Computer Science Theses and Dissertations

Permanent URI for this collectionhttp://hdl.handle.net/1903/2756

Browse

Subject: search.filters.subject.Active Learning

Search Results

Now showing 1 - 3 of 3
  • Thumbnail Image
    Item
    TOWARDS AN EFFICIENT SEMANTIC SEGMENTATION PIPELINE FOR 3D ELECTRON MICROSCOPY DATA.
    (2022) Emam, Zeyad Ali Sami; Czaja, Wojciech; Goldstein, Thomas; Applied Mathematics and Scientific Computation; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
    In recent years, deep neural networks revolutionized many aspects of computer vision. However, their success relies on massive high-quality annotated datasets that are costly to curate. This thesis is composed of three major parts. In Chapter 3, we use novel high dimensional visualization methods to explore connections between the loss landscape of neural networks and their intriguing ability to generalize to unseen test data. Next, in Chapter 4, we tackle a difficult computer vision task, namely the segmentation of anisotropic 3D electron microscopy image volumes. Deep neural networks tend to struggle in this scenario due to the lack of sufficient training data and the 3 dimensional nature of the images, as such we develop a novel state-of-the-art architecture and training workflow to improve the overall segmentation pipeline. Finally, in Chapter 5 we propose a novel state-of-the-art deep active learning algorithm for image classification to alleviate the costs of data annotations and allow networks to train effectively using less data.
  • Thumbnail Image
    Item
    Identifying Graphs from Noisy Observational Data
    (2012) Namata Jr., Galile Mark Supapo; Getoor, Lise; Computer Science; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
    There is a growing amount of data describing networks -- examples include social networks, communication networks, and biological networks. As the amount of available data increases, so does our interest in analyzing the properties and characteristics of these networks. However, in most cases the data is noisy, incomplete, and the result of passively acquired observational data; naively analyzing these networks without taking these errors into account can result in inaccurate and misleading conclusions. In my dissertation, I study the tasks of entity resolution, link prediction, and collective classification to address these deficiencies. I describe these tasks in detail and discuss my own work on each of these tasks. For entity resolution, I develop a method for resolving the identities of name mentions in email communications. For link prediction, I develop a method for inferring subordinate-manager relationships between individuals in an email communication network. For collective classification, I propose an adaptive active surveying method to address node labeling in a query-driven setting on network data. In many real-world settings, however, these deficiencies are not found in isolation and all need to be addressed to infer the desired complete and accurate network. Furthermore, because of the dependencies typically found in these tasks, the tasks are inherently inter-related and must be performed jointly. I define the general problem of graph identification which simultaneously performs these tasks; removing the noise and missing values in the observed input network and inferring the complete and accurate output network. I present a novel approach to graph identification using a collection of Coupled Collective Classifiers, C3, which, in addition to capturing the variety of features typically used for each task, can capture the intra- and inter-dependencies required to correctly infer nodes, edges, and labels in the output network. I discuss variants of C3 using different learning and inference paradigms and show the superior performance of C3, in terms of both prediction quality and runtime performance, over various previous approaches. I then conclude by presenting the Graph Alignment, Identification, and Analysis (GAIA) open-source software library which not only provides an implementation of C3 but also algorithms for various tasks in network data such as entity resolution, link prediction, collective classification, clustering, active learning, data generation, and analysis.
  • Thumbnail Image
    Item
    IMAGE RETRIEVAL BASED ON COMPLEX DESCRIPTIVE QUERIES
    (2011) Siddiquie, Behjat; DAVIS, LARRY S; Computer Science; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
    The amount of visual data such as images and videos available over web has increased exponentially over the last few years. In order to efficiently organize and exploit these massive collections, a system, apart from being able to answer simple classification based questions such as whether a specific object is present (or absent) in an image, should also be capable of searching images and videos based on more complex descriptive questions. There is also a considerable amount of structure present in the visual world which, if effectively utilized, can help achieve this goal. To this end, we first present an approach for image ranking and retrieval based on queries consisting of multiple semantic attributes. We further show that there are significant correlations present between these attributes and accounting for them can lead to superior performance. Next, we extend this by proposing an image retrieval framework for descriptive queries composed of object categories, semantic attributes and spatial relationships. The proposed framework also includes a unique multi-view hashing technique, which enables query specification in three different modalities - image, sketch and text. We also demonstrate the effectiveness of leveraging contextual information to reduce the supervision requirements for learning object and scene recognition models. We present an active learning framework to simultaneously learn appearance and contextual models for scene understanding. Within this framework we introduce new kinds of labeling questions that are designed to collect appearance as well as contextual information and which mimic the way in which humans actively learn about their environment. Furthermore we explicitly model the contextual interactions between the regions within an image and select the question which leads to the maximum reduction in the combined entropy of all the regions in the image (image entropy).