2018 Winners

Winners of the 2023-2024 Neukom Institute CompX Faculty Grants Program for Dartmouth faculty have been announced for one-year projects. We received over $650K in total requests and awarded a total of $250K.

The program seeks to fund both the development of novel computational techniques as well as the application of computational methods to research across the campus and professional schools.

Dartmouth College faculty including the undergraduate, graduate, and professional schools were eligible to apply for these competitive grants.

Note: * indicates an award that is partnered with assistance from Dartmouth College Research Computing.

+ indicates an award that is partnered with an RA provided through the Scholars program

ArchaeoRover: A Robotic System for Geophysical Survey of Archaeological Landscapes

casana_sm2018.png

Archaeological excavation is a time-consuming, labor-intensive, and fundamentally destructive process, and for these reasons, it is only capable of revealing a tiny fraction of the remains of past human activities that lie below the ground.  Today many archaeologists instead rely on geophysical instruments such as ground penetrating radar (GPR) or magnetic gradiometry to discover, map, and interpret the archaeological record—ancient architecture, artifacts, and features such as roads or field systems—without ever digging.  However, our ability to deploy these instruments over large areas, which could potentially offer an unparalleled perspective on the broader archaeological landscape, is limited primarily by the effort required to map in precise grids, pull ropes and tape measures along transects, and laboriously drag, push, or carry instruments in a lawnmower pattern. This project seeks to develop a new, more efficient approach to the collection of large-scale archaeological geophysics that brings together recent technological developments in robotics, autonomous vehicle navigation software, lightweight RTK GNSS systems, and geophysical instrumentation. 

We plan to work with a small robotics firm to construct a customized, all-terrain, remote-controlled robotic vehicle, the ArchaeoRover, that is capable of carrying a GPR and magnetic gradiometer.  Using open source navigation and mission planning software and hardware designed for autonomous drones, in combination with a new generation of low-cost RTK GNSS units, we will then develop a system that enables surveys to be precisely planned and executed autonomously by the robotic vehicle.  Simultaneously collecting both GPR and magnetic data and moving at around 2-3 meters/second, we anticipate that we will be able to collect geophysical data at comparable resolution to conventional approaches, but over areas many times larger than is currently feasible.  We will then deploy the ArchaeoRover system for testing at several archaeological sites including Etzanoa (Kansas), Midewin National Tallgrass Prairie (Illinois), and the Enfield Shaker Village (New Hampshire), with plans to publish both methodological papers on our approach as well as results of surveys at individual sites.

Investigating the Utility of Learned Latent Dimensions of DNA Methylation in Human Tumors

christensen_sm.png

Our goal is to apply unsupervised deep learning approaches to investigate the utility of learned latent dimensions of DNA methylation in both individual tumor types and across tumor types. Despite the rapid growth of available deep learning approaches, the community of molecular biologists and cancer biologists have been slower to test their utility due to limitations on the scope of available samples with data, and the lack of ground truth labels required for the supervised methods. However, the scope of samples with available molecular data, and newer unsupervised approaches in deep learning that do not require labels overcome these limitations and have shown promise in their application to gene expression data from human tumors. 

We hypothesize that learned, low dimensional latent methylation DNA profiles will reveal meaningful information about the interaction of DNA methylation features in carcinogenesis and disease progression, as well as novel relationships among tumor types. In addition to applying unsupervised deep learning methods, we plan to develop approaches for interpreting the biological significance of learned latent dimensions.

Ancestral Soundscapes and the Origins of Spoken Language

dominy_sm2018.png

hannah_sm.png

symes_field.png

Spoken language is a structured form of communication, containing words and grammar that can be combined in infinite ways. The origin of spoken language is an enduring topic of interest because it is unique to humans and a milestone of human evolution. Given that the frequency range of spoken language is closely aligned with our hearing sensitivity, which differs from nonhuman primates, it has been argued that human speech and hearing are co-evolved traits.

This hypothesis is difficult to test because the extent of human derivation is uncertain (because the hearing of chimpanzees is uncertain) and because few studies have examined the acoustic properties of landscapes ("soundscapes") germane to human evolution. We will use this CompX grant to (a) produce standardized audiograms for chimpanzees and (b) identify habitats in Africa that favor sound transmission in the frequency range that we hear best. Our overarching goal is to better understand the ecological conditions that shaped the acoustics of spoken language.

Force of Things: An Open for Objects

fure_sm.png

The Force of Things: an Opera for Objects is an immersive work of music theater that wrestles with the animate vitality of matter and the mounting hum of ecological anxiety around us. Created with architect Adam Fure and the International Contemporary Ensemble, the project is driven by a desire to tune our focus toward a rate of change and a scope of alteration at odds with the scale of human life. Audience members move throughout a field of sculpted matter ringed by speaker cones projecting subaudible sounds. Though vibrating under the threshold of human hearing, this choir of subwoofers cause waveforms that ripple through and disturb the saturated material space.

