Indiana University-Purdue University Indianapolis (IUPUI) / Every year the most effective Deep learning models, CNN architectures are showcased based on their compatibility and performance on the embedded edge hardware, especially for applications like image classification. These deep learning models necessitate a significant amount of computation and memory, so they can only be used on high-performance computing systems like CPUs or GPUs. However, they often struggle to fulfill portable specifications due to resource, energy, and real-time constraints. Hardware accelerators have recently been designed to provide the computational resources that AI and machine learning tools need. These edge accelerators have high-performance hardware which helps maintain the precision needed to accomplish this mission. Furthermore, this classification dilemma that investigates channel interdependencies using either depth-wise or group-wise convolutional features, has benefited from the inclusion of Bottleneck modules. Because of its increasing use in portable applications, the classic inverted residual block, a well-known architecture technique, has gotten more recognition. This work takes it a step forward by introducing a design method for porting CNNs to lowresource embedded systems, essentially bridging the difference between deep learning models and embedded edge systems. To achieve these goals, we use closer computing strategies to reduce the computer’s computational load and memory usage while retaining excellent deployment efficiency. This thesis work introduces HBONext, a mutated version of Harmonious Bottlenecks (DHbneck) combined with a Flipped version of Inverted Residual (FIR), which outperforms the current HBONet architecture in terms of accuracy and model size miniaturization. Unlike the current definition of inverted residual, this FIR block performs identity mapping and spatial transformation at its higher dimensions. The HBO solution, on the other hand, focuses on two orthogonal dimensions: spatial (H/W) contraction-expansion and later channel (C) expansion-contraction, which are both organized in a bilaterally symmetric manner. HBONext is one of those versions that was designed specifically for embedded and mobile applications. In this research work, we also show how to use NXP Bluebox 2.0 to build a real-time HBONext image classifier. The integration of the model into this hardware has been a big hit owing to the limited model size of 3 MB. The model was trained and validated using CIFAR10 dataset, which performed exceptionally well due to its smaller size and higher accuracy. The validation accuracy of the baseline HBONet architecture is 80.97%, and the model is 22 MB in size. The proposed architecture HBONext variants, on the other hand, gave a higher validation accuracy of 89.70% and a model size of 3.00 MB measured using the number of parameters. The performance metrics of HBONext architecture and its various variants are compared in the following chapters.

Indiana University-Purdue University Indianapolis (IUPUI) / Computer vision is a domain which deals with the goal of making technology as efficient as human vision. To achieve that goal, after decades of research, researchers have developed algorithms that are able to work efficiently on resource constrained hardware like mobile or embedded devices for computer vision applications. Due to their constant efforts, such devices have become capable for tasks like Image Classification, Object Detection, Object Recognition, Semantic Segmentation, and many other applications. Autonomous systems like self-driving cars, Drones and UAVs, are being successfully developed because of these advances in AI. Deep Learning, a part of AI, is a specific domain of Machine Learning which focuses on developing algorithms for such applications. Deep Learning deals with tasks like extracting features from raw image data, replacing pipelines of specialized models with single end-to-end models, making models usable for multiple tasks with superior performance. A major focus is on techniques to detect and extract features which provide better context for inference about an image or video stream. A deep hierarchy of rich features can be learned and automatically extracted from images, provided by the multiple deep layers of CNN models. CNNs are the backbone of Computer Vision. The reason that CNNs are the focus of attention for deep learning models is that they were specifically designed for image data. They are complicated but very effective in extracting features from an image or a video stream. After AlexNet won the ILSVRC in 2012, there was a drastic increase in research related with CNNs. Many state-of-the-art architectures like VGG Net, GoogleNet, ResNet, Inception-v4, Inception-Resnet-v2, ShuffleNet, Xception, MobileNet, MobileNetV2, SqueezeNet, SqueezeNext and many more were introduced. The trend behind the research depicts an increase in the number of layers of CNN to make them more efficient but with that, the size of the model increased as well. This problem was fixed with the advent of new algorithms which resulted in a decrease in model size. As a result, today we have CNN models, which are implemented on mobile devices. These mobile models are compact and have low latency, which in turn reduces the computational cost of the embedded system. This thesis resembles similar idea, it proposes two new CNN architectures, A-MnasNet and R-MnasNet, which have been derived from MnasNet by Design Space Exploration. These architectures outperform MnasNet in terms of model size and accuracy. They have been trained and tested on CIFAR-10 dataset. Furthermore, they were implemented on NXP Bluebox 2.0, an autonomous driving platform, for Image Classification.

