1,721,031 research outputs found
OrpheusDB: an attempt towards data version control on relational database
Get PDFThis thesis will cover the deign, manual, and implementation detail of OrpheusDB.Submission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2018-12-01The student, Sili Hui, accepted the attached license on 2016-12-06 at 22:28.The student, Sili Hui, submitted this Thesis for approval on 2016-12-06 at 22:34.This Thesis was approved for publication on 2016-12-07 at 09:14.DSpace SAF Submission Ingestion Package generated from Vireo submission #10439 on 2017-02-28 at 14:37:18Made available in DSpace on 2017-03-01T16:37:06Z (GMT). No. of bitstreams: 2
HUI-THESIS-2016.pdf: 370407 bytes, checksum: 5d25b70350d18de43d3eb78eb2ceacaa (MD5)
LICENSE.txt: 4205 bytes, checksum: 6646600d2ec552a4affb59d50f854faa (MD5)
Previous issue date: 2016-12-07Embargo set by: Seth Robbins for item 98627
Lift date: 2019-03-01T16:37:19Z
Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD systemU of I Only Restriction Lifted for Item 98627 on 2019-03-02T10:15:17Z
Towards open-ended crowd-powered data processing: a case study of clustering and counting
Get PDFDue to the widespread use and importance of crowdsourcing in gathering training data at scale, the data management community has devoted its efforts in understanding and optimizing fundamental primitives like filters and joins. These primitive boolean operations, where the human responses come from a small, finite space of possible answers, are inadequate for a number of data analysis tasks, especially those involving images, videos and maps. There is, thus, a need for open-ended crowdsourcing in order to get more fine-grained information from humans that can be used in developing sophisticated AI systems. In this thesis, we study two popular open-ended crowdsourcing problems. The first, clustering, is the problem of organizing a collection of objects (images, videos) by allowing workers to form as many clusters as they would like and organize items across them. The second, counting, is the problem of counting objects in images. In this thesis, we develop models to reason about human behavior for both problems, and use these models to design provably cost-efficient algorithms that provide high-quality results, as compared to currently available approaches.Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2016-11-09 without embargo termsThe student, Ayush Jain, accepted the attached license on 2016-07-18 at 16:26.The student, Ayush Jain, submitted this Thesis for approval on 2016-07-18 at 16:39.This Thesis was approved for publication on 2016-07-19 at 09:11.DSpace SAF Submission Ingestion Package generated from Vireo submission #9998 on 2016-11-09 at 10:25:26Made available in DSpace on 2016-11-10T17:55:19Z (GMT). No. of bitstreams: 2
JAIN-THESIS-2016.pdf: 18941114 bytes, checksum: 510809166171ffa8f59f0e8aee01d958 (MD5)
LICENSE.txt: 4207 bytes, checksum: ab36553fc08758b3f1fc1cb0e7ffcf5b (MD5)
Previous issue date: 2016-07-1
Enabling effective visual data exploration for solvent discovery in material science
Get PDFData visualization has become increasingly important in almost all scientific fields. However, current visual analytics tools usually require redundant manual processing, resulting in the visualization process remaining overwhelming and error-prone. Zenvisage automates the process of querying for desired visual patterns, thereby speeding up visual exploration. In this work, we collaborate with material scientists, whose goal is to identify battery solvents with favorable properties while considering economical, physical and chemical tradeoffs in their manufacture. We extend Zenvisage to allow material scientists to compare among subsets of data dynamically and employ non-line chart visualizations to explore their data.