Premiered at Kasser Theater in Montclair, New Jersey in October 2017, this CompX grant will support the fabrication of digitally milled topological surfaces designed for The Force of Things’ New York Premiere during Lincoln Center’s Mostly Mozart Festival in August 2018.

Associating Fingerprint Patterns with Age and Sex: A Quantifiable Approach

hruby_sm2018.png

A wide range of prehistoric and ancient Greek ceramic objects, including vessels, ceramic sculpture, seal impressions, and writing tablets preserve the fingerprint impressions of their producers. Traditionally, archaeologists have matched prints in an effort to understand ancient labor systems, but in part because of various pragmatic impediments to matching prints and in part to broaden the range of questions that can be asked of ancient prints, archaeologists have begun to ask questions that require less specific print material. The ages and sexes of producers are among those questions, but so far, the techniques that have been used to reconstruct those factors have typically been able to work on the level of populations rather than individuals, and they have also been subject to challenges posed by differential clay shrinkage rates.

The current project will improve the accuracy of sexing and aging ancient Greek samples by using fingerprints accidentally impressed in objects made by modern Greek ceramicists as a reference sample. Fingerprint impressions from modern Greek adult potters of known sexes and age grades have already been collected and scanned with a high-resolution 3D scanner. A Greek attorney has been hired who will assist us to understand the requirements of both European Union and Greek law as they relate to the collection of prints from juveniles, and we will begin the process of collecting and scanning the prints in the summer of 2018.

Becoming a Bright Lifer: Sentiment Change and Sustainable Weight-Loss*

lively_sm2018.png

Last year over two-thirds of American adults were classified as either overweight or obese. And every year, millions of dollars are spent on weight loss programs, gym memberships, supplements, or coaches. Despite the effort and resources that many Americans put into their attempts to lose weight, most of these efforts are doomed to fail. Recent studies suggest that only one percent of obese individuals who attempt to lose all of their excess weight will be successful, and of that one percent, the majority will have gained their weight back within less than a year.

This research attempts to understand the social factors and social psychological mechanisms that underlie successful long-term weight loss, by examining the sentiment changes and narrative accounts of individual experiences of weight loss and the transition into weight maintenance, within the context of a specific for-profit on-line weight loss community, Bright Line Eating, LLC (BLE). Although BLE is a relatively new company, industry data suggests that individuals who follow the program (which was founded by a cognitive psychologist) are fifty-five times more likely to achieve their weight loss goals within the course of a single year than individuals who follow other weight loss programs, including Jenny Craig and Weight Watchers.

In order to systemically address how social and social psychological mechanisms are related to successful sustainable weight loss, this research will draw on the experiences of individuals who have been enrolled in the BLE community for a minimum of 8 weeks. Currently, individuals in BLE range in age from 18-89 and reside in over 157 countries, making this one of the most ambitious sociological studies on successful weight loss ever conducted. This project will develop, implement, and test a suite of innovative computational tools for the collection, transcription, and analysis of qualitative data, including, but not limited to, in-depth open-ended surveys, and in-depth open ended interviews.

Optical Wavefront Shaping for Deep tissue Imaging: Combining Advanced Computational Ultrasound-based Methods with Optical Wavefront Shaping to Enable Focusing of Light Deep in Tissue

luke_sm2018.png

Optical imaging of living cells and organisms has led to innumerable discoveries in the past century. A steady advance of technologies ranging from fluorescence imaging to precise optics to super-resolution imaging have enabled these discoveries. Despite this progress, however, one critical barrier remains in optical imaging of living tissue: tissue is a highly scattering material. The ability to resolve images is limited to depths of 1 to 2 millimeters. Beyond this depth, photons are scattered and images are irreversibly blurred. In this project, we combine advanced computational ultrasound-based methods with optical wavefront shaping to enable focusing of light deep in tissue.

We are working to develop methods to use ultrasound to generate single microbubbles in tissue. These microbubbles will act as an embedded lens, distorting the light that travels through it. By measuring the optical wavefront before and after generation of the microbubble, the light which traveled through the microbubble can be isolated. Then digital optical phase conjugation can be applied to the sensed light to refocus the light to the location of the microbubble. Thus a small optical focus can be achieved even in highly scattering tissue. This approach will be applied to fluorescence imaging with the end goal of deep-tissue high-resolution imaging in living organisms.

Web-based Platform for Creating Chinese Language and Culture Multimedia Materials*

mowry_sm2018.png

qi_sm2018.png

The aim of this project is to create a digital platform for compiling and authoring web-based Chinese language and culture instructional and research multimedia materials. We will prototype this platform to produce two original pieces of courseware based on two classic Chinese animation films: Buffalo Boy and the Flute (1963) and Three Monks (1980).

