Yes / Optimization plays a pivotal role in computer graphics and vision. Learning-based optimization algorithms have emerged as a powerful optimization technique for solving problems with robustness and accuracy because it learns gradients from data without calculating the Jacobian and Hessian matrices. The key aspect of the algorithms is the least-squares method, which formulates a general parametrized model of unconstrained optimizations and makes a residual vector approach to zeros to approximate a solution. The method may suffer from undesirable local optima for many applications, especially for point cloud registration, where each element of transformation vectors has a different impact on registration. In this paper, Reweighted Discriminative Optimization (RDO) method is proposed. By assigning different weights to components of the parameter vector, RDO explores the impact of each component and the asymmetrical contributions of the components on fitting results. The weights of parameter vectors are adjusted according to the characteristics of the mean square error of fitting results over the parameter vector space at per iteration. Theoretical analysis for the convergence of RDO is provided, and the benefits of RDO are demonstrated with tasks of 3D point cloud registrations and multi-views stitching. The experimental results show that RDO outperforms state-of-the-art registration methods in terms of accuracy and robustness to perturbations and achieves further improvement than non-weighting learning-based optimization.

<p dir="ltr">Traditional methods of behavior classification on videos of mice often rely on manually annotated datasets, which can be labor-intensive and resource-demanding to create. This research aims to address the challenges of behavior classification in mouse studies by leveraging an algorithmic framework employing self-supervised learning techniques capable of analyzing unlabeled datasets. This research seeks to develop a novel approach that eliminates the need for extensive manual annotation, making behavioral analysis more accessible and cost-effective for researchers, especially those in laboratories with limited access to annotated datasets.</p>

This thesis explores adapting self-supervised representation learning to visual domains beyond natural scenes, focusing on medical imaging. The research addresses the central question: “How can self-supervised representation learning be specifically adapted for detecting liver cancer in histopathology images?” The study utilizes the PAIP 2019 dataset for liver cancer segmentation and employs a self-supervised approach based on the VICReg method. The evaluation results demonstrated that the ImageNet-pretrained model achieved superior performance on the test set, with a clipped Jaccard index of 0.7747 at a threshold of 0.65. The VICReg-pretrained model followed closely with a score of 0.7461, while the model initialized with random weights trailed behind at 0.5420. These findings indicate that while ImageNet-pretrained models outperformed VICReg-pretrained models, the latter still captured essential data characteristics, suggesting the potential of self-supervised learning in diverse visual domains. The research attempts to contribute to advancing self-supervised learning in non-natural scenes and provides insights into model pretraining strategies.

This thesis proposes a novel approach for efficiently estimating gaze points in dual camera transportation datasets. Traditional methods for gaze point estimation are dependent on large amounts of labeled data, which can be both expensive and time-consuming to collect. Additionally, alignment and calibration of the two camera views present significant challenges. To overcome these limitations, this thesis investigates the use of self-learning techniques such as semi-supervised learning and self-training, which can reduce the need for labeled data while maintaining high accuracy. The proposed method is evaluated on the DGAZE dataset and achieves a 57.2\% improvement in performance compared to the previous methods. This approach can prove to be a valuable tool for studying visual attention in transportation research, leading to more cost-effective and efficient research in this field. / Master of Science / This thesis presents a new method for efficiently estimating the gaze point of drivers while driving, which is crucial for understanding driver behavior and improving transportation safety. Traditional methods require a lot of labeled data, which can be time-consuming and expensive to obtain. This thesis proposes a self-learning approach that can learn from both labeled and unlabeled data, reducing the need for labeled data while maintaining high accuracy. By training the model on labeled data and using its own estimations on unlabeled data to improve its performance, the proposed approach can adapt to new scenarios and improve its accuracy over time. The proposed method is evaluated on the DGAZE dataset and achieves a 57.2\% improvement in performance compared to the previous methods. Overall, this approach offers a more efficient and cost-effective solution that can potentially help improve transportation safety by providing a better understanding of driver behavior. This approach can prove to be a valuable tool for studying visual attention in transportation research, leading to more cost-effective and efficient research in this field.

