Skip to main content
[an error occurred while processing this directive]

[an error occurred while processing this directive]
HomePrograms >

Neukom Undergraduate Research

2017 Neukom Funded Undergrad Impacts

neural test

 

Info Theoretics, Very Large Datasets & the Human Limits of Teamwork

Student • Bruno Korbar, CS
Mentor • Seth Frey, PBS, Digital Humanities, Engineering, Government, Neukom Postdoctoral Fellow

Bruno Korbar is a third (now fourth) year CS student is an outstanding, talented machine learning enthusiast. To supplement what he saw as an inability of Dartmouth courses to fully satisfy his appetite for a focused education in machine learning, he has taken on research projects with faculty. We worked together on an information theoretic approach to a very large basketball dataset that promises to offer new insights into the human limits of teamwork. Bruno blew me away not only with his competence and interest, but his ability to find the weak points in my own thinking. The resulting discussions always made me both more clear and correct. Around the 8th week he reported in a class discussion that I'd finally gotten through to him (at least a little): he'd always noticed that social science was far behind the other sciences in terms of methods and answers and he had always assumed that that was because social scientists weren't very rigorous. But as a result of my course and his work with me, he’d discovered that studying humans and sociality scientifically is just much harder than he'd thought.  Bruno has performed well enough in his three years (and counting), that he earned an internship at Microsoft. As proud of him as I am, I'll be eager to have him back this Fall.

 

Linear Glacier Modelling

Student • David Cavagnaro
Mentor • Alice Doughty, Earth Sciences, Thayer, Neukom Postdoctoral Fellow

The Neukom Scholars Program gave me the first opportunity to mentor an undergraduate research assistant, and gave a student, David Cavagnaro, his first opportunity to conduct computational scientific research during the winter term of 2015. David came to me with impressive programming skills and a love for the Earth sciences. He successfully applied for the Neukom Scholar Program to code a linear glacier model from published manuscripts to conduct glacier length sensitivity tests.

David coded the glacier model and seemed to enjoy the process, so I invited him to work with me on a separate research project analyzing climate data from Uganda in 2016 (funded by the John Lindsley Fund). His friendly personality, strong work ethic, and excitement for the research impressed my mentor, Meredith Kelly, and she invited him to participation in the June 2016 Ugandan field expedition to the Rwenzori Mountains (funded by the Comer Science and Education Foundation). David graduated this past weekend after completing a senior research thesis on dating the age of a landslide in the Rwenzori Mountains. None of this would have happened without the first successful funding opportunity from the Neukom Institute. Thank you for your support!  David will continue to combine his computational skills with analyzing environmental issues.

17cavagnaro 

David and Alice (and Ranger George) at a lunch break about 3500 m above sea level in the Rwenzori Mountains.


 

Neukom Prize for Outstanding Undergraduate Research

Two 1st Prize winners

 

ATLAS (Automated Tone Level Annotation System)

Student • Emily Grabowski 
Mentor • Laura McPherson, Linguistics

Emily has created a tool ATLAS (Automated Tone Level Annotation System) that will import audio from field recordings of spoken languages, extract and normalize the pitch, then automatically assign each syllable in the recording a “tone level”, a number representing which part of the speaker’s range it is pronounced in. The number of levels can be set by the researcher depending upon how much detail about the pitch contours of a speaker’s utterance is necessary. These annotations are an enormous timesaver for the researcher, who currently must annotate each syllable by hand using either his or her ear or raw acoustic measures. The annotations that are computationally produced by ATLAS are also superior in that they are objective and replicable, based on the actual acoustic signal rather than the researcher’s (often imperfect) ear.

  

Selective Sharing of Health Data

Student • Emily Greene
Mentor • Dave Kotz, CS

In her senior honors thesis Emily has leveraged advanced cryptographic techniques to make it possible for wearers of Amulet, or similar mobile Health devices, to upload their mobile Health data to the cloud and then selectively share that data with family, friends, caregivers, and researchers. This challenge – selective sharing of health data – is one of the great unsolved problems in the mobile Health sector.

  

2nd Prize

Augmented Reality for the Visually Impaired 

Student • Jonathan Huang 
Mentors • Emily Cooper (PBS) and Wojciech Jarosz (CS)

“A HoloLens Application to Aid People who are Visually Impaired in Navigation Tasks” Department of Computer Science

Jonathan developed an application for an Augmented Reality (Microsoft HoloLens) device that assists users with severely impaired visual acuity to locate, identify, and read text information. His TextSpotting application makes use of the Microsoft HoloLens’s sensors and combines them with the capabilities of the Google Vision API to locate, highlight and read text.

 

 

Neukom Funded 2014 CompX Project Includes Dartmouth Undergraduate Work

Published in PNAS: Quantitative criticism of literary relationships

Students • Jorge Bonilla Lopez, James Brofos, Ajay Kannan, Lea Schroeder

Mentor • Pramit Chaudhuri 

Abstract

Authors often convey meaning by referring to or imitating prior works of literature, a process that creates complex networks of literary relationships (“intertextuality”) and contributes to cultural evolution. In this paper, we use techniques from stylometry and machine learning to address subjective literary critical questions about Latin literature, a corpus marked by an extraordinary concentration of intertextuality. Our work, which we term “quantitative criticism,” focuses on case studies involving two influential Roman authors, the playwright Seneca and the historian Livy. We find that four plays related to but distinct from Seneca’s main writings are differentiated from the rest of the corpus by subtle but important stylistic features. We offer literary interpretations of the significance of these anomalies, providing quantitative data in support of hypotheses about the use of unusual formal features and the interplay between sound and meaning. The second part of the paper describes a machine-learning approach to the identification and analysis of citational material that Livy loosely appropriated from earlier sources.We extend our approach to map the stylistic topography of Latin prose, identifying the writings of Caesar and his near-contemporary Livy as an inflection point in the development of Latin prose style. In total, our results reflect the integration of computational and humanistic methods to investigate a diverse range of literary questions.

 

 

Last Updated: 7/21/17