Indiana University-Purdue University Indianapolis (IUPUI) / Computer vision is a domain which deals with the goal of making technology as efficient as human vision. To achieve that goal, after decades of research, researchers have developed algorithms that are able to work efficiently on resource constrained hardware like mobile or embedded devices for computer vision applications. Due to their constant efforts, such devices have become capable for tasks like Image Classification, Object Detection, Object Recognition, Semantic Segmentation, and many other applications. Autonomous systems like self-driving cars, Drones and UAVs, are being successfully developed because of these advances in AI. Deep Learning, a part of AI, is a specific domain of Machine Learning which focuses on developing algorithms for such applications. Deep Learning deals with tasks like extracting features from raw image data, replacing pipelines of specialized models with single end-to-end models, making models usable for multiple tasks with superior performance. A major focus is on techniques to detect and extract features which provide better context for inference about an image or video stream. A deep hierarchy of rich features can be learned and automatically extracted from images, provided by the multiple deep layers of CNN models. CNNs are the backbone of Computer Vision. The reason that CNNs are the focus of attention for deep learning models is that they were specifically designed for image data. They are complicated but very effective in extracting features from an image or a video stream. After AlexNet won the ILSVRC in 2012, there was a drastic increase in research related with CNNs. Many state-of-the-art architectures like VGG Net, GoogleNet, ResNet, Inception-v4, Inception-Resnet-v2, ShuffleNet, Xception, MobileNet, MobileNetV2, SqueezeNet, SqueezeNext and many more were introduced. The trend behind the research depicts an increase in the number of layers of CNN to make them more efficient but with that, the size of the model increased as well. This problem was fixed with the advent of new algorithms which resulted in a decrease in model size. As a result, today we have CNN models, which are implemented on mobile devices. These mobile models are compact and have low latency, which in turn reduces the computational cost of the embedded system. This thesis resembles similar idea, it proposes two new CNN architectures, A-MnasNet and R-MnasNet, which have been derived from MnasNet by Design Space Exploration. These architectures outperform MnasNet in terms of model size and accuracy. They have been trained and tested on CIFAR-10 dataset. Furthermore, they were implemented on NXP Bluebox 2.0, an autonomous driving platform, for Image Classification.

No description available.

京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第18561号 / 工博第3922号 / 新制||工||1603(附属図書館) / 31461 / 京都大学大学院工学研究科社会基盤工学専攻 / (主査)教授 田村 正行, 准教授 須﨑 純一, 准教授 田中 賢治 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

No description available.

This thesis examines adversarial examples in machine learning, specifically in the im- age classification domain. State-of-the-art deep learning models are able to recognize patterns better than humans. However, we can significantly reduce the model's accu- racy by adding imperceptible, yet intentionally harmful noise. This work investigates various methods of creating adversarial images as well as techniques that aim to defend deep learning models against these malicious inputs. We choose one of the contemporary defenses and design an attack that utilizes evolutionary algorithms to deceive it. Our experiments show an interesting difference between adversarial images created by evolu- tion and images created with the knowledge of gradients. Last but not least, we test the transferability of our created samples between various deep learning models. 1

No description available.

In this study, we systematically investigate the impact of class imbalance on classification performance of convolutional neural networks and compare frequently used methods to address the issue. Class imbalance refers to significantly different number of examples among classes in a training set. It is a common problem that has been comprehensively studied in classical machine learning, yet very limited systematic research is available in the context of deep learning. We define and parameterize two representative types of imbalance, i.e. step and linear. Using three benchmark datasets of increasing complexity, MNIST, CIFAR-10 and ImageNet, we investigate the effects of imbalance on classification and perform an extensive comparison of several methods to address the issue: oversampling, undersampling, two-phase training, and thresholding that compensates for prior class probabilities. Our main evaluation metric is area under the receiver operating characteristic curve (ROC AUC) adjusted to multi-class tasks since overall accuracy metric is associated with notable difficulties in the context of imbalanced data. Based on results from our experiments we conclude that (i) the effect of class imbalance on classification performance is detrimental and increases with the extent of imbalance and the scale of a task; (ii) the method of addressing class imbalance that emerged as dominant in almost all analyzed scenarios was oversampling; (iii) oversampling should be applied to the level that totally eliminates the imbalance, whereas undersampling can perform better when the imbalance is only removed to some extent; (iv) thresholding should be applied to compensate for prior class probabilities when overall number of properly classified cases is of interest; (v) as opposed to some classical machine learning models, oversampling does not necessarily cause overfitting of convolutional neural networks. / I den här studien undersöker vi systematiskt effekten av klassobalans på prestandan för klassificering hos konvolutionsnätverk och jämför vanliga metoder för att åtgärda problemet. Klassobalans avser betydlig ojämvikt hos antalet exempel per klass i ett träningsset. Det är ett vanligt problem som har studerats utförligt inom maskininlärning, men tillgången av systematisk forskning inom djupinlärning är starkt begränsad. Vi definerar och parametriserar två representiva typer av obalans, steg och linjär. Med hjälpav tre dataset med ökande komplexitet, MNIST, CTFAR-10 och ImageNet, undersöker vi effekterna av obalans på klassificering och utför en omfattande jämförelse av flera metoder för att åtgärda problemen: översampling, undersampling, tvåfasträning och avgränsning för tidigare klass-sannolikheter. Vår huvudsakliga utvärderingsmetod är arean under mottagarens karaktäristiska kurva (ROC AUC) justerat för multi-klass-syften, eftersom den övergripande noggrannheten är förenad med anmärkningsvärda svårigheter i samband med obalanserade data. Baserat på experimentens resultat drar vi slutsatserna att (i) effekten av klassens obalans påklassificeringprestanda är skadlig och ökar med mängden obalans och omfattningen av uppgiften; (ii) metoden att ta itu med klassobalans som framträdde som dominant i nästan samtliga analyserade scenarier var översampling; (iii) översampling bör tillämpas till den nivå som helt eliminerar obalansen, medan undersampling kan prestera bättre när obalansen bara avlägsnas i en viss utsträckning; (iv) avgränsning bör tillämpas för att kompensera för tidigare sannolikheter när det totala antalet korrekt klassificerade fall är av intresse; (v) i motsats till hos vissa klassiska maskininlärningsmodeller orsakar översampling inte nödvändigtvis överanpassning av konvolutionsnätverk.

