Recursos de colección

DSpace at MIT (101.070 recursos)

This site is a university repository providing access to the publication output of the institution. Registered users can set up email alerts to notify them of newly added relevant content. A certain level of encryption and security is embedded in the site which may cause some users accessibility problems.

Electrical Engineering and Computer Sciences - Master's degree

Mostrando recursos 1 - 20 de 6.547

  1. Low frequency active sonar performance in the Arctic Beaufort Lens

    Carper, Scott Adams
    A newly discovered double ducted acoustic environment present throughout much of the Beaufort Sea in the Arctic has a major effect on active acoustic transmissions. This work performs an in depth analysis of how the lower duct impacts the propagation of various active signals used commonly for acoustic communications or active sonar. First, this thesis performs a thorough modal analysis of the effect of the double ducted environment on long range propagation of a 300 Hz and 3500 Hz pulse. Signal excess is determined for the two different source pulses to quantify the effect of the lower duct on noise...

  2. Automatic 3D surface area measurement for vitiligo lesions

    Huang, Jiarui, M. Eng. Massachusetts Institute of Technology
    Vitiligo is a long term skin depigmentation disease that may result in psychological stress. Monitoring changes in vitiligo lesion area allows assessment of treatment efficacy and aids in clinical decision making. Currently existing approaches for vitiligo lesion measurement are either inefficient or inaccurate. Using a RGB-D camera (Kinect) and imaging processing techniques, we develop an automated skin lesion screening system (VLAMS) that can be widely adopted in clinics. VLAMS is tested using clinical medical data. Results show VLAMS can consistently segment target lesion region and accurately measure lesion area of any skin surface.

  3. SibylFuzzer : stateful fuzzing for file systems

    Zuo, Catherine (Catherine W.)
    Correct file system behavior is vital to developing robust higher-level software and applications. However, correctly and efficiently investigating the wide range of file system behavior makes testing file systems a difficult task. In this thesis, I designed and implemented SibylFuzzer, a stateful fuzzer for testing file system behavior. Sibyl- Fuzzer is based on SibylFS, a third-party system comprised of a model for acceptable file system behavior and a procedure for comparing real-life file system implementation behavior against that model. SibylFuzzer uses SibylFS in two ways: first, as a source of file system knowledge to produce in-depth and meaningful tests; second,...

  4. ModelDB : tools for machine learning model management and prediction storage

    Viswanathan, Srinidhi
    Building a machine learning model is often an iterative process. Data scientists train hundreds of models before finding a model that meets acceptable criteria. But tracking these models and remembering the insights obtained from them is an arduous task. In this thesis, we present two main systems for facilitating better tracking, analysis, and querying of scikit-learn machine learning models. First, we introduce our scikit-learn client for ModelDB, a novel end-to-end system for managing machine learning models. The client allows data scientists to easily track diverse scikit-learn workflows with minimal changes to their code. Then, we describe our extension to ModelDB,...

  5. High voltage sampling scheme independent of capacitor voltage coefficient for a delta sigma modulator

    Uyehara, Chad P
    The rapidly growing market of electric vehicles has motivated the research for safe and effective monitoring of large battery packs. This paper presents a unique sampling scheme for a delta sigma modulator that is independent of common mode voltage. The rejection of common mode voltage is important in the application of electric vehicles and other large battery packs in order to accurately measure every cell on the stack with varying common mode voltages. A switched capacitor implementation is the proposed solution, and simulations in Cadence provide the results. The simulations of the circuit supported the claim that it is independent...

  6. Theory and application of neural and graphical models in early cancer diagnostics

    Radhakrishnan, Adityanarayanan
    With the recent availability of large training datasets and graphics processing units (GPUs), we address challenges in the application of graphical models and neural networks to prediction sensitive areas such as healthcare. We begin by presenting our work in the context of learning graphical models from biological data. Namely, we present a combinatorial perspective of Markov Equivalence Classes (MECs), which defines the size of solution spaces when attempting to learn a graphical model from data. Through our analysis, we show that the size of these MECs can be exponential with respect to features of the graph (such as average degree)....

