DSpace at MIT
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
Optimizations to a massively parallel database and support of a shared scan architecture - Ahwal, Saher B
This thesis presents a new architecture and optimizations to MapD, a database server which uses a hybrid of multi-CPU/multi-GPU architecture for query execution and analysis. We tackle the challenge of partitioning the data across multiple nodes with many CPUs and GPUs by means of an indexing framework. We implement a QuadTree spatial partitioning scheme and demonstrate how it improves the latencies of many queries when using the index as opposed to not using any. Moreover, we tackle the challenge of processing many queries (perhaps issued concurrently) where queries have very fast latency constraints, e.g, for visualization. We implement a software...
Enhancing the classroom experience with faculty curated discussions - Alsharaiha, Eyas
We identify a number of metrics that attempt to recognize collaborative discussion in online annotation tools and class forums. We hypothesize that posts and questions that lead to collaborative discussion are likely to achieve the same effect if posed again. To study the effects of comment reuse, we build content curation tools that allow content that appears to stimulate collaborative discussion to be promoted or reused from previous class offerings. NB, a social annotation and discussion platform, is extended to provide such tools for annotation reuse, content promotion and curation, and content analysis based on certain metrics as potential indicators...
Simulating tax evasion using agent based modelling And evolutionary search - Badar, Osama
We present a design and model for Simulating Co-Evolution of Tax and Evasion (SCOTE). The system performs agent based modeling of the tax ecosystem and searches for tax evasion strategies using a variant of a Genetic Algorithm with a grammar. Current methodologies and tools to detect, discover or recognize tax evasion are not sufficient. In recent years the tax gap, the aggregate sum of the difference between the tax owed in principle and tax paid in practice was calculated to exceed 450 billion dollars. Numerous tax evasion schemes have surfaced that perform seemingly legal transactions but once observed closely their...
BayesDB : querying the probable implications of tabular data - Baxter, Jay
BayesDB, a Bayesian database table, lets users query the probable implications of their tabular data as easily as an SQL database lets them query the data itself. Using the built-in Bayesian Query Language (BQL), users with little statistics knowledge can solve basic data science problems, such as detecting predictive relationships between variables, inferring missing values, simulating probable observations, and identifying statistically similar database entries. BayesDB is suitable for analyzing complex, heterogeneous data tables with no preprocessing or parameter adjustment required. This generality rests on the model independence provided by BQL, analogous to the physical data independence provided by the relational...
Reactive trajectory adjustment for motion execution using Chekhov - Burke, Sean E
Robots are becoming more and more prominent in the world of manufacturing for assembling products. Currently most of these robots, such as the ones used in automobile manufacturing have specific pre-programmed tasks and motions, and no sense of their surrounding environment. In many of today's applications, this method will not be sufficient, as many real world environments are unstructured and could cause disturbances to the robots requiring the motion and task plans to be modified. If a robot has a task to complete, a planner, such as Bidirectional RRT , will generate a motion plan to complete the task. If...
Design, implementation, and verification of a multipurpose, flexible, three-phase back-to-back voltage-source inverter - Chai, Elaina T
In this thesis, I designed, implemented, and verified a multi-purpose,three-phase back-to- back voltage source inverter for grid connected applications. This inverter features extensive hardware protection, opto-isolation between the power stage and the controller, and reconfigurable I/O routing. The converter is designed for both research and teaching to enable rapid prototyping and verification of advanced control algorithms for grid-connected converters, motor drives etc. I also demonstrate the utility of the power electronics converter as a vector controlled PMSM motor drive. Finally, I use the inverter to perform frequency domain validation of a hardware-in-the-loop simulator. I then design and implement a virtual...
TaleBlazer zap and go with organizations : a platform for managing and broadcasting location-based augmented reality games - Chang, Stephanie N. (Stephanie Ni-Wen)
The TaleBlazer mobile application, developed by the MIT STEP Lab, allows users to create and play custom location-based augmented reality games. Many of the current games are created by TaleBlazer partner organizations, but the means of sharing these games with the general public is difficult and restricted. Zap and Go was conceived as an easier way for organizations to broadcast games to the public, and for the public to download, find, and play these games on their personal mobile devices. This paper presents the backend modifications and user interface designs implemented to support the desired Zap and Go features. This...
InMind : mobile application for sharing the status of serious stuff - Chen, Joy C., M. Eng. Massachusetts Institute of Technology
Individuals under unusual stress, whether due to bereavement, illness, serious conflict, or other transitions can benefit from social support. Support is a valuable resource, but there are many psychological and technical obstacles that can get between an individual and those he gets support from. I am studying how the needs of these individuals are met or not met, how technology is involved, and what can help. To compliment and improve upon existing communication strategies, I designed InMind, an Android application that mediates life-issue tracking between close relations. InMind helps individuals communicate the status of important topics in their lives and...
