Browsing by Subject "deep learning"

Now showing 1 - 6 of 6
  • Thumbnail Image
    Item
    GeoComputational Approaches to Evaluate the Impacts of Communication on Decision-making in Agriculture
    (2018-12) Runck, Bryan
    This dissertation proposes a new geocomputational approach to evaluate how communication-based interventions impact outcomes in agriculture. The decisions that people make in agriculture over the next ten to fifteen years will have long-term global consequences because agriculture is going to need to broadly change in order to meet the needs of the future. Many of the technical requirements and economic demands needed to enhance agriculture’s sustainability have been articulated with relative clarity. What remains opaque are the details of who should change, when, where, and how. A growing number of organizations are turning to communication-based interventions to answer these questions with people who will be impacted by changes. Evaluating these interventions is difficult because they are qualitative, affective, meaning-oriented, and discursive. This dissertation builds on existing trends in geocomputation around qualitative geographic information systems and incorporates new methods from machine learning into spatial agent-based modeling. Doing so allows for largely automated creation of agents from natural language text. The dissertation expands on these new tools in each chapter and applies them to the challenge of evaluating communication- based interventions focused on Midwest agriculture. Results suggest that novel insights can be gained into the inner workings of communication-based interventions for improving decision-making using the approaches described in this dissertation.
  • Thumbnail Image
    Item
    How to Use Brains and Computers to Enhance Brain Computer Interfacing
    (2020-11) Stieger, James
    Brain computer interfaces (BCIs) are assistive devices that provide individuals with paralysis access to the world. Through decoding brain data in real-time, BCIs can translate user intent into actionable commands that can control computer cursors, wheelchairs, and robotic arms. However, many individuals struggle to learn how to control these devices. In this investigation, we explore two methods to improve brain computer interface performance. First, mind-body awareness training was shown to enhance BCI skill through increasing control over alpha band EEG power during rest. Next, deep learning methods were shown to increase the BCI classification accuracy and highlight the merit of EEG with full scalp coverage. In conclusion, we were able to demonstrate BCI performance can be improved through both behavioral and computational methods, which may increase the effectiveness of BCI in the large population who could benefit from alternatives to direct motor control.
  • Thumbnail Image
    Item
    Physics-Driven Deep Learning Techniques for High-Resolution MRI
    (2023-05) Demirel, Omer
    Magnetic resonance imaging (MRI) is a non-invasive diagnostic tool used in clinics to evaluate the functional properties of the human body with superior soft-tissue contrast. Scan duration is a major issue in MRI that requires trade-offs between signal-to-noise ratio (SNR), spatio-temporal resolution, and coverage leading to numerous challenges. The need for faster MRI acquisition is particularly important in cardiac imaging and functional MRI (fMRI), where improved spatio-temporal resolution is essential for better coverage and evaluation. To tackle these issues, accelerated MRI techniques have been developed, such as parallel imaging, simultaneous multi-slice (SMS) imaging and compressed sensing (CS). Although advanced image reconstruction techniques are applied to reduce scan time while maintaining high-quality images, these techniques are limited in certain ways. Hand-crafted sparsity assumptions, blocky artifacts due to reconstruction errors, time-consuming parameter tuning, and long reconstruction times due to the iterative nature of the algorithms are the main limitations of CS. Recently, physics-guided deep learning (PG-DL) or physics-driven deep learning (PD-DL) reconstruction has gained immense interest in fast