Recent adaptations of machine learning models in many businesses has underscored the need for quality training data. Typically, training supervised machine learning systems involves using large amounts of human-annotated data. Labeling data is expensive and can be a limiting factor in using machine learning models. To enable continued integration of machine learning systems in businesses and also easy access by users, researchers have proposed several alternatives to supervised learning. Weak supervision is one such alternative. Weak supervision or weakly supervised learning involves using noisy labels (weak signals of the data) from multiple sources to train machine learning systems. A weak supervision model aggregates multiple noisy label sources called weak signals in order to produce probabilistic labels for the data. The main allure of weak supervision is that it provides a cheap yet effective substitute for supervised learning without need for labeled data. The key challenge in training weakly supervised machine learning models is that the weak supervision leaves ambiguity about the possible true labelings of the data. In this dissertation, we aim to address the challenge associated with training weakly supervised learning models by developing new weak supervision methods. Our work focuses on learning with constraint-based weak supervision algorithms. Firstly, we will propose an adversarial labeling approach for weak supervision. In this method, the adversary chooses the labels for the data and a model learns by minimising the error made by the adversarial model. Secondly, we will propose a simple constrained based approach that minimises a quadratic objective function in order to solve for the labels of the data. Next we explain the notion of data consistency for weak supervision and propose a data consistent method for weakly supervised learning. This approach combines weak supervision labels with features of the training data to make the learned labels consistent with the data. Lastly, we use this data consistent approach to propose a general approach for improving the performance of weak supervision models. In this method, we combine weak supervision with active learning in order to generate a model that outperforms each individual approach using only a handful of labeled data. For each algorithm we propose, we report extensive empirical validation of it by testing it on standard text and image classification datasets. We compare each approach against baseline and state-of-the-art methods and show that in most cases we match or outperform the methods we compare against. We report significant gains of our method on both binary and multi-class classification tasks. / Doctor of Philosophy / Machine learning models learn to make predictions from data. In supervised learning, a machine learning model is fed data and corresponding labels for the data so that the model can learn to predict labels for new unseen test data. Curation of large fully supervised datasets is expensive and time consuming since it involves subject matter experts providing labels for each individual data example. The cost of collecting labels has become one of the major roadblocks for training machine learning models. An alternative to supervised training of machine learning models is weak supervision. Weak supervision or weakly supervised learning trains with cheap, and easy to define signals that noisily label the data. We refer to these signals as weak signals. A weak supervision model combines various weak signals to produce training labels for the data. The key challenge in weak supervision is how to combine the different weak signals while navigating misleading correlations in their errors. In this dissertation, we propose several algorithms for weakly supervised learning. We classify our methods as constraint-based weak supervision since weak supervision is provided as constraints to our algorithms. We use experiments on different text and image classification datasets to show that our methods are effective and outperform competing methods that we compare against. Lastly, we propose a general framework for improving the performance of weak supervision models by incorporating a few labeled data. With this method we are able to close the gap to supervised learning without the need for labeling all the data examples.

