Indiana University-Purdue University Indianapolis (IUPUI) / In recent years, deep learning models have become popular in the real-time embedded application, but there are many complexities for hardware deployment because of limited resources such as memory, computational power, and energy. Recent research in the field of deep learning focuses on reducing the model size of the Convolution Neural Network (CNN) by various compression techniques like Architectural compression, Pruning, Quantization, and Encoding (e.g., Huffman encoding). Network pruning is one of the promising technique to solve these problems. This thesis proposes methods to prune the convolution neural network (SqueezeNet) without introducing network sparsity in the pruned model. This thesis proposes three methods to prune the CNN to decrease the model size of CNN without a significant drop in the accuracy of the model. 1: Pruning based on Taylor expansion of change in cost function Delta C. 2: Pruning based on L2 normalization of activation maps. 3: Pruning based on a combination of method 1 and method 2. The proposed methods use various ranking methods to rank the convolution kernels and prune the lower ranked filters afterwards SqueezeNet model is fine-tuned by backpropagation. Transfer learning technique is used to train the SqueezeNet on the CIFAR-10 dataset. Results show that the proposed approach reduces the SqueezeNet model by 72% without a significant drop in the accuracy of the model (optimal pruning efficiency result). Results also show that Pruning based on a combination of Taylor expansion of the cost function and L2 normalization of activation maps achieves better pruning efficiency compared to other individual pruning criteria and most of the pruned kernels are from mid and high-level layers. The Pruned model is deployed on BlueBox 2.0 using RTMaps software and model performance was evaluated.

京都大学 / 0048 / 新制・課程博士 / 博士(情報学) / 甲第21210号 / 情博第663号 / 新制||情||114(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 黒橋 禎夫, 教授 河原 達也, 教授 森 信介 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM

No description available.

<p>Domain adaptation is one of the hottest directions in solving annotation insufficiency problem of deep learning. General domain adaptation is not consistent with the practical scenarios in the industry. In this thesis, we focus on two concerns as below.</p> <p>  </p> <p>  First is that labeled data are generally collected from multiple domains. In other words, multi-source adaptation is a more common situation. Simply extending these single-source approaches to the multi-source cases could cause sub-optimal inference, so specialized multi-source adaptation methods are essential. The main challenge in the multi-source scenario is a more complex divergence situation. Not only the divergence between target and each source plays a role, but the divergences among distinct sources matter as well. However, the significance of maintaining consistency among multiple sources didn't gain enough attention in previous work. In this thesis, we propose an Enhanced Consistency Multi-Source Adaptation (EC-MSA) framework to address it from three perspectives. First, we mitigate feature-level discrepancy by cross-domain conditional alignment, narrowing the divergence between each source and target domain class-wisely. Second, we enhance multi-source consistency via dual mix-up, diminishing the disagreements among different sources. Third, we deploy a target distilling mechanism to handle the uncertainty of target prediction, aiming to provide high-quality pseudo-labeled target samples to benefit the previous two aspects. Extensive experiments are conducted on several common benchmark datasets and demonstrate that our model outperforms the state-of-the-art methods.</p> <p>  </p> <p>  Second is that data privacy and security is necessary in practice. That is, we hope to keep the raw data stored locally while can still obtain a satisfied model. In such a case, the risk of data leakage greatly decreases. Therefore, it is natural for us to combine the federated learning paradigm with domain adaptation. Under the source-private setting, the main challenge for us is to expose information from the source domain to the target domain while make sure that the communication process is safe enough. In this thesis, we propose a method named Fourier Transform-Assisted Federated Domain Adaptation (FTA-FDA) to alleviate the difficulties in two ways. We apply Fast Fourier Transform to the raw data and transfer only the amplitude spectra during the communication. Then frequency space interpolations between these two domains are conducted, minimizing the discrepancies while ensuring the contact of them and keeping raw data safe. What's more, we make prototype alignments by using the model weights together with target features, trying to reduce the discrepancy in the class level. Experiments on Office-31 demonstrate the effectiveness and competitiveness of our approach, and further analyses prove that our algorithm can help protect privacy and security.</p>

The automotive industry is becoming more dependent on sustainable and efficient systems within vehicles. With the diverse combination of conditions affecting vehicle performance, such as environmental conditions and drivers' behaviour, the interest in monitoring machine health increases. This master thesis examines the machine learning approach to sensor reconstruction of hauler engine sensors for deviation detection applications across multiple domains. A novel proposal for sequence learning and modelling was by introducing a weighted difference of sequence derivatives. Impacts of including differences of derivatives assisted the learning capabilities of sequential data for the majority of the target sensors across multiple operating domains. Robust sensor reconstruction was also examined by using inductive transfer learning with a Long Short-Term Memory-Domain Adversarial Neural Network. Obtained results implied an improvement in using the Long Short-Term Memory-Domain Adversarial Neural Network, then using a regular Long Short-Term Memory network trained on both source and target domains. Suggested methods were evaluated towards model-based performance and computational limitations. The combined aspects of model performance and system performance are used to discuss the trade-offs using each proposed method.

