The Internet of Things (IoT) computing paradigm has connected smart objects “things” and has brought new services at the proximity of the user. Edge Computing, a natural evolution of the traditional IoT, has been proposed to deal with the ever-increasing (i) number of IoT devices and (ii) the amount of data traffic that is produced by the IoT endpoints. EC promises to significantly reduce the unwanted latency that is imposed by the multi-hop communication delays and suggests that instead of uploading all the data to the remote cloud for further processing, it is beneficial to perform computation at the “edge” of the network, close to where the data is produced. However, bringing computation at the edge level has created numerous challenges as edge devices struggle to keep up with the growing application requirements (e.g. Neural Networks, or video-based analytics). In this thesis, we adopt the EC paradigm and we aim at addressing the open challenges. Our goal is to bridge the performance gap that is caused by the increased requirements of the IoT applications with respect to the IoT platform capabilities and provide latency- and energy-efficient computation at the edge level. Our first step is to study the performance of IoT applications that are based on Deep Neural Networks (DNNs). The exploding need to deploy DNN-based applications on resource-constrained edge devices has created several challenges, mainly due to the complex nature of DNNs. DNNs are becoming deeper and wider in order to fulfill users expectations for high accuracy, while they also become power hungry. For instance, executing a DNN on an edge device can drain the battery within minutes. Our solution to make DNNs more energy and inference friendly is to propose hardware-aware method that re-designs a given DNN architecture. Instead of proxy metrics, we measure the DNN performance on real edge devices and we capture their energy and inference time. Our method manages to find alternative DNN architectures that consume up to 78.82% less energy and are up to35.71% faster than the reference networks. In order to achieve end-to-end optimal performance, we also need to manage theedge device resources that will execute a DNN-based application. Due to their unique characteristics, we distinguish the edge devices into two categories: (i) a neuromorphic platform that is designed to execute Spiking Neural Networks (SNNs), and (ii) a general-purpose edge device that is suitable to host a DNN. For the first category, we train a traditional DNN and then we convert it to a spiking representation. We target the SpiNNaker neuromorphic platform and we develop a novel technique that efficiently configures the platform-dependent parameters, in order to achieve the highest possible SNN accuracy.Experimental results show that our technique is 2.5× faster than an exhaustive approach and can reach up to 0.8% higher accuracy compared to a CPU-based simulation method. Regarding the general-purpose edge devices, we show that a DNN-unaware platform can result in sub-optimal DNN performance in terms of power and inference time. Our approachconfigures the frequency of the device components (GPU, CPU, Memory) and manages to achieve average of 33.4% and up to 66.3% inference time improvements and an average of 42.8% and up to 61.5% power savings compared to the predefined configuration of an edge device. The last part of this thesis is the offloading optimization between the edge devicesand the gateway. The offloaded tasks create contention effects on gateway, which can lead to application slowdown. Our proposed solution configures (i) the number of application stages that are executed on each edge device, and (ii) the achieved utility in terms of Quality of Service (QoS) on each edge device. Our technique manages to (i) maximize theoverall QoS, and (ii) simultaneously satisfy network constraints (bandwidth) and user expectations (execution time). In case of multi-gateway deployments, we tackled the problem of unequal workload distribution. In particular, we propose a workload-aware management scheme that performs intra- and inter-gateway optimizations. The intra-gateway mechanism provides a balanced execution environment for the applications, and it achieves up to 95% performance deviation improvement, compared to un-optimized systems. The presented inter-gateway method manages to balance the workload among multiple gateways and is able to achieve a global performance threshold.

The main objective of this thesis is to propose a localization method for a UAV using various computer vision and machine learning techniques. It plays a major role in planning the strategy for the flight, and acts as a navigational contingency method, in event of a GPS failure. The implementation of the algorithms employs high processing capabilities of the graphics processing unit, making it more efficient. The method involves the working of various neural networks, working in synergy to perform the localization. This thesis is a part of a collaborative project between The University of North Texas, Denton, USA, and the University of Windsor, Ontario, Canada. The localization has been divided into three phases namely object detection, recognition, and location estimation. Object detection and position estimation were discussed in this thesis while giving a brief understanding of the recognition. Further, future strategies to aid the UAV to complete the mission, in case of an eventuality, like the introduction of an EDGE server and wireless charging methods, was also given a brief introduction.

Indiana University-Purdue University Indianapolis (IUPUI) / 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. 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.