The project combines traditional archival research with contemporary educational and research technologies. Each of the final multimedia products will represent an in-depth study of the featured animated film and a leading-edge digital Chinese language textbook with unique cultural content. The main multimedia components of the courseware are:

The original videos (whose audio components consist solely of background, instrumental music and sounds of nature) with added subtitles and voice-over that offer simple audiovisual interpretations of the animations.

Chinese text written in either traditional or simplified characters: The text is a much longer version of the video’s subtitles, presented as a “voiced book” in Chinese. Users can also click to cross-reference either the embedded vocabulary or the section on grammar at any time while reading/listening.
English texts, and still and audiovisual images on or about the animations. This part of the project constitutes a traditional scholarly examination of Chinese animated film and should prove useful for any student or instructor of Chinese film or animation.

Measuring Misinformation and Countering its Spread in the Era of “Fake News”

nyhan_sm2018.png

The spread of online misinformation and widespread distrust of the media threaten the health of American democracy. With support from the Neukom Institute, Nyhan and his co-authors will collect paired behavioral and survey data from a nationally representative panel during the 2018 election campaign to assess the causes and consequences of the consumption of political information from hyperpartisan and "fake news" outlets.  

Smart Rocks: Quantifying Stream Channel Responses to Natural and Human-Induced Disturbances  

renshaw_sm2018.png

magilligan_photo_2.jpg

The increasing frequency and magnitude of extreme precipitation events caused by on-going climate change has already cost the U.S. more than $350 billion over the last decade.  Efforts to mitigate future impacts of increasing extreme precipitation require the ability to predict the response of fluvial systems to increased flows. This response is fundamentally controlled by the movement and residence time of channel sediment which, in turn, requires detailed analyses of channel hydraulics and morphology.  We propose to track gravel mobilization and transport by embedding data logging accelerometers and Radio Frequency Identification (RFID) tags into natural (and perhaps artificial) clasts across a range of sizes. 

The project has four major computational components: i) The high resolution characterization of the spatial variation in channel bed grain size using automated image analysis.  ii) The synthesis of the RFID antenna data (to detect the RFID tags) with the high resolution GPS receiver data (to determine digitally based precise location once a signal is detected).  iii) The use of a 2-D river flow model to determine velocity and shear stress conditions throughout the study reaches over the duration of a storm event.  And iv) the synthesis of the tracer location and mobilization time data with the time-varying modeled shear stress distributions to determine the shear stress at the onset of particle motion.  The merging of these computationally robust approaches will help to better constrain predictive models of stream channel recovery to natural and anthropogenic disturbances.

Assessing the Effects of Deep Brain Stimulation on Agency*

roskies_sm.png

Philosophers have long relied upon the notion of agency in discussions in ethics, philosophy of mind, and moral psychology, but there is no general consensus about what agency is. With the advent of novel techniques for manipulating the brain and affecting behavior, both pharmacologically and technologically, there is renewed attention to agency and the potential risks involved in manipulating or circumventing it. Our overall aim is to explore the ways in which structured questionnaires and online behavioral assays that assess a broad array of aspects of thought and behavior with relevance to agency can be used to inform our conception of agency.

We will apply machine learning techniques and data reduction methods to high-dimensional data sets describing agency from normal populations. Comparing these data-derived factors with intuitive conceptions of agency will inform our understanding of agency, writ large. Although this is an experiment in computational philosophy, this work can ultimately be used to examine how agency is altered in disease and with interventions such as DBS, enabling us to address neuroethical questions about these changes from a new vantage point.

Spatiotemporal Analysis of Communicable Diseases based on Epidemic Trees: A Case Study of Dengue Fever in China*

shi_sm2018.png

Vector-borne diseases (re)emerged as a response to global climate change, globalization, and urbanization, annually causing more than 700,000 deaths globally. Among all vector-borne diseases, dengue is most prevalent, evolving from a sporadic disease to a major public health problem. The global annual dengue incidence has increased 30-fold in the past 50 years, mostly in Asia, followed by Latin America and Africa. 

Using a dataset of dengue cases in China for a 10-year period, we will explores a novel approach to spatiotemporal modeling of vector-borne diseases that takes advantage of computational methodologies and big data. Specifically, we will employ and further develop a method called epidemic trees. We will build the epidemic trees based on detailed patient data, and enhance the modeling with big data of human mobility (transportation and social media). With the constructed epidemic trees, we intend to achieve detailed characterizations of epidemic processes, detection of their associations with environmental factors, and ultimately prediction of future spatiotemporal patterns of the disease.