In this thesis, we introduce the notion of dynamic class creation, which targets the seamless creation of subsets of data and comparison of properties among them. We address the non-time-series data issue by conducting visual property search queries directly on scatter plots. We implemented polygon-bound queries and drag-and-drop queries for scatter plots, along with two similarity metrics. We also introduce a new approach for material scientists to upload their datasets using scripts. Our work would enable material scientists to get insights more quickly on increasingly large datasets.Submission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2021-05-01The student, Renxuan Wang, accepted the attached license on 2019-04-19 at 19:44.The student, Renxuan Wang, submitted this Thesis for approval on 2019-04-19 at 19:50.This Thesis was approved for publication on 2019-04-22 at 13:59.DSpace SAF Submission Ingestion Package generated from Vireo submission #13802 on 2019-08-22 at 15:07:37Made available in DSpace on 2019-08-23T20:36:05Z (GMT). No. of bitstreams: 2
WANG-THESIS-2019.pdf: 2290782 bytes, checksum: c07219fe08f665e5dcb705e26ed73c03 (MD5)
LICENSE.txt: 4209 bytes, checksum: 9fdd9d05d094df0827e4ee7af555a9f8 (MD5)
Previous issue date: 2019-04-22Embargo set by: Seth Robbins for item 112187
Lift date: 2021-08-23T20:36:18Z
Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD systemU of I Only Restriction Lifted for Item 112187 on 2021-08-24T09:15:16Z
DataSpread: scaling spreadsheets using relational databases
Get PDFSpreadsheet software is the tool of choice for ad-hoc tabular data management, manipulation, querying, and visualization with adoption by billions of users. However, spreadsheets are not scalable, unlike database systems. We develop DataSpread, a system that holistically unifies databases and spreadsheets with a goal to work with massive spreadsheets: DataSpread retains all of the advantages of spreadsheets, including ease of use, ad-hoc analysis and visualization capabilities, and a schema-free nature, while also adding the scalability and collaboration abilities of traditional relational databases. We design DataSpread with a spreadsheet front-end and a regular relational database back-end. To integrate spreadsheets and databases, in this thesis, we develop a storage and indexing engine for spreadsheet data. We first formalize and study the problem of representing and manipulating spreadsheet data within a relational database. We demonstrate that identifying the optimal representation is NP-Hard via a reduction from partitioning of rectangles; however, under certain reasonable assumptions, can be solved in PTIME. We develop a collection of mechanisms for representing spreadsheet data, and evaluate these representations on a workload of typical data manipulation operations. We augment our mechanisms with novel positionally-aware indexing structures that further improve performance. DataSpread can scale to billions of cells, returning results for common operations within seconds. Lastly, to motivate our research questions, we perform an extensive survey of spreadsheet use for ad-hoc tabular data management.Submission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2019-05-01The student, Vipul Venkataraman, accepted the attached license on 2017-04-12 at 12:19.The student, Vipul Venkataraman, submitted this Thesis for approval on 2017-04-12 at 12:20.This Thesis was approved for publication on 2017-04-12 at 15:22.DSpace SAF Submission Ingestion Package generated from Vireo submission #10722 on 2017-08-10 at 15:05:27Made available in DSpace on 2017-08-10T20:32:52Z (GMT). No. of bitstreams: 2
VENKATARAMAN-THESIS-2017.pdf: 1068361 bytes, checksum: d09adf1e73a044e0e5e4ea1dc393e475 (MD5)
LICENSE.txt: 4215 bytes, checksum: f35d5860899c10463eff04f36479665c (MD5)
Previous issue date: 2017-04-12Embargo set by: Colleen Fallaw for item 102745
Lift date: 2019-08-10T21:27:21Z
Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD systemU of I Only Restriction Lifted for Item 102745 on 2019-08-11T09:15:28Z
Towards understanding and simplifying human-in-the-loop machine learning
Get PDF"Machine learning application developers and data scientists spend inordinate amount of time iterating on machine learning (ML) workflows, by modifying the data pre-processing, model training, and post-processing steps, via trial-and-error to achieve the desired model performance. As a result, developers are ""in-the-loop"" of the development cycle. Under this ""human-in-the-loop"" setting, the ultimate goal of a ML system becomes shortening the time to obtain deployable models from scratch. However, some of the existing ML systems ignore this iterative aspect, and only optimize the one-shot execution of the workflow, while some of them don't provide enough support for system users to make iterative changes. Here, we first conduct a mini-survey of the applied machine learning literature to quantitatively study the user behavior in iterative ML application development. Then, we propose Helix, a declarative machine learning system implemented in Scala. Helix mainly focuses on the optimization of the execution across iterations by reusing or recomputing intermediate results as appropriate. Finally, we describe our collaboration system on top of Helix, that includes a workflow management module and a visualization tool, to make the machine learning system easier to use. In our evaluations, Helix achieved a 60% magnitude reduction in cumulative running time compared to state-of-the-art machine learning tools."Submission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2020-05-01The student, Litian Ma, accepted the attached license on 2018-04-25 at 20:27.The student, Litian Ma, submitted this Thesis for approval on 2018-04-25 at 20:28.This Thesis was approved for publication on 2018-04-26 at 09:01.DSpace SAF Submission Ingestion Package generated from Vireo submission #12508 on 2018-08-31 at 17:21:34Made available in DSpace on 2018-09-04T20:41:58Z (GMT). No. of bitstreams: 3