Ecosystems respond to stress factors that may have a natural or anthropogenic origin. Natural stress factors include flood, wildfire, drought, insect infestations, etc. and anthropogenic stress factors include pollution, land cover changes, and the introduction of alien invasive species. These stressors can degrade ecosystems and result in biodiversity loss and lowered resilience. In this thesis, I investigate the spatial and temporal dynamics of ecosystem stress caused by natural and anthropogenic factors in both aquatic and terrestrial ecosystems. The large study areas and long-term changes in my research have mandated the use of Remote Sensing (RS) and Geographic Information Systems (GIS) techniques in ways that have not been previously considered in ecological studies. In the first two chapters, I developed new approaches to monitor Phragmites australis, one of the most aggressive alien plant species that has invaded wetland ecosystems throughout N. America, as well as roadside ditches where management is costly and logistically challenging. I have developed innovative methods to accurately map invasive Phragmites under two conditions: 1) when plant biomass and densities are high so that managers can evaluate the effectiveness of treatment methods and 2) when plant biomass and densities are small and sparse so that these stands can be quantified and eradicated. I found that freely available, low to moderate resolution satellite imagery (Landsat 7/8 and Sentinel 2), acquired in late July or early August, can be used to produce highly accurate maps of dense Phragmites populations. I also found that commercial satellite imagery (WorldView 2/3) can be used to map Phragmites in the early stages of invasion and when plants have regenerated following herbicide treatment. In the latter half of my thesis, I examined how pre-fire canopy species composition and forest health influence the response of boreal forests to wildfires in Alberta, Canada. Forest fires occur naturally in boreal forests and usually affect very large spatial extents that remove accumulated fire fuel from the system. Following these outbreaks, the forests will regenerate and eventually become restored to their initial state. Climate-change induced droughts and flooding may change the frequency and location of these forest fires. To quantify the burn severity of each fire, I used Landsat images to calculate the differenced Normalized Burn Ratio (dNBR); then combined dNBR for all affected areas to develop the Standardized Burn Impact Score (SBIS), which quantifies the average impact of each fire based on the size of the burned area and the mean burn severity per pixel. In general, pre-fire dominance of coniferous species (jack pine and spruce) led to higher SBIS values while pre-fire dominance of broad-leaved species (aspen, birch, and poplar) led to lower values. Mean burn severity and SBIS values increased significantly when fire outbreaks occurred at a distance of 1 km or greater from water features (e.g. lakes, rivers, streams, wetlands). I also investigated the post-fire recovery process using indices of vegetation health and accounting for the effect of distance from the water features with respect to different levels of human activity. My results show that the post-fire recovery patterns are altered due to human activities and can affect the long-term fire regimes in boreal forests of northern Alberta. Overall, my thesis has advanced the use of novel remote-sensing techniques to study ecosystem stress factors on wetland and boreal ecosystems in Canada. / Thesis / Doctor of Philosophy (PhD) / Ecosystem stress is caused by natural or anthropogenic factors and results degradation of ecosystems. I investigated the spatial and temporal dynamics of ecosystem stress on aquatic and terrestrial ecosystems using Remote Sensing and Geographic Information Systems techniques. I mapped Phragmites australis, a notorious invasive grass, in wetlands to aid the Phragmites management programs. My research shows that images collected in late summer or fall provide high Phragmites mapping accuracy. Furthermore, I successfully mapped small, low-density Phragmites stands in the early stages of invasions. I also investigated the pre-and post-fire vegetation dynamics in the boreal forests of Alberta. I show that the species composition and water features influence the burn severity. The human influence on these ecosystems alters the natural post-fire vegetation recovery processes. Overall, my thesis advances the use of novel remote-sensing techniques to investigate the ecosystem stress factors on wetland and boreal ecosystems in Canada.