Title: Converting Prose into Poetry with Neural Networks Author: Memduh Gokirmak Institute: Institute of Formal and Applied Linguistics Supervisor: Martin Popel, Institute of Formal and Applied Linguistics Abstract: We present here our attempts to create a system that generates poetry based on a sequence of text provided to it by a user. We explore the use of machine translation and language model technologies based on the neural network architecture. We use different types of data across three languages in our research, and employ and develop metrics to track the quality of the output of the systems we develop. We find that combining machine translation techniques to generate training data to this end with fine-tuning of pre-trained language models provides the most satisfactory generated poetry. Keywords: poetry machine translation language models iii

京都大学 / 0048 / 新制・課程博士 / 博士(情報学) / 甲第21909号 / 情博第692号 / 新制||情||119(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)教授 神谷 之康, 教授 熊田 孝恒, 教授 西野 恒 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DGAM

No description available.

With the development of small satellites it has become easier and cheaper to deploy satellites for earth observation from space. While optical sensors capture high-resolution data, this data is traditionally sent to earth for analysis which puts a high constraint on the data link and increases the time for making data based decisions. This thesis explores the possibilities of deploying an AI model in small satellites for detecting objects in satellite imagery and therefore reduce the amount of data that needs to be transmitted. The neural network model YOLOv8 was trained on the xView and DIOR dataset and evaluated in a hardware restricted execution environment. The model achieved a mAP50 of 0.66 and could process satellite images at a speed of 309m2/s.

Battery Electric Vehicles (BEV) are gaining market share but still must overcome several engineering challenges related to the lithium-ion battery packs powering them. The batteries must be carefully managed to optimize safety and performance. The estimation of battery states, which cannot be measured directly, is an important part of battery management and remains an active area of research since small gains in estimation accuracy can help reduce cost and increase BEV range. This thesis presents several improvements to battery state estimation using different methods. Electrochemical Impedance Spectroscopy (EIS) is receiving increased attention from researchers as a method for state estimation and diagnostics for real-time applications. Due to battery relaxation behaviour, long rest times are commonly used before performing the EIS measurement. In this work, methods were developed to significantly shorten the required rest times, and a State of Health (SoH) estimation strategy was proposed by taking advantage of the relaxation effect as measured by EIS. This method was demonstrated to have an estimation error of below 1%. At low temperatures, the accuracy of the battery model becomes poor due to the non-linear battery response to current. By using an adaptive filter called the Interacting Multiple Model (IMM) filter, the next part of this work showed how to significantly improve low temperature State of Charge (SoC) estimation. Further reduction in estimation errors was achieved by pairing the IMM with the Smooth Variable Structure Filter (SVSF), for SoC estimation errors below 2%. The work presented in this thesis also includes the application of Deep Neural Networks (DNN) for SoC estimation from EIS data. Finally, an extensive aging study was conducted and an accelerated protocol was compared to a realistic drive cycle based protocol using EIS as a characterization tool. / Thesis / Doctor of Philosophy (PhD) / Replacing conventional gasoline/diesel powered cars with battery powered vehicles is part of a solution to the climate crisis. However, the initial costs paired with range anxiety stops many from switching to electric cars. Both cost and range are related to the battery pack. To achieve the best possible range for the lowest possible cost, battery packs must be carefully controlled by sophisticated algorithms. Unfortunately, battery range or health cannot be measured directly, but must be inferred through measurable indicators. This thesis explores battery behavior under different operating conditions and develops improved methods which can be used to determine battery health and/or range. A powerful method usually used only in laboratory settings is studied and improved to make it more suitable for implementation in electric cars. In this work it is used for accurate battery health determination. Furthermore, a strategy for improving battery range determination at low temperatures is also proposed.