Hybrid laser with CMOS photonics - Chong, Johanna S
In this thesis, an interesting approach for a photonic laser source is presented. By using integrated photonic resonators with an external gain medium, we are able to build a laser that offers a number of advantages including reducing the electrical and thermal load on the integrated chip socket, eliminating the challenges of integrating gain mediums into CMOS processes, allowing for lasing at virtually arbitrary wavelengths, the possibility of multiwavelength operation with a shared gain medium, elimination of closed-loop control of wavelength tuning, ability to control laser output and wavelength on-chip, and the potential for wavelength modulation using novel resonator tuning...
Modular Lorentz force actuators for efficient biomimetic propulsion of Autonomous Underwater Vehicles - Church, Joseph Christopher
In this thesis, we developed a highly scalable design for modular Lorentz force actuators for use in segmented flexible-hull undersea vehicles such as the RoboTuna being developed at Franklin W, Olin College of Engineering. The actuators were designed to directly drive tail foil sections, or vertebrae, in an oscillatory motion to provide thrust. The design process was automated to facilitate implementation in different sized vertebrae. A set of prototype actuators was manufactured and tested to evaluate the feasibility of the design. A test stand was constructed to evaluate both the static and dynamic performance of the actuators. The prototype actuators...
Evaluation of QUIC on web page performance - Das, Somak R
This work presents the first study of a new protocol, QUIC, on Web page performance. Our experiments test the HTTP/1.1, SPDY, and QUIC multiplexing protocols on the Alexa U.S. Top 500 websites, across 100+ network configurations of bandwidth and round-trip time (both static links and cellular networks). To do so, we design and implement QuicShell, a tool for measuring QUIC's Web page performance accurately and reproducibly. Using QuicShell, we evaluate the strengths and weaknesses of QUIC. Due to its design of stream multiplexing over UDP, QUIC outperforms its predecessors over low-bandwidth links and high-delay links by 10 - 60%. It...
Learning a semantic database from unstructured text - Dhandhania, Keshav
In this paper, we aim to learn a semantic database given a text corpus. Specifically, we focus on predicting whether or not a pair of entities are related by the hypernym relation, also known as the 'is-a' or 'type-of' relation. We learn a neural network model for this task. The model is given as input a description of the words and the context from the text corpus in which a pair of nouns (entities) occur. In particular, among other things the description includes pre-trained embeddings of the words. We show that the model is able to predict hypernym noun pairs...
Characterizing ferrofluid spin-up flow in rotating uniform magnetic fields - Dozier, Kahlil A
A ferrofluid is a collection of nanoscale ferromagnetic particles with a stabilizing surfactant in a liquid to form a colloid. The dynamic behavior of ferrofluids in the presence of magnetic fields has long been an area of research interest. A particular area of interest deals with the "spin-up" mechanisms of ferrofluids, which describe how a container of ferrofluid comes to a steady state of bulk flow when subjected to a uniform rotating external magnetic field. There are two prevailing theories that attempt to explain the spin-up mechanisms of ferrofluids: spin diffusion theory, and the presence of non-uniformities in the magnetic...
Using distributed machine learning to predict arterial blood pressure - Emeagwali, Ijeoma
This thesis describes how to build a flow for machine learning on large volumes of data. The end result is EC-Flow, an end to end tool for using the EC-Star distributed machine learning system. The current problem is that analysing datasets on the order of hundreds of gigabytes requires overcoming many engineering challenges apart from the theory and algorithms used in performing the machine learning and analysing the results. EC-Star is a software package that can be used to perform such learning and analysis in a highly distributed fashion. However, there are many complexities to running very large datasets through...
Tournament based task allocation in a parallel MIS algorithm - Ezeaka, Chidubem L
This paper explores the use of a tournament data structure for task allocation and dependency tracking in a parallel Maximal Independent Set (MIS) algorithm as a way to reduce contention in counter updates and improve runtime. The tournament data structure adds a noticeable overhead to the algorithm that causes T time of the algorithm to increase but then there is a steady improvement in performance as we increase the number of worker threads. Running the tournament algorithm with 12 threads in our experimental setup, we are able to get an average speedup of 1.13 for our test suite of 7...
Hallucination machine : a body centric model of space perception - Zaman, C̦ăgri Hakan
In this thesis I present a novel approach to space perception. I provide a body-centric computational model, The Hallucination Machine, that integrates bodily knowledge with senses in a common modality which I call "the sphere of embodiment". Understanding the human experience of space is an important inquiry not only in the context of design and architecture, but in a broad range of scholarly disciplines where humans are the subject of study, whether as biological, social, or cognitive entities. My vision is that in order to create a knowledge of space shared through different disciplines and to develop tools and methods...
Design and development of a network architecture for a chemical sensor network - Cheung, Jimmy
A real-time continuous chemical sensor network can obtain detailed data to analyze the dynamic behavior of water systems such as a lake. The behaviors of interest to us in Upper Mystic Lake are the effects of stratification on methane water chemistry and the results of water mixing between layers. To monitor the water chemistry, a network of three buoys is populated with various sensors. This paper covers the design implementation of the network architecture for transmitting sensor data between buoys and a shore station. The buoys' sensors and construction are also included.