the Nord Pool market, a simpler DNN model is more accurate for price forecasting. In both markets, the models exhibit behaviours inconsistent with reality, making it challenging to trust the models’ predictions. Overall, the study highlights the importance of understanding the underlying causality of forecasting models and the limitations of relying solely on overall accuracy metrics. / Priserna på den liberaliserade europeiska elmarknaden spelar en avgörande roll för att främja konkurrens, säkerställa stabilitet i elnätet och maximera aktörernas vinster. Exakta prisprognoalgoritmer har därför blivit allt viktigare på denna konkurrensutsatta marknad. Existerande utvärderingar av prognosverktyg fokuserar emellertid på den övergripande noggrannheten och förbiser de underliggande orsakssambanden i prognoserna. Denna rapport utforskar två moderna prognosverktyg, DNN (Deep Neural Network) och LEAR (Lasso Estimated AutoRegressive) på elmarknaderna i Belgien respektive Norden. Målsättningen är att förstå om deras prognoser är pålitliga i mer allmänna sammanhang än det begränsade sammahang som de är tränade i. Om modellerna producerar dåliga prognoser under extrema förhållanden eller om deras prognoser inte överensstämmer med verkligheten så kan man inte förlita sig på dem i den verkliga världen, där prognoserna ligger till grund för beslutsfattande aktiviteter. Resultaten för Belgien visar att DNN-modellen överträffar LEAR-modellen när det gäller övergripande noggrannhet. LEAR-modellen presterar dock bättre när det gäller att förutse negativa priser, medan DNN-modellen presterar bättre när det gäller prisspikar. På den nordiska elmarknaden är en enklare DNN-modell mer noggrann för prisprognoser. På båda marknaden visar modellerna beteenden som inte överensstämmer med verkligheten, vilket gör det utmanande att lita på modellernas prognoser. Sammantaget belyser studien vikten av att förstå de underliggande orsakssambanden i prognosmodellerna och begränsningarna med att enbart förlita sig på övergripande mått på noggrannhet.

Future computing will require significant development in new computing device paradigms. This is motivated by CMOS devices reaching their technological limits, the need for non-Von Neumann architectures as well as the energy constraints of wearable technologies and embedded processors. The first device proposal, an energy-efficient voltage-controlled domain wall device for implementing an artificial neuron and synapse is analyzed using micromagnetic modeling. By controlling the domain wall motion utilizing spin transfer or spin orbit torques in association with voltage generated strain control of perpendicular magnetic anisotropy in the presence of Dzyaloshinskii-Moriya interaction (DMI), different positions of the domain wall are realized in the free layer of a magnetic tunnel junction to program different synaptic weights. Additionally, an artificial neuron can be realized by combining this DW device with a CMOS buffer. The second neuromorphic device proposal is inspired by the brain. Membrane potential of many neurons oscillate in a subthreshold damped fashion and fire when excited by an input frequency that nearly equals their Eigen frequency. We investigate theoretical implementation of such “resonate-and-fire” neurons by utilizing the magnetization dynamics of a fixed magnetic skyrmion based free layer of a magnetic tunnel junction (MTJ). Voltage control of magnetic anisotropy or voltage generated strain results in expansion and shrinking of a skyrmion core that mimics the subthreshold oscillation. Finally, we show that such resonate and fire neurons have potential application in coupled nanomagnetic oscillator based associative memory arrays.

京都大学 / 新制・課程博士 / 博士(情報学) / 甲第23311号 / 情博第747号 / 新制||情||128(附属図書館) / 京都大学大学院情報学研究科知能情報学専攻 / (主査)准教授 吉井 和佳, 教授 河原 達也, 教授 西野 恒, 教授 鹿島 久嗣 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM

Humans can visually see things and can differentiate objects easily but for computers, it is not that easy. Computer Vision is an interdisciplinary field that allows computers to comprehend, from digital videos and images, and differentiate objects. With the Introduction to CNNs/DNNs, computer vision is tremendously used in applications like ADAS, robotics and autonomous systems, etc. This thesis aims to propose an architecture, RMNv2, that is well suited for computer vision applications such as ADAS, etc.<br><div>RMNv2 is inspired by its original architecture Mobilenet V2. It is a modified version of Mobilenet V2. It includes changes like disabling downsample layers, Heterogeneous kernel-based convolutions, mish activation, and auto augmentation. The proposed model is trained from scratch in the CIFAR10 dataset and produced an accuracy of 92.4% with a total number of parameters of 1.06M. The results indicate that the proposed model has a model size of 4.3MB which is like a 52.2% decrease from its original implementation. Due to its less size and competitive accuracy the proposed model can be easily deployed in resource-constrained devices like mobile and embedded devices for applications like ADAS etc. Further, the proposed model is also implemented in real-time embedded devices like NXP Bluebox 2.0 and NXP i.MX RT1060 for image classification tasks. <br></div>