  7. Experimental implementations of stereo matching algorithms in Halide

    Zhang, Min, M. Eng. Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science (2016)
    Currently, most stereo matching algorithms focus their efforts on increasing accuracy at the price of losing run-time performance. However, applications such as robotics require high performance stereo algorithms to perform real time tasks. The problem is due to the difficulty of hand optimizing the complicated stereo matching pipelines. Halide is a programming language that has been widely used in writing high-performance image processing codes. In this work, we explore the usability of Halide in the area of real-time stereo algorithms by implementing several stereo algorithms in Halide. Because of Halide's ability to reduce the computation cost of dense algorithms, we...

  8. An Improved TaleBlazer Editor to encourage and facilitate collaboration between multiple TaleBlazer game designers

    Wang, Linda, M. Eng. Massachusetts Institute of Technology
    TaleBlazer is a platform for creating and playing educational location-based augmented reality games. These games are made by both adults and children game designers using the online TaleBlazer Editor. Oftentimes, building a TaleBlazer game is a group effort, but the previous Editor made collaboration inefficient, difficult, error-prone, and frustrating. This thesis describes a suite of features designed to improve version control for a game and assist designers in sharing their work with others. Together, these features encourage and facilitate collaboration between multiple TaleBlazer designers.

  9. DCG-UPUP-Away : automatic symbol acquisition through grounding to unknowns

    Tucker, Mycal (Mycal D.)
    Research in automatic natural language grounding, in which robots understand how phrases relate to real-world objects or actions, offers a compelling reality in which untrained humans can operate highly sophisticated robots. Current techniques for training robots to understand natural language, however, assume that there is a fixed set of phrases or objects that the robot will encounter during deployment. Instead, the real world is full of confusing jargon and unique objects that are nearly impossible to anticipate and therefore train for. This thesis presents a model called the Distributed Correspondence Graph - Unknown Phrase, Unknown Percept - Away (DCG-UPUP-Away) that...

  10. An investigation of sparse tensor formats for tensor libraries

    Tew, Parker Allen
    Tensors provide a generalized structure to store arbitrary indexable data, which is applicable in fields such as chemometrics, physics simulations, signal processing and lies at the heart of machine learning. Many naturally occurring tensors are considered sparse as they contain mostly zero values. As with sparse matrices, various techniques can be employed to more efficiently store and compute on these sparse tensors. This work explores several sparse tensor formats while ultimately evaluating two implementations; one based on explicitly storing coordinates and one that compresses these coordinates. The two formats, Coordinate and CSF2, were evaluated by comparing their execution time of...

  11. Long-term multi-object tracking in indoor environments using network flows

    Zhang, Thomas, S.M. (Thomas R.). Massachusetts Institute of Technology
    This thesis presents a system for long-term tracking of multiple people in indoor environments using a min cost flow algorithm to optimize the association of short-term tracklets. This system is built on top of an existing RF-based indoor localization system called WiTrack, which is able to track a number of people without requiring them to hold or wear any special devices. However, WiTrack relies primarily on being able to detect motion from a person to track him, and has its effectiveness limited by other factors such as obstructions and multipath effects. Consequently, WiTrack's strength lies in tracking over shorter time...

  12. Compiling Gallina to go for the FSCQ file system

    Ziegler, Daniel (Daniel M.)
    Over the last decade, systems software verification has become increasingly practical. Many verified systems have been written in the language of a proof assistant, proved correct, and then made runnable using code extraction. However, due to the rigidity of extraction and the overhead of the target languages, the resulting code's CPU performance can suffer, with limited opportunity for optimization. This thesis contributes CoqGo, a proof-producing compiler from Coq's Gallina language to Go. We created Go', a stylized semantics of Go which enforce linearity, and implemented proof-producing compilation tactics from Gallina to Go' plus a straightforward translation from Go' to Go....

  13. Splinter : practical private queries on public data

    Yun, Catherine (Catherine T.)
    Every day, millions of people rely on third party services like Google Maps to navigate from A to B. With existing technology, each query provides Google and their affiliates with a track record of where they've been and where they're going. In this thesis, I design, engineer, and implement a solution that offers absolute privacy when making routing queries, through the application of the Function Secret Sharing (FSS) cryptographic primitive. I worked on a library in Golang that applied an optimized FSS protocol, and exposed an API to generate and evaluate different kinds of queries. I then built a system...