MA-THESIS-2018.pdf: 1940500 bytes, checksum: 56decd2f5bb97ad45eda64078970cf6f (MD5)
UIUCGraduateThesis-master.zip: 7812584 bytes, checksum: 64980208eeb010e16294ee2a53d23a2b (MD5)
LICENSE.txt: 4206 bytes, checksum: b45654ad0207ba8fe8e23231ac95d97e (MD5)
Previous issue date: 2018-04-26Embargo set by: Seth Robbins for item 107316
Lift date: 2020-09-04T20:42:08Z
Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD systemU of I Only Restriction Lifted for Item 107316 on 2020-09-05T09:15:13Z
Efficient pattern-based querying of trend line visualizations
Get PDFEmbargo set by: Seth Robbins for item 107918
Lift date: 2020-09-27T16:47:41Z
Reason: Author requested closed access (OA after 2yrs) in Vireo ETD systemLimited Restriction Lifted for Item 107918 on 2020-09-28T09:15:13Z.Finding visualizations with desired patterns is a common goal during data exploration. However, due to the limited expressiveness and flexibility of existing visual analytics systems, pattern-based querying of visualizations has largely been a manual process. We present ShapeSearch, a system that enables users to express their desired patterns using multiple flexible mechanisms—including natural language and visual regular expressions— and automates the search via an optimized execution engine. Internally, the system leverages an expressive ShapeQuery algebra that supports a range of operators and primitives for representing ShapeSearch queries. We will describe how the various components of ShapeSearch help accelerate scientific discovery by automating the search for meaningful patterns in multiple domains such as genomics and material science.Submission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2020-08-01The student, Zesheng Wang, accepted the attached license on 2018-07-11 at 16:31.The student, Zesheng Wang, submitted this Thesis for approval on 2018-07-11 at 16:45.This Thesis was approved for publication on 2018-07-12 at 14:57.DSpace SAF Submission Ingestion Package generated from Vireo submission #12835 on 2018-09-27 at 11:37:06Made available in DSpace on 2018-09-27T16:47:33Z (GMT). No. of bitstreams: 2
WANG-THESIS-2018.pdf: 809820 bytes, checksum: cafd9d45961f43bfe26ecda691e5be93 (MD5)
LICENSE.txt: 4209 bytes, checksum: 75aaf58a61aedb85bd3910a82e9fdf83 (MD5)
Previous issue date: 2018-07-1
Effortless data exploration with zenvisage: an expressive and interactive visual analytics system
Get PDF"Data visualization is by far the most commonly used mechanism to explore data, especially by novice data analysts and data scientists. And yet, current visual analytics tools are rather limited in their ability to guide data scientists to interesting or desired visualizations: the process of visual data exploration remains cumbersome and time-consuming. We propose zenvisage, a platform for effortlessly visualizing interesting patterns, trends, or insights from large datasets. We describe zenvisage's general purpose visual query language, ZQL (""zee-quel"") for specifying the desired visual trend, pattern, or insight — ZQL draws from use-cases in a variety of domains, including biology, mechanical engineering, climate science, and commerce. We formalize the expressiveness of ZQL via a visual exploration algebra, and demonstrate that ZQL is at least as expressive as that algebra. While analysts are free to use ZQL directly, we also expose ZQL via a visual specification interface. We then describe our architecture and optimizations, preliminary experiments in supporting and optimizing for ZQL queries in our initial zenvisage prototype, and a user study to evaluate whether data scientists are able to effectively use zenvisage for real applications."Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2016-11-09 without embargo termsThe student, Tarique Ashraf Siddiqui, accepted the attached license on 2016-07-12 at 20:43.The student, Tarique Ashraf Siddiqui, submitted this Thesis for approval on 2016-07-12 at 20:47.This Thesis was approved for publication on 2016-07-14 at 11:49.DSpace SAF Submission Ingestion Package generated from Vireo submission #9913 on 2016-11-09 at 10:24:32Made available in DSpace on 2016-11-10T17:55:03Z (GMT). No. of bitstreams: 2
SIDDIQUI-THESIS-2016.pdf: 1609861 bytes, checksum: 7f7b1e6330cff4f99f1107d1da33adb4 (MD5)
LICENSE.txt: 4220 bytes, checksum: a8d19e8d2f5f1afd983859f36115b7d3 (MD5)
Previous issue date: 2016-07-1
Towards understanding and simplifying human-in-the-loop machine learning
