2019 Winners

About the CompX Faculty Grants Program

Winners of the 2019-2020 Neukom Institute CompX Faculty Grants Program for Dartmouth faculty have been announced for one-year projects. We received over $1M 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.


Susan Brison

Integrating Moral Reasoning into Information Technology and Artificial Intelligence


Susan Brison

Recent developments in information technology (IT), especially regarding the rapid growth of artificial intelligence (AI), have prompted much public and scholarly debate about these technologies in terms of both their real and potential impacts on society. From universities and government agencies to private corporations and kitchen tables, people now commonly speak of "tech ethics" as one of the most pressing concerns of this era. Dominating these discussions are such questions as: What is ethics, anyway? How can we create ethical tech? Who ought to decide which values are most important and ought to be prioritized? These are challenging questions for all of us. Unfortunately, though, until now, there have been few opportunities for philosophers and political theorists to engage in these debates on an academic level; instead, the ethics and IT landscape has been largely dominated by the fields of law, computer science and engineering. But the rise in interdisciplinary initiatives to address ethical dilemmas surrounding AI, e.g. co-taught courses, workshops, and shared post-docs, at universities around the country, suggests that this is changing.

This project seeks to apply the methods used in ethics and political theory to some current problems arising in the use of computational techniques. It will involve planning and hosting a series of three lectures on Ethics, Society, and Information Technology to take place in 20X in conjunction with a new Dartmouth course—PHIL 9.08: Ethics and Information Technology. Each of the public lectures will bring to Dartmouth a moral philosopher or political theorist to address an ethical issue posed by new information technologies. This project also involves planning and hosting a two-day workshop at Dartmouth in 20X that will bring Dartmouth faculty, post-docs, and grad students together with external faculty, post-docs, and grad students to discuss original papers on ethical and social issues in AI. This workshop, which will be co-organized with the Princeton Dialogues on AI and Ethics project, will provide an opportunity for Dartmouth to join peer institutions in developing academic programs focused specifically on the ethical challenges of AI and IT.


Jonathan Chipman

A Virtual-Reality Node for Interdisciplinary 3D Visualization in the Spatial Sciences


Jonathan Chipman

Geographic data visualization (geovisualization) is rapidly becoming central to the spatial sciences, due to advances in remote sensing (lidar, photogrammetry), big databases, and computational technology. Yet the field is only beginning to consider the advantages of immersive 3D, virtual reality (VR), and augmented reality (AR) for visual exploration, interpretation, and analysis of geospatial data.  There are a wide array of research and teaching applications that could be revolutionized by the adoption of VR: for example, enabling field work in inaccessible or environmentally-sensitive sites, exploring past landscapes and archaeological sites, allowing students to contribute to field research in remote areas where travel is prohibitively expensive, and facilitating the interpretation and mapping of landscape features such as glacial moraines by immersive visualization across scales.

We are developing this new "Virtual Reality Node" within Dartmouth's Citrin GIS/Applied Spatial Analysis lab, for immersive 3D visualization. This facility will include hardware (VR-enabled desktop and mobile workstations), software, and visualization equipment, for use in both teaching and research. Faculty and staff from departments across campus will be able to use the equipment in situ or borrow mobile setups for use elsewhere on campus as needed.

Computer Science & Thayer

Bo Zhu & Zi Chen

Intelligent Computational System for Automated Exploration and Identification of Design Space for Functional Mechanical Structures


Bo Zhu


Zi Chen

Functional mechanical structures, including examples of Origami, Kirigami, reconfigurable, and multi-stable structures, cover a broad range of scientific research topics and industrial applications including robotics, energy harvesting devices, and energy absorption structures. The collisions and intertwinement among the forefronts of the advancing computation and fabrication techniques are changing the way and pace scientists and engineers discover and prototype new functional structures. We propose to develop a fully automated computational approach to explore and understand functional structures. We plan to create an intelligent computational system to understand mechanical relations, explore property boundaries, and synthesize high-level functions by co-designing the geometry, topology, and material of a functional structure in a fully automated algorithmic fashion. Further, we aim to connect this computational infrastructure to the advancing manufacturing solutions to close the loop between virtual designs and physical fabrication.