MRI. PD-DL is particularly useful because it combines the benefits of MRI physics with advanced neural network-based regularization techniques. On the other hand, PD-DL has already shown improved image quality compared to parallel imaging and CS and has led to unprecedented acceleration rates. Yet, PD-DL has its own limitations some of which are, limited training raw-data availability, overregularization or artificially hallucinated image features, generalizability issues across different SNRs, and sensitivity to noise. In this thesis, novel reconstruction methods were introduced to address these challenges using parallel imaging, cutting-edge SMS techniques, and state-of-the-art PD-DL reconstruction. First, we introduced an SMS reconstruction technique that was applied to cardiac MRI (CMR) to achieve faster heart coverage without compromising the image quality. This method addressed noise amplification and inter-slice leakage problems in accelerated SMS imaging using an extended k-space approach where SMS acceleration was characterized as a uniform sub-sampling in the readout direction. Second, we proposed to encode signal intensity variations across image phases into the forward operator of the MRI inverse problem without altering coil sensitivities to tackle the generalizability issue across different SNRs/contrast. This led to a more uniform contrast across the image series, which in turn facilitated generalizability for PD-DL methods. Third, we proposed to use a subject-specific self-supervised physics-guided deep learning reconstruction that exploits spatio-temporal correlations by using a 3D convolutional neural network. This network was trained on a subject of interest without a database to overcome the challenging database learning process of cardiac motion patterns for free-breathing CMR. Fourth, we extended a self-supervised PG-DL reconstruction to 20-fold accelerated 7T fMRI to show functional precision and temporal effects in the subsequent fMRI analysis were not altered by deep learning reconstruction leading to ultra-high acceleration rates with SMS and in-plane acceleration. Lastly, we re-envisioned the conventional fMRI computational imaging pipeline. Instead of performing reconstruction followed by denoising, we achieved improved image quality by employing PG-DL reconstruction after denoising the raw k-space leading to a synergistic combination of thermal noise suppression followed by deep learning reconstruction which leveraged the best of both worlds.
  • Thumbnail Image
    Item
    Predicting Therapy Adherence : A Machine Learning Approach
    (2021-12) Lima Diniz Araujo, Matheus
    Ensuring adherence to medical therapy has been an open problem in health care practices since the Hippocrates Oath (400 BC) to modern medicine. In an ideal world, people would follow their doctor's recommendations. They would stick to their diet to lose weight, take their medication on time, and use their electronic health devices as recommended by the doctors. But the planned routine is rarely followed, causing a financial burden in the order of billions of dollars for the national healthcare system and many billions of dollars worldwide. A key mechanism to revert a tendency of non-adherence is early personalized intervention, targeting the key factors of undesired behavior, but this task is not trivial. After starting their therapy, individuals have an unpredictable series of life events that may impact their willingness to keep with the necessary therapy routine. Only recently, we achieved the ability to passively collect individual-level therapy data as patients progress in their treatments using digital devices. In this thesis, we proposes various machine-learning strategies that aim to leverage the data collected at the early stages of medical therapies to predict future adherence and recommend early accurate interventions that align with each individual's desired outcomes. We narrow most of the analysis in two sleep apnea therapies, Continuous Positive Airway Pressure (CPAP) and Upper-Airway Stimulation (UAS). But to reinforce the generalization of our methods we also show how they can be applied for the growth-hormone therapy management.