No full textMachine learning application developers and data scientists spend inordinate amount of time iterating on machine learning (ML) workflows, by modifying the data pre-processing, model training, and post-processing steps, via trial-and-error to achieve the desired model performance. As a result, developers are "in-the-loop" of the development cycle. Under this "human-in-the-loop" setting, the ultimate goal of a ML system becomes shortening the time to obtain deployable models from scratch. However, some of the existing ML systems ignore this iterative aspect, and only optimize the one-shot execution of the workflow, while some of them don't provide enough support for system users to make iterative changes. Here, we first conduct a mini-survey of the applied machine learning literature to quantitatively study the user behavior in iterative ML application development. Then, we propose Helix, a declarative machine learning system implemented in Scala. Helix mainly focuses on the optimization of the execution across iterations by reusing or recomputing intermediate results as appropriate. Finally, we describe our collaboration system on top of Helix, that includes a workflow management module and a visualization tool, to make the machine learning system easier to use. In our evaluations, Helix achieved a 60% magnitude reduction in cumulative running time compared to state-of-the-art machine learning tools.U of I OnlyAuthor requested U of Illinois access only (OA after 2yrs) in Vireo ETD syste
Assisting data exploration via in-situ adaptive visualizations
Get PDFVisual analytics has been widely used by data scientists to shed light on complex problems. Despite the prevalence of many visual analytics tools that empower human decision making with data-driven insights, challenges still exist that hinder users from genuinely capitalizing on insights from visualizations. The two biggest challenges we identify are the lack of task support and disconnected workflow. Visual analytics tools lack task support because they do not actively suggest insights to the users, requiring users to pick each individual step during exploration manually. These tools also suffer from disconnected workflows by keeping interactive exploration via dashboards separate from data preparation and cleaning tools like computational notebooks.
To address these challenges, we introduce Lux, a visualization recommendation library that automatically generates useful insights for data exploration, and seamlessly integrates into a user’s data exploration workflow by augmenting the Pandas library. In this thesis, we document the design decisions made and the implementation details of Lux as well as how users can easily unlock intelligent analytical capabilities by adding our library to their code. Furthermore, we share how predecessor visual analytics tools such as Zenvisage that we contributed to guided the development of Lux.Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2020-08-25 without embargo termsThe student, Jaewoo Kim, accepted the attached license on 2020-05-12 at 11:58.The student, Jaewoo Kim, submitted this Thesis for approval on 2020-05-12 at 12:04.This Thesis was approved for publication on 2020-05-12 at 15:15.DSpace SAF Submission Ingestion Package generated from Vireo submission #15348 on 2020-08-25 at 17:14:19Made available in DSpace on 2020-08-26T21:58:05Z (GMT). No. of bitstreams: 2
KIM-THESIS-2020.pdf: 3181874 bytes, checksum: d547f70663d2e85f8094329337bb14c3 (MD5)
LICENSE.txt: 4207 bytes, checksum: 83985ae90e10ba1b6870117c294849dc (MD5)
Previous issue date: 2020-05-1
Prostate cancer diagnosis with deep learning
No full textProstate cancer is one of the most common cancers and the second leading
cause of death among American men. However, prostate cancer diagnosis
is one of the most urgent problems confronted by scientific research.
Accurate prostate cancer diagnosis needs a great degree of medical knowledge and is
usually based on experience. It is hard for ordinary men to diagnose
prostate cancer by themselves. This project aims to eliminate the
knowledge
barrier and provide a precise and effective method using deep learning.
This project uses a deep learning neural network to build a binary classifier for
prostate needle biopsies from patients. The project enlarges the prostate
cancer needle biopsies dataset using randomly cutting, builds the deep learning
network binary classifier, and generates predictions for the biopsies. The
classifier will assign a benign or malignant label to every biopsy with accuracy near 100%.U of I Only Restriction set for Item 102956 on 2017-08-28T20:49:00Z with date by [email protected] by Janice Progen ([email protected]) on 2017-08-28T21:12:32Z
No. of bitstreams: 1
ECE499-Sp2017-wang-Bangqi.pdf: 2804978 bytes, checksum: 011a9d97a56681dad20825e54cbb22ef (MD5)Approved for entry into archive by James Hutchinson ([email protected]) on 2017-08-28T21:31:24Z (GMT) No. of bitstreams: 1
ECE499-Sp2017-wang-Bangqi.pdf: 2804978 bytes, checksum: 011a9d97a56681dad20825e54cbb22ef (MD5)Made available in DSpace on 2017-08-28T21:31:24Z (GMT). No. of bitstreams: 1
ECE499-Sp2017-wang-Bangqi.pdf: 2804978 bytes, checksum: 011a9d97a56681dad20825e54cbb22ef (MD5)
Previous issue date: 2017-12Embargo set by: James Hutchinson for item 102956
Lift date: 10000-01-01
Reason: undergraduate senior thesis not recommended for open accessundergraduate senior thesis not recommended for open accessU of I Onl
- …