Data augmentation is a technique that acquires more training data by augmenting available samples, where the training data is used to fit model parameters. Data augmentation is utilized due to a shortage of training data in certain domains and to reduce overfitting. Augmenting a training dataset for image classification with a Generative Adversarial Network (GAN) has been shown to increase classification accuracy. This report investigates if transfer learning within a GAN can further increase classification accuracy when utilizing the augmented training dataset. The method section describes a specific GAN architecture for the experiments that includes a label condition. When using transfer learning within the specific GAN architecture, a statistical analysis shows a statistically significant increase in classification accuracy for a classification problem with the EMNIST dataset, which consists of images of handwritten alphanumeric characters. In the discussion section, the authors analyze the results and motivates other use cases for the proposed GAN architecture. / Datautökning är en metod som skapar mer träningsdata genom att utöka befintlig träningsdata, där träningsdatan används för att anpassa modellers parametrar. Datautökning används på grund av en brist på träningsdata inom vissa områden samt för att minska overfitting. Att utöka ett träningsdataset för att genomföra bildklassificering med ett generativt adversarialt nätverk (GAN) har visats kunna öka precisionen av klassificering av bilder. Denna rapport undersöker om transferlärande inom en GAN kan vidare öka klassificeringsprecisionen när ett utökat träningsdataset används. Metoden beskriver en specific GANarkitektur som innehåller ett etikettvillkor. När transferlärande används inom den utvalda GAN-arkitekturen visar en statistisk analys en statistiskt säkerställd ökning av klassificeringsprecisionen för ett klassificeringsproblem med EMNIST datasetet, som innehåller bilder på handskrivna bokstäver och siffror. I diskussionen diskuteras orsakerna bakom resultaten och fler användningsområden nämns.

Breast cancer is one of the most common cancers among women in the world, and the average error rate among radiologists during diagnosis is 30%. Computer-aided medical diagnosis aims to assist doctors by giving them a second opinion, thus decreasing the error rate. Convolutional neural networks (CNNs) have shown to be good for visual detection and recognition tasks, and have been explored in combination with transfer learning. However, the performance of a deep learning model does not only rely on the model itself, but on the nature of the dataset as well In breast cancer diagnosis, the area surrounding a mass provides useful context for diagnosis. In this study, we explore providing different amounts of context to the CNN model ResNet50, to see how it affects the model’s performance. We test masses with no additional context, twice the amount of original context and four times the amount of original context, using 10-fold cross-validation with ROC AUC and average precision (AP ) as our metrics. The results suggest that providing additional context does improve the model’s performance. However, giving two and four times the amount of context seems to give similar performance. / Bröstcancer är en av de vanligaste cancersjukdomar bland kvinnor i världen, och den genomsnittliga felfrekvensen under diagnoser är 30%. Datorstödd medicinsk diagnos syftar till att hjälpa läkare genom att ge dem en andra åsikt, vilket minskar felfrekvensen. Konvolutionella neurala nätverk (CNNs) har visat sig vara bra för visuell detektering och igenkännande, och har utforskats i samband med det s.k. “transfer learning”. Prestationen av en djup inlärningsmodell är däremot inte enbart beroende på modellen utan också på datasetets natur. I bröstcancerdiagnos ger området runt en bröstmassa användbar kontext för diagnos. I den här studien testar vi att ge olika mängder kontext till CNNmodellen ResNet50, för att se hur det påverkar modellens prestanda. Vi testar bröstmassor utan ytterligare kontext, dubbelt så mycket som den originala mängden kontext och fyra gånger så mycket som den orginala mängden kontext, med hjälp av “10-fold cross-validation” med ROC AUC och “average precision” (AP ) som våra mätvärden. Resultaten visar att mer kontext förbättrar modellens prestanda. Däremot verkar att ge två och fyra gånger så mycket kontext resultera i liknande prestanda.