The primary anticipated outcome of this pilot project is an effective and robust process for constructing epidemic trees through incorporating detailed patient data, geospatial analysis, and machine-learning process. A software tool that implements this process will be created and will be included in ArcHealth, a software package that we have been developing for spatial epidemiology.

Reconstructing the Genetic and Epigenetic Effects of Ancient Urbanism and Geopolitical Transformation in the Teotihuacan Valley, Mexico

17_neuk_fellow_rick2_2_copy.jpg

nichols_sm.png

Teotihuacan, an archaeological site northeast of present day Mexico City, was one of the great urban centers of the ancient world. At its height between 250-550 CE, the city was home to more than 100,000 people, and served as the multi-ethnic epicenter of the Mesoamerican world. But after 550 CE, the city fell into decline, and by 600-650 CE, the city was de-populated and its public buildings were burned.

Today, Teotihuacan provides one of the best known archaeological models of early urbanism in the ancient world, yet the precise circumstances surrounding the rise and decline of the city remain controversial among archaeologists. In this research, we will apply emerging methods in paleogenomics and computational biology to reconstruct ancient genomes from human remains that were excavated from multiple mortuary sectors throughout Teotihuacan and its surrounding hinterland.

Using these ancient genomes, our goals are to assess whether the rise and expansion of Teotihuacan reshaped the genetic diversity of the surrounding hinterland, whether patterns of urban segregation that developed along class and ethnic lines led to genetic population structures within Teotihuacan’s urban core, and to what extent the decline of Teotihuacan shaped patterns of genetic diversity throughout central Mexico today. Taken together, this research aims to provide an unprecedented glimpse into how the rise and decline of Teotihuacan reshaped the genetic landscape of central Mexico. 

Unimagined Sounds: Computational Sound Sculpture and Instrument Design

topel_sm2018_2.png

Generative Computer-Aided Design (CAD) continues to build momentum in both virtual and real contexts. Programs such as AutoCAD and SolidWorks offer generative methods as well as the means to inject your own algorithms into their designs. In my own creative practice, generative techniques increasingly provide new ways to generate compelling physical forms via scripting.

This research aims to develop tools that combine generative processes in such a way as to allow artists to audition and control elements in both the sonic and visual domains. To do this, I propose bridging two well-developed modeling approaches from both domains: Physical Modeling Synthesis and Finite Element Analysis (FEA), to yield a tool that will account for both the forces applied to it (e.g. vibration from a speaker or string) and the resulting dynamical system responses, which can affect everything from the sound characteristics over time to material fatigue and integrity. 

Once built, the second phase of the project is to incorporate this tool into a generative CAD environment with the aim to deploy 3D fabricated prototypes. From these prototypes we will. Artists could then generate designs for objects, structures, and instruments and immediately listen to how they might sound if actuated or driven by some arbitrary physical force, such as air, vibration, or actuation. Alternatively, an artist might apply the process in reverse, and use parameters of sound to generate unimaginable structures and instruments. 

Continuous Motion Sensing Using Computational Fabrics

zhou_sm2018.png

Fine-grained, continuous sensing of body joint motion is critical for a wide range of applications such as health monitoring, sports analytics, and athlete training. Supporting these applications requires motion sensing systems that are portable, convenient to wear, low-power, and capable of sensing subtle (millimeter-level) joint motion. Existing approaches, however, either demand heavy instrumentation of the environment, achieve coarse sensing granularity, or impose on users the burden of constantly wearing on-body electrical sensors. 

We propose to study a new sensing paradigm that requires neither infrastructure support nor any on-body electrical sensors. It exploits the use of fabrics/textile alone as an unobtrusive on-body sensing medium that continuously senses and recovers the motion of body joints with fine granularity. While the concept of e-textile has been proposed in the past, prior works either offer coarse sensing capabilities, or still require embedding electronics into textiles by using textile as a substrate for sensor attachment. We plan to advance the state-of-the-art by realizing fine-grained sensing of body joint motion and more importantly, focusing on the use of off-the-shelf, low-cost conductive fabrics alone for motion sensing without adding any electrical sensors. Such a minimalist sensing approach renders the sensing system comfortable to wear and low-power. 

Study Dartmouth Students’ Learning by Gathering Evidence About Their Experiences

van_kley_sm2018.png

At Dartmouth's Institute for Writing and Rhetoric, faculty from diverse disciplines come together to study writing and speaking. Over the past eight years, the Institute has gathered thousands of student-produced artifacts in an attempt to better understand how students develop as writers and speakers at the College. Dozens of faculty have studied these artifacts in independent and collaborative research and assessment projects.

This project aims to develop a relational database to put these and future projects into dialogue. The database will promote enhanced collaboration across faculty, permit more synthetic understandings of student learning across time, and reduce barriers for faculty looking to participate in computational research on writing for the first time.