One aim in the area of artificial intelligence (AI) is to develop a smart agent with high intelligence that is able to perceive and understand the complex visual environment around us. More ambitiously, it should be able to interact with us about its surroundings in natural languages. Thanks to the progress made in deep learning, we've seen huge breakthroughs towards this goal over the last few years. The developments have been extremely rapid in visual recognition, in which machines now can categorize images into multiple classes, and detect various objects within an image, with an ability that is competitive with or even surpasses that of humans. Meanwhile, we also have witnessed similar strides in natural language processing (NLP). It is quite often for us to see that now computers are able to almost perfectly do text classification, machine translation, etc. However, despite much inspiring progress, most of the achievements made are still within one domain, not handling inter-domain situations. The interaction between the visual and textual areas is still quite limited, although there has been progress in image captioning, visual question answering, etc. In this dissertation, we design models and algorithms that enable us to build in-depth connections between images and natural languages, which help us to better understand their inner structures. In particular, first we study how to make machines generate image descriptions that are indistinguishable from ones expressed by humans, which as a result also achieved better quantitative evaluation performance. Second, we devise a novel algorithm for measuring review congruence, which takes an image and review text as input and quantifies the relevance of each sentence to the image. The whole model is trained without any supervised ground truth labels. Finally, we propose a brand new AI task called Image Aspect Mining, to detect visual aspects in images and identify aspect level rating within the review context. On the theoretical side, this research contributes to multiple research areas in Computer Vision (CV), Natural Language Processing (NLP), interactions between CVandNLP, and Deep Learning. Regarding impact, these techniques will benefit related users such as the visually impaired, customers reading reviews, merchants, and AI researchers in general. / Doctor of Philosophy / One aim in the area of artificial intelligence (AI) is to develop a smart agent with high intelligence that is able to perceive and understand the complex visual environment around us. More ambitiously, it should be able to interact with us about its surroundings in natural languages. Thanks to the progress made in deep learning, we’ve seen huge breakthroughs towards this goal over the last few years. The developments have been extremely rapid in visual recognition, in which machines now can categorize images into multiple classes, and detect various objects within an image, with an ability that is competitive with or even surpasses that of humans. Meanwhile, we also have witnessed similar strides in natural language processing (NLP). It is quite often for us to see that now computers are able to almost perfectly do text classification, machine translation, etc. However, despite much inspiring progress, most of the achievements made are still within one domain, not handling inter-domain situations. The interaction between the visual and textual areas is still quite limited, although there has been progress in image captioning, visual question answering, etc. In this dissertation, we design models and algorithms that enable us to build in-depth connections between images and natural languages, which help us to better understand their inner structures. In particular, first we study how to make machines generate image descriptions that are indistinguishable from ones expressed by humans, which as a result also achieved better quantitative evaluation performance. Second, we devise a novel algorithm for measuring review congruence, which takes an image and review text as input and quantifies the relevance of each sentence to the image. The whole model is trained without any supervised ground truth labels. Finally, we propose a brand new AI task called Image Aspect Mining, to detect visual aspects in images and identify aspect level rating within the review context. On the theoretical side, this research contributes to multiple research areas in Computer Vision (CV), Natural Language Processing (NLP), interactions between CV&NLP, and Deep Learning. Regarding impact, these techniques will benefit related users such as the visually impaired, customers reading reviews, merchants, and AI researchers in general.

The proposed master's thesis investigates the adaptability and efficacy of self-supervised representation learning (SSL) in medical image analysis, focusing on Mammographic Imaging to develop robust representation learning models. This research will build upon existing studies in Mammographic Imaging that have utilized contrastive learning and knowledge distillation-based self-supervised methods, focusing on SimCLR (Chen et al 2020) and SimSiam (Chen et al 2020) and evaluate approaches to increase the classification performance in a transfer learning setting. The thesis will critically evaluate and integrate recent advancements in these SSL paradigms (Chhipa 2023, chapter 2), and incorporating additional SSL approaches. The core objective is to enhance robust generalization and label efficiency in medical imaging analysis, contributing to the broader field of AI-driven diagnostic methodologies. The proposed master's thesis will not only extend the current understanding of SSL in medical imaging but also aims to provide actionable insights that could be instrumental in enhancing breast cancer detection methodologies, thereby contributing significantly to the field of medical imaging and cancer research.