  • Thumbnail Image
    Item
    Robustness and Stability of Deep Learning
    (2021-06) Lai, Chieh-Hsin
    This dissertation serves as a collection of my three projects after I received the Ph.D. candidacy in 2018. The first two projects (in Chapter 2 and 3, respectively), joint works with Dongmian Zou and Gilad Lerman, are about novel algorithms for unsupervised and semi-supervised anomaly detection tasks, respectively. Our new methods allow datasets with a high ratio of corruption by outliers. The third project (in Chapter 4), a joint work with Kshitij Tayal, Raunak Manekar, Zhong Zhuang, Vipin Kumar and Ju Sun, brings out a methodology for improving the performance of end-to-end deep learning approaches for inverse problems with many-to-one forward mappings. General features of these three projects are introduced in the following. In Chapter 2, we propose a neural network for unsupervised anomaly detection with a novel robust subspace recovery layer (RSR layer). This layer seeks to extract the underlying subspace from a latent representation of the given data and removes outliers that lie away from this subspace. It is used within an autoencoder. The encoder maps the data into a latent space, from which the RSR layer extracts the subspace. The decoder then smoothly maps back the underlying subspace to a ``manifold" close to the original inliers. Inliers and outliers are distinguished according to the distances between the original and mapped positions (small for inliers and large for outliers). Extensive numerical experiments with both image and document datasets demonstrate state-of-the-art precision and recall. In Chapter 3, we propose a new method for novelty detection that can tolerate high corruption of the training points, whereas previous works assumed either no or very low corruption. Our method trains a robust variational autoencoder (VAE), which aims to generate a model for the uncorrupted training points. To gain robustness to high corruption, we incorporate the following four changes to the common VAE: 1. Extracting crucial features of the latent code by a carefully designed dimension reduction component for distributions; 2. Modeling the latent distribution as a mixture of Gaussian low-rank inliers and full-rank outliers, where the testing only uses the inlier model; 3. Applying the Wasserstein-1 metric for regularization, instead of the Kullback-Leibler (KL) divergence; and 4. Using a robust error for reconstruction. We establish both robustness to outliers and suitability to low-rank modeling of the Wasserstein metric as opposed to the KL divergence. We illustrate state-of-the-art results on standard benchmarks. In Chapter 4, we propose a methodology to resolve the irregular approximation of the inverse mapping in some inverse problems with many-to-one forward mappings; especially, we focus on 2D Fourier phase retrieval problem. In many physical systems, inputs related by intrinsic system symmetries generate the same output. So when inverting such systems, an input is mapped to multiple symmetry-related outputs. This causes fundamental difficulties for tackling these inverse problems by the emerging end-to-end deep learning approach. Taking phase retrieval as an illustrative example, we show that careful symmetry breaking on the training data can help get rid of the difficulties and significantly improve learning performance in real data experiments. We also extract and highlight the underlying mathematical principle of the proposed solution, which is directly applicable to other inverse problems.
  • Thumbnail Image
    Item
    Supporting Data for "Development of a PointNet for Detecting Morphologies of Self-Assembled Block Oligomers in Atomistic Simulations"
    (2021-08-30) Shen, Zhengyuan; Sun, Yangzesheng; Lodge, Timothy P; Siepmann, J Ilja; siepmann@umn.edu; Siepmann, J Ilja; University of Minnesota MRSEC
    Molecular simulations with atomistic or coarse-grained force fields are a powerful approach for understanding and predicting the self-assembly phase behavior of complex molecules. Amphiphiles, block oligomers, and block polymers can form mesophases with different ordered morphologies describing the spatial distribution of the blocks, but entirely amorphous nature for local packing and chain conformation. Screening block oligomer chemistry and architecture through molecular simulations to find promising candidates for functional materials is aided by effective and straightforward morphology identification techniques. Capturing 3-dimensional periodic structures, such as ordered network morphologies, is hampered by the requirement that the number of molecules in the simulated system and the shape of the periodic simulation box need to be commensurate with those of the resulting network phase. Common strategies for structure identification include structure factors and order parameters, but these fail to identify imperfect structures in simulations with incorrect system sizes. Building upon pioneering work by DeFever et al. [Chem. Sci.2019, 10, 7503–7515] who implemented a PointNet (i.e., a neural network designed for computer vision applications using point clouds) to detect local structure in simulations of single-bead particles and water molecules, we present a PointNet for detection of nonlocal ordered morphologies of complex block oligomers. Our PointNet was trained using atomic coordinates from molecular dynamics simulation trajectories and synthetic point clouds for ordered network morphologies that were absent from previous simulations. In contrast to prior work on simple molecules, we observe that large point clouds with 1000 or more points are needed for the more complex block oligomers. The trained PointNet model achieves an accuracy as high as 0.99 for globally ordered morphologies formed by linear diblock, linear triblock, and 3-arm and 4-arm star-block oligomers, and it also allows for the discovery of emerging ordered patterns from nonequilibrium systems.