Interconnected devices are becoming attractive solutions to integrate physical parameters and making them more accessible for further analysis. Edge devices, located at the end of the physical world, measure and transfer data to the remote server using either wired or wireless communication. The exploding number of sensors, being used in the Internet of Things (IoT), medical fields, or industry, are demanding huge bandwidth and computational capabilities in the cloud, to be processed by Artificial Neural Networks (ANNs) – especially, processing audio, video and images from hundreds of edge devices. Additionally, continuous transmission of information to the remote server not only hampers privacy but also increases latency and takes more power. Deep Neural Network (DNN) is proving to be very effective for cognitive tasks, such as speech recognition, object detection, etc., and attracting researchers to apply it in edge devices. Microcontrollers and single-board computers are the most commonly used types of edge devices. These have gone through significant advancements over the years and capable of performing more sophisticated computations, making it a reasonable choice to implement DNN. In this thesis, a DNN model is trained and implemented for Keyword Spotting (KWS) on two types of edge devices: a bare-metal embedded device (microcontroller) and a robot car. The unnecessary components and noise of audio samples are removed, and speech features are extracted using Mel-Frequency Cepstral Co-efficient (MFCC). In the bare-metal microcontroller platform, these features are efficiently extracted using Digital Signal Processing (DSP) library, which makes the calculation much faster. A Depth wise Separable Convolutional Neural Network (DSCNN) based model is proposed and trained with an accuracy of about 91% with only 721 thousand trainable parameters. After implementing the DNN on the microcontroller, the converted model takes only 11.52 Kbyte (2.16%) RAM and 169.63 Kbyte (8.48%) Flash of the test device. It needs to perform 287,673 Multiply-and-Accumulate (MACC) operations and takes about 7ms to execute the model. This trained model is also implemented on the robot car, Jetbot, and designed a voice-controlled robotic vehicle. This robot accepts few selected voice commands-such as “go”, “stop”, etc. and executes accordingly with reasonable accuracy. The Jetbot takes about 15ms to execute the KWS. Thus, this study demonstrates the implementation of Neural Network based KWS on two different types of edge devices: a bare-metal embedded device without any Operating System (OS) and a robot car running on embedded Linux OS. It also shows the feasibility of bare-metal offline KWS implementation for autonomous systems, particularly autonomous vehicles.<br>

Syftet med studien är att jämföra tre deep learning nätverk med varandra för att ta reda på vilket nätverk som kan producera den högsta uppmätta noggrannheten. Noggrannheten mäts genom att nätverken försöker förutspå antalet personer som vistas i rummet där observation äger rum. Utöver att jämföra de tre djupinlärningsnätverk med varandra, kommer vi även att jämföra dem med en traditionell metoder inom maskininlärning - i syfte för att ta reda på ifall djupinlärningsnätverken presterar bättre än vad traditionella metoder gör. I studien används design and creation. Design and creation är en forskningsmetodologi som lägger stor fokus på att utveckla en IT produkt och använda produkten som dess bidrag till ny kunskap. Metodologin har fem olika faser, vi valde att göra en iterativ process mellan utveckling- och utvärderingfaserna. Observation är den datagenereringsmetod som används i studien för att samla in data. Datagenereringen pågick under tre veckor och under tiden hann 31287 rader data registreras i vår databas. Ett av våra nätverk fick vi en noggrannhet på 78.2%, de andra två nätverken fick en noggrannhet på 45.6% respektive 40.3%. För våra traditionella metoder använde vi ett beslutsträd med två olika formler, de producerade en noggrannhet på 61.3% respektive 57.2%. Resultatet av denna studie visar på att utav de tre djupinlärningsnätverken kan endast en av djupinlärningsnätverken producera en högre noggrannhet än de traditionella maskininlärningsmetoderna. Detta resultatet betyder nödvändigtvis inte att djupinlärningsnätverk i allmänhet kan producera en högre noggrannhet än traditionella maskininlärningsmetoder. Ytterligare arbete som kan göras är följande: ytterligare experiment med datasetet och hyperparameter av djupinlärningsnätverken, samla in mer data och korrekt validera denna data och jämföra fler djupinlärningsnätverk och maskininlärningsmetoder. / The purpose of the study is to compare three deep learning networks with each other to evaluate which network can produce the highest prediction accuracy. Accuracy is measured as the networks try to predict the number of people in the room where observation takes place. In addition to comparing the three deep learning networks with each other, we also compare the networks with a traditional machine learning approach - in order to find out if deep learning methods perform better than traditional methods do. This study uses design and creation. Design and creation is a methodology that places great emphasis on developing an IT product and uses the product as its contribution to new knowledge. The methodology has five different phases; we choose to make an iterative process between the development and evaluation phases. Observation is the data generation method used to collect data. Data generation lasted for three weeks, resulting in 31287 rows of data recorded in our database. One of our deep learning networks produced an accuracy of 78.2% meanwhile, the two other approaches produced an accuracy of 45.6% and 40.3% respectively. For our traditional method decision trees were used, we used two different formulas and they produced an accuracy of 61.3% and 57.2% respectively. The result of this thesis shows that out of the three deep learning networks included in this study, only one deep learning network is able to produce a higher predictive accuracy than the traditional ML approaches. This result does not necessarily mean that deep learning approaches in general, are able to produce a higher predictive accuracy than traditional machine learning approaches. Further work that can be made is the following: further experimentation with the dataset and hyperparameters, gather more data and properly validate this data and compare more and other deep learning and machine learning approaches.

No description available.