Antenna tilt is the angle subtended by the radiation beam and horizontal plane. This angle plays a vital role in determining the coverage and the interference of the network with neighbouring cells and adjacent base stations. Traditional methods for network optimization rely on rule-based heuristics to do decision making for antenna tilt optimization to achieve desired network characteristics. However, these methods are quite brittle and are incapable of capturing the dynamics of communication traffic. Recent advancements in reinforcement learning have made it a viable solution to overcome this problem but even this learning approach is either limited to its simulation environment or is limited to off-policy offline learning. So far, there has not been any effort to overcome the previously mentioned limitations, so as to make it applicable in the real world. This work proposes a method that consists of transferring reinforcement learning policies from a simulated environment to a real environment i.e. sim-to-real transfer through the use of offline learning. The approach makes use of a simulated environment and a fixed dataset to compensate for the underlined limitations. The proposed sim-to-real transfer technique utilizes a hybrid policy model, which is composed of a portion trained in simulation and a portion trained on the offline real-world data from the cellular networks. This enables to merge samples from the real-world data to the simulated environment consequently modifying the standard reinforcement learning training procedures through knowledge sharing between the two environment’s representations. On the one hand, simulation enables to achieve better generalization performance with respect to conventional offline learning as it complements offline learning with learning through unseen simulated trajectories. On the other hand, the offline learning procedure enables to close the sim-to-real gap by exposing the agent to real-world data samples. Consequently, this transfer learning regime enable us to establish optimal antenna tilt control which in turn results in improved coverage and reduced interference with neighbouring cells in the cellular network. / Antennlutning är den vinkel som dämpas av strålningsstrålen och det horisontella planet. Denna vinkel spelar en viktig roll för att bestämma täckningen och störningen av nätverket med angränsande celler och intilliggande basstationer. Traditionella metoder för nätverksoptimering förlitar sig på regelbaserad heuristik för att göra beslutsfattande för antennlutningsoptimering för att uppnå önskade nätverksegenskaper. Dessa metoder är dock ganska styva och är oförmögna att fånga dynamiken i kommunikationstrafiken. De senaste framstegen inom förstärkningsinlärning har gjort det till en lönsam lösning att lösa detta problem, men även denna inlärningsmetod är antingen begränsad till dess simuleringsmiljö eller är begränsad till off-policy offline inlärning. Hittills har inga ansträngningar gjorts för att övervinna de tidigare nämnda begränsningarna för att göra det tillämpligt i den verkliga världen. Detta arbete föreslår en metod som består i att överföra förstärkningsinlärningspolicyer från en simulerad miljö till en verklig miljö, dvs. sim-till-verklig överföring genom användning av offline-lärande. Metoden använder en simulerad miljö och en fast dataset för att kompensera för de understrukna begränsningarna. Den föreslagna sim-till-verkliga överföringstekniken använder en hybridpolicymodell, som består av en del utbildad i simulering och en del utbildad på offline-verkliga data från mobilnätverk. Detta gör det möjligt att slå samman prover från verklig data till den simulerade miljön och därmed modifiera standardutbildningsförfarandena för förstärkning genom kunskapsdelning mellan de två miljöernas representationer. Å ena sidan möjliggör simulering att uppnå bättre generaliseringsprestanda med avseende på konventionellt offlineinlärning eftersom det kompletterar offlineinlärning med inlärning genom osynliga simulerade banor. Å andra sidan möjliggör offline-inlärningsförfarandet att stänga sim-till-real-klyftan genom att exponera agenten för verkliga dataprov. Följaktligen möjliggör detta överföringsinlärningsregime att upprätta optimal antennlutningskontroll som i sin tur resulterar i förbättrad täckning och minskad störning med angränsande celler i mobilnätet.

Anomaly detection in high-dimensional financial transaction data is challenging and resource-intensive, particularly when the dataset is unlabeled. Sometimes, one can alleviate the computational cost and improve the results by utilizing a pre-trained model, provided that the features learned from the pre-training are useful for learning the second task. Investigating this issue was the main purpose of this thesis. More specifically, it was to explore the potential gain of pre-training a detection model on one trader's transaction history and then retraining the model to detect anomalous trades in another trader's transaction history. In the context of transfer learning, the pre-trained and the retrained model are usually referred to as the source model and target model, respectively.  A deep LSTM autoencoder was proposed as the source model due to its advantages when dealing with sequential data, such as financial transaction data. Moreover, to test its anomaly detection ability despite the lack of labeled true anomalies, synthetic anomalies were generated and included in the test set. Various experiments confirmed that the source model learned to detect synthetic anomalies with highly distinctive features. Nevertheless, it is hard to draw any conclusions regarding its anomaly detection performance due to the lack of labeled true anomalies. While the same is true for the target model, it is still possible to achieve the thesis's primary goal by comparing a pre-trained model with an identical untrained model. All in all, the results suggest that transfer learning offers a significant advantage over traditional machine learning in this context.

<p>Deep neural networks have made significant advances in computer vision. However, several challenges limit their real-world applications. For example, domain shifts in vision data degrade model performance; visual appearance variances affect model robustness; it is also non-trivial to extend a model trained on one task to novel tasks; and in many applications, large-scale labeled data are not even available for learning powerful deep models from scratch. This research focuses on improving the transferability of deep features and the efficiency of deep vision model adaptation, leading to enhanced generalization and new capabilities on computer vision tasks. Specifically, we approach these problems from the following two directions: architectural adaptation and label-efficient transferable feature learning. From an architectural perspective, we investigate various schemes that permit network adaptation to be parametrized by multiple copies of sub-structures, distributions of parameter subspaces, or functions that infer parameters from data. We also explore how model adaptation can bring new capabilities, such as continuous and stochastic image modeling, fast transfer to new tasks, and dynamic computation allocation based on sample complexity. From the perspective of feature learning, we show how transferable features emerge from generative modeling with massive unlabeled or weakly labeled data. Such features enable both image generation under complex conditions and downstream applications like image recognition and segmentation. By combining both perspectives, we achieve improved performance on computer vision tasks with limited labeled data, enhanced transferability of deep features, and novel capabilities beyond standard deep learning models.</p>