  14. An efficient score alignment algorithm and its applications

    Zhang, Emily H
    String alignment and comparison in Computer Science is a well-explored space with classic problems such as Longest Common Subsequence that have practical application in bioinformatic genomic sequencing and data comparison in revision control systems. In the field of musicology, score alignment and comparison is a problem with many similarities to string comparison and alignment but also vast differences. In particular we can use ideas in string alignment and comparison to compare a music score in the MIDI format with a music score generated from Optical Musical Recognition (OMR), both of which have incomplete or wrong information, and correct errors that...

  15. Design and testing of wearable and long-term subdermal implantable electroencephalograms

    Yang, Jason, M. Eng. Massachusetts Institute of Technology
    Electroencephalography (EEG) has long been used by neurologists to aid in diagnosing and treating neurological disorders ranging from sleep apnea to epilepsy. However, inherent difficulties still exist in capturing EEG data for extended periods of time on the order of days to weeks in humans. Such difficulties brought on by implementation challenges and usability ultimately lead to patient non-compliance. These challenges also curtail EEG use in sleep studies due to complexity of setup. This work aims to address these issues by extending the functionality and performance of a previously designed EEG ASIC. A design for a subdermal, implantable EEG recording...

  16. Database engine integration and performance analysis of the BigDAWG polystore system

    Yu, Katherine (Katherine X.)
    The BigDAWG polystore database system aims to address workloads dealing with large, heterogeneous datasets. The need for such a system is motivated by an increase in Big Data applications dealing with disparate types of data, from large scale analytics to realtime data streams to text-based records, each suited for different storage engines. These applications often perform cross-engine queries on correlated data, resulting in complex query planning, data migration, and execution. One such application is a medical application built by the Intel Science and Technology Center (ISTC) on data collected from an intensive care unit (ICU). This thesis presents work done...

  17. ELK : an online, role-playing game to better elicit learner knowledge

    Wong, Tiffany Chi Man
    Teachers that elicit learner knowledge at the beginning of new units are more aware of student preconceptions and reasoning. Teachers who understand student preconceptions are able better to anticipate misconceptions and help students assimilate new knowledge. However, many pre-service and in-service teachers neither collect student ideas nor know how to do so. For my thesis, I have designed and implemented Eliciting Learner Knowledge (ELK), an online, role-playing game for pre-service teachers. While playing ELK, users experiment and learn skills and strategies for gathering students ideas. The game has been iteratively user tested during Teaching System Lab's Dine and Play sessions...

  18. Improving information extraction by acquiring external evidence with reinforcement learning

    Yala, Adam
    Most successful information extraction systems operate with access to a large collection of documents. In this work, we explore the task of acquiring and incorporating external evidence to improve extraction accuracy in domains where the amount of training data is scarce. This process entails issuing search queries, extraction from new sources and reconciliation of extracted values, which are repeated until sufficient evidence is collected. We approach the problem using a reinforcement learning framework where our model learns to select optimal actions based on contextual information. We employ a deep Q-network, trained to optimize a reward function that reflects extraction accuracy...

  19. Build your own deep learner

    Wong, David, M. Eng. (David Y.). Massachusetts Institute of Technology
    BYODL is a framework for building deep learning-based mobile apps to solve domain-specific image recognition problems. Domain-specific image recognition problems are challenging due to lack of labeled data - few have the expertise to assign labels to the images. By using the mobile app to collect data, our framework speeds up the process of improving the model's performance and makes the updated version readily available to app users. By handling the details of setting up the infrastructure and the mobile app boilerplate, BYODL helps users produce a functional image recognition app in a matter of hours instead of months. We...

  20. Enabling scalable multicolor connectomics through expansion microscopy

    Wohlwend, Jeremy
    Decades of work in computational power and high resolution imaging have made it possible to observe specific neurons and their connections at a microscopic level. This level of resolution is known as the connectomic level. While we have been able to observe the brain at such resolution for some time thanks to electron microscopy (EM), the analysis of the resulting data has proven to be a very challenging task, mainly because it relies only on membrane contrast to differentiate cells. Expansion microscopy (ExM) offers an exciting alternative to EM by expanding physical brain tissue, with the added benefits that regular...

Aviso de cookies: Usamos cookies propias y de terceros para mejorar nuestros servicios, para análisis estadístico y para mostrarle publicidad. Si continua navegando consideramos que acepta su uso en los términos establecidos en la Política de cookies.