The construction sector is embossed with uncertainty, where cash flow prediction, time delays, and complex feature interaction make it hard to predict which future projects will be profitable or not. This thesis explores the prediction of project yield and project success for a company in the construction industry using supervised learning models. Leveraging historical project data, parametric traditional regression and machine learning techniques are employed to develop predictive models for project yield and project success. The models were chosen based on previously related work and consultations with employees with domain knowledge in the industry. The study aims to identify the most effective modeling approach for yield prediction and success in construction projects through comprehensive analysis and comparison. The features influencing project yield are investigated using SHAP (SHapley Additive exPla-nations) and permutation feature importance (PFI) values. These explainability techniquesprovide insights into feature importance in the models, thereby enhancing the understandingof the underlying factors driving project yield and project success. The results of this research contribute to the advancement of predictive modeling in the construction industry, offering valuable insights for project planning and decision-making. Construction companies can optimize resource allocation, mitigate risks, and improve projectoutcomes by accurately predicting project yield and success and understanding the keyfactors influencing it. The results in this thesis reveal that the machine-learning models outperform the parametric models overall. The best-performing models, based primarily on accuracy and ROI, were the random forest models with both binary and continuous outputs, leading to a suggested data-driven guideline for the company to use in their project decision-making process.

Modern indoor positioning systems encounter challenges inherent to indoor environments. Signal changes can stem from various factors like object movement, signal propagation, or obstructed line of sight. This thesis explores a supervised machine learning approach that integrates Bluetooth Low Energy (BLE) and inertial sensor data to achieve consistent angle and distance estimations. The method relies on BLE angle estimations and signal strength alongside additional sensor data from an Inertial Measurement Unit (IMU). Relevant features are extracted and a supervised learning model is trained and then validated on familiar environment tests. The model is then gradually introduced to more unfamiliar test environments, and its performance is evaluated and compared accordingly. This thesis project was conducted at the u-blox office and presents a comprehensive methodology utilizing their existing hardware. Several extensive experiments were conducted, refining both data collection procedures and experimental setups. This iterative approach facilitated the improvement of the supervised learning model, resulting in a proposed model architecture based on transformers and convolutional layers. The provided methodology encompasses the entire process, from data collection to the evaluation of the proposed supervised learning model, enabling direct comparisons with existing angle estimation solutions employed at u-blox. The results of these comparisons demonstrate more accurate outcomes compared to existing solutions when validated in familiar environments. However, performance gradually declines when introduced to a new environment, encountering a wider range of signal conditions than the supervised model had trained on. Distance estimations are then compared with the path loss propagation equation, showing an overall improvement. / Moderna inomhuspositioneringssystem möter utmaningar som förekommer i inomhusmiljöer. Signalförändringar kan bero på olika faktorer som objektets rörelse, signalutbredning eller blockerad siktlinje. Denna kandidat avhandling undersöker ett övervakat maskininlärningssätt som integrerar Bluetooth Low Energy (BLE) och tröghetssensorer för att uppnå konsekventa vinkel- och avståndsberäkningar. Metoden bygger på BLE-vinkelberäkningar och signalstyrka tillsammans med ytterligare sensordata från en Inertial Measurment Unit (IMU). Relevanta funktioner extraheras och en övervakad inlärningsmodell tränas och valideras sedan på tester i bekanta miljöer. Modellen introduceras sedan gradvis till mer obekanta testmiljöer, och dess prestanda utvärderas och jämförs därefter. Detta examensarbete genomfördes på u-blox kontor och presenterar en omfattande metodik som utnyttjar deras befintliga hårdvara. Flera omfattande experiment genomfördes, vilket förfinade både datainsamlingsprocedurer och experimentuppsättningar. Detta iterativa tillvägagångssätt underlättade förbättringen av den övervakade inlärningsmodellen, vilket resulterade i en föreslagen modellarkitektur baserad på transformatorer och konvolutionella lager. Den tillhandahållna metodiken omfattar hela processen, från datainsamling till utvärdering av den föreslagna övervakade inlärningsmodellen, vilket möjliggör direkta jämförelser med befintliga vinkelberäkningslösningar som används på u-blox. Resultaten av dessa jämförelser visar mer exakta resultat jämfört med befintliga lösningar när de valideras i bekanta miljöer. Dock minskar prestandan gradvis när den introduceras till en ny miljö, där den möter ett bredare spektrum av signalförhållanden än vad inlärningsmodellen har tränats på. Avståndsberäkningar jämförs sedan med en matematisk formel, kallat path loss propagation ekvationen, som ger distans som en funktion av uppmätt signalstyrka.