Atypical Femur Fracture(AFF) is a type of stress fracture that occurs in conjunction with prolonged bisphosphonate treatment. In practice, AFF is very rarely identified from Normal Femur Fracture(NFF) correctly on the first diagnostic X-ray examination. This project aims at developing an algorithm based on deep neural networks to assist clinicians with the diagnosis of atypical femurfracture. Two diagnostic pipelines were constructed using the Convolutional Neural Network (CNN) as the core classifier. One is a fully automatic pipeline, where the X-rays image is directly input into the network with only standardized pre-processing steps. Another interactive pipeline requires the user to re-orient the femur bones above the fractures to a vertical position and move the fracture line to the image center, before the repositioned image is sent to the CNNs. Three most popular CNNs architectures, namely VGG19, InceptionV3 and ResNet50,were tested for classifying the images to either AFF or NFF. Transfer learning technique was used to pre-train these networks using images form ImageNet. The diagnosis accuracy was evaluated using 5-fold cross-validation. With the fully automatic diagnosis pipeline, we achieved diagnosis accuracy of 82.7%, 89.4%, 90.5%, with VGG19, InceptionV3 and ResNet50, respectively. With the interactive diagnostic pipeline, the diagnosis accuracy was improved to 92.2%, 93.4% and 94.4%, respectively. To further validate the results, class activation mapping is used for indicating the discriminative image regions that the neural networks learn to identify a certain class.

The ability to extract useful information from texts and present them in the form of structured attributes is an important step towards making product comparison algorithm in fashion smarter and better. Some previous work exploits statistical features like the word frequency and graph models to predict keyphrases. In recent years, deep neural networks have proved to be the state-of-the-art methods to handle language modeling. Successful examples include Long Short Term Memory (LSTM), Gated Recurrent Units (GRU), Bidirectional Encoder Representations from Transformers(BERT) and their variations. In addition, some word embedding techniques like word2vec[1] are also helpful to improve the performance. Besides these techniques, a high-quality dataset is also important to the effectiveness of models. In this project, we aim to develop reliable and efficient machine learning models for keyphrase extraction. At Norna AB, we have a collection of product descriptions from different vendors without keyphrase annotations, which motivates the use of unsupervised methods. They should be capable of extracting useful keyphrases that capture the features of a product. To further explore the power of deep neural networks, we also implement several deep learning models. The dataset has two parts, the first part is called the fashion dataset where keyphrases are extracted by our unsupervised method. The second part is a public dataset in the domain of news. We find that deep learning models are also capable of extracting meaningful keyphrases and outperform the unsupervised model. Precision, recall and F1 score are used as evaluation metrics. The result shows that the model that uses LSTM and CRF achieves the optimal performance. We also compare the performance of different models with respect to keyphrase lengths and keyphrase numbers. The result indicates that all models perform better on predicting short keyphrases. We also show that our refined model has the advantage of predicting long keyphrases, which is challenging in this field. / Förmågan att extrahera användbar information från texter och presentera den i form av strukturerade attribut är ett viktigt steg mot att göra produktjämförelsesalgoritmen på ett smartare och bättre sätt. Vissa tidigare arbeten utnyttjar statistiska funktioner som ordfrekvens och grafmodeller för att förutsäga nyckelfraser. Under de senaste åren har djupa neurala nätverk visat sig vara de senaste metoderna för att hantera språkmodellering. Framgångsrika exempel inkluderar Long Short Term Memory (LSTM), Gated Recurrent Units (GRU), Bidirectional Encoder Representations from Transformers (BERT) och deras variationer. Dessutom kan vissa ordinbäddningstekniker som word2vec[1] också vara till hjälp för att förbättra prestandan. Förutom dessa tekniker är en datauppsättning av hög kvalitet också viktig för modellernas effektivitet. I detta projekt strävar vi efter att utveckla pålitliga och effektiva maskininlärningsmodeller för utvinning av nyckelfraser. På Norna AB har vi en samling produktbeskrivningar från olika leverantörer utan nyckelfrasnoteringar, vilket motiverar användningen av metoder utan tillsyn. De bör kunna extrahera användbara nyckelfraser som fångar funktionerna i en produkt. För att ytterligare utforska kraften i djupa neurala nätverk implementerar vi också flera modeller för djupinlärning. Datasetet har två delar, den första delen kallas modedataset där nyckelfraser extraheras med vår metod utan tillsyn. Den andra delen är en offentlig dataset i nyhetsdomänen. Vi finner att deep learning-modeller också kan extrahera meningsfulla nyckelfraser och överträffa den oövervakade modellen. Precision, återkallning och F1-poäng används som utvärderingsmått. Resultatet visar att modellen som använder LSTM och CRF uppnår optimal prestanda. Vi jämför också prestanda för olika modeller med avseende på keyphrase längder och nyckelfras nummer. Resultatet indikerar att alla modeller presterar bättre på att förutsäga korta tangentfraser. Vi visar också att vår raffinerade modell har fördelen att förutsäga långa tangentfraser, vilket är utmanande inom detta område.