Phenotypic profiling systematically maps and analyzes observable traits (phenotypes) exhibited in cells, tissues, organisms or systems in response to various conditions, including chemical, genetic and disease perturbations. This approach seeks to comprehensively understand the functional consequences of perturbations on biological systems, thereby informing diverse research areas such as drug discovery, disease modeling, functional genomics and systems biology. Corresponding techniques should capture high-dimensional features to distinguish phenotypes affected by different conditions. Current methods mainly include fluorescence imaging, mass spectrometry and omics technologies, coupled with computational analysis, to quantify diverse features such as morphology, metabolism and gene expression in response to perturbations. Yet, they face challenges of high costs, complicated operations and strong batch effects. Vibrational imaging offers an alternative for phenotypic profiling, providing a sensitive, cost-effective and easily operated approach to capture the biochemical fingerprint of phenotypes. Among vibrational imaging techniques, infrared (IR) imaging has further advantages of high throughput, fast imaging speed and full spectrum coverage compared with Raman imaging. However, current biomedical applications of IR imaging mainly concentrate on "digital disease pathology", which uses label-free IR imaging with machine learning for tissue pathology classification and disease diagnosis. The thesis contributes as the first comprehensive study of using IR imaging for phenotypic profiling, focusing on three key areas. First, IR-active vibrational probes are systematically designed to enhance metabolic specificity, thereby enriching measured features and improving sensitivity and specificity for phenotype discrimination. Second, experimental workflows are established for phenotypic profiling using IR imaging across biological samples at various levels, including cellular, tissue and organ, in response to drug and disease perturbations. Lastly, complete data analysis pipelines are developed, including data preprocessing, statistical analysis and machine learning methods, with additional algorithmic developments for analyzing and mapping phenotypes. Chapter 1 lays the groundwork for IR imaging by delving into the theory of IR spectroscopy theory and the instrumentation of IR imaging, establishing a foundation for subsequent studies. Chapter 2 discusses the principles of popular machine learning methods applied in IR imaging, including supervised learning, unsupervised learning and deep learning, providing the algorithmic backbone for later chapters. Additionally, it provides an overview of existing biomedical applications using label-free IR imaging combined with machine learning, facilitating a deeper understanding of the current research landscape and the focal points of IR imaging for traditional biomedical studies. Chapter 3-5 focus on applying IR imaging coupled with machine learning for novel application of phenotypic profiling. Chapter 3 explores the design and development of IR-active vibrational probes for IR imaging. Three types of vibrational probes, including azide, 13C-based probes and deuterium-based probes are introduced to study dynamic metabolic activities of protein, lipids and carbohydrates in cells, small organisms and mice for the first time. The developed probes largely improve the metabolic specificity of IR imaging, enhancing the sensitivity of IR imaging towards different phenotypes. Chapter 4 studies the combination of IR imaging, heavy water labeling and unsupervised learning for tissue metabolic profiling, which provides a novel method to map metabolic tissue atlas in complex mammalian systems. In particular, cell type-, tissue- and organ-specific metabolic profiles are identified with spatial information in situ. In addition, this method further captures metabolic changes during brain development and characterized intratumor metabolic heterogeneity of glioblastoma, showing great promise for disease modeling. Chapter 5 developed Vibrational Painting (VIBRANT), a method using IR imaging, multiplexed vibrational probes and supervised learning for cellular phenotypic profiling of drug perturbations. Three IR-active vibrational probes were designed to measure distinct essential metabolic activities in human cancer cells. More than 20,000 single-cell drug responses were collected, corresponding to 23 drug treatments. Supervised learning is used to accurately predict drug mechanism of action at single-cell level with minimal batch effects. We further designed an algorithm to discover drug candidates with novel mechanisms of action and evaluate drug combinations. Overall, VIBRANT has demonstrated great potential across multiple areas of phenotypic drug screening.