Institute for Writing and Rhetoric

Christiane Donahue +

Textual Moves and the Voices of Others: Longitudinal Research on Student Source Use

Taine Donahue

Taine Donahue

Tiane Donahue, Isaac Feldman, Nick Van Kley, Sarah Smith, Annika Konrad

Student writers work frequently with written assignments that involve interacting with other texts. The strategies they use and the ways they position themselves in relation to these other voices tell us volumes about their entrance into the scholarly community. Many of the texts students post to their college digital portfolios use and interact with sources, in relation to particular text types assigned and the context of different courses. Our project analyzes the qualitative data of these texts via two quantitative methods: human coding (a social sciences research approach) and NLP-driven automated coding, in order to identify statistically significant trends and put these quantitative results in conversation with our qualitative analyses, notably case study. We will use open source AntConc, and Atlas.ti, for some features analysis, similar to work in big data and digital humanities studies, and grounded in textual corpus analysis which is best adapted to studying change over time, patterns, and comparisons across contexts.

This descriptive study will produce simple analytics (for example, frequency of different text types and textual phenomena); identify statistically significant trends in source use, text use, and use of other kinds of evidence across different text types and course types; track statistically significant change over time in student productions and approaches; and track significant correlations among various contextual factors. The study will inform national conversations about how students work with sources to establish their authority, create strong arguments, and synthesize available knowledge. It will also enable network-building with other US and international teams studying these questions.


Feng Fu

Online Influence Manipulation in Social Media Reposts: Detection, Evaluation, and Mitigation


Feng Fu

We will closely integrate computational and mathematical models of information propagation with massive scale social media repost/retweet data collected from Weibo and Twitter, study the structural and temporal pattern of reposting/retweeting dynamics driven by online influence manipulation, and provide early detection, evaluation, and mitigation strategies for controlling the spread of misinformation. We will address two foundational questions that are central to stem fake news: (i) How does the topological structure of repost networks influence the temporal pattern of reposts and vice versa? (ii) Which structure promotes the viral reposting activity, and which structure diminishes it? In answering these questions, we will discern structural and temporal features driven by online influence manipulation as well as ways to restructure the multiplex networks of social media reposts in order to hinder any potential influence manipulation. Through the lens of mathematical data science, the project will help us gain previously unattainable insights into understanding the presence and spread of misinformation in social media platforms. We will provide innovative intervention solutions that can minimize the effects of misinformation and reduce fake news. We will share our collected repost dataset and computational models that are suited for studying online influence manipulation.


Julie Hruby

Associating Fingerprint Patterns with Age and Sex: A Quantifiable Approach


Photo of Julie Hruby

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 more recently, we have also begun to ask a much wider range of questions. 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, and a Greek attorney has assisted us in complying with both European Union and Greek law as they relate to the collection of prints from juveniles. We are beginning the process of collecting and scanning the juvenile prints in the summer of 2019, and we will spend a month working at the site of ancient Corinth, scanning fingerprints from the interiors of half life-size to nearly life-sized ceramic sculpture.


Geoffery Luke

Computational Compressed Ultrafast Holography


Photo of Geoffrey Luke

Light behaves like a wave, propagating with both a magnitude and a phase. Conventional cameras, however, are only able to detect the magnitude of the light; phase information is lost. The technology of holography is able to capture both the phase and magnitude information by interfering light from the object of interest with a reference laser beam. This concept has led to many advances in imaging and three-dimensional displays. The ability to capture fast-moving processes, however, is limited by the camera technology.

Our goal in this project is to create a new ultra-fast digital holography system. The system relies on spatio-spectral phase encoding of the reference beam and mechanical sweeping of the image across the camera. The end result is that the hologram at different time points is recorded on different pixels on the camera. We then propose to apply sparsity-constrained optimization methods to reconstruct an entire holographic movie from a single camera exposure. With advanced camera technology this could result in frame rates in the range of billions of frames per second.


Bjoern Muetzel

Linear Games - games based on algorithms from Linear Algebra


Bjoern Muetzel

Only four out of ten high school students feel engaged in class (Gallup, 2015) and half of the students feel bored and tired. Especially concerned by this trend is the subject math. A very recent idea to solve this problem is the gamification of math. Following this approach we will develop several computer games that teach high school and college students math in a fun and open setting. Here we focus on Linear Algebra whose mastery is essential for almost all basic sciences.

The games are puzzle games which can be solved using methods from Linear Algebra. To focus on the algorithmic side instead of complicated calculations we replace the real numbers by a simpler number system which in these games is represented by pieces on a Go board. In playing these games children gain an intuitive understanding of the underlying algorithms from Linear Algebra and the new number system from Abstract Algebra.

The games will be accompanied by a detailed description of the mathematics behind them and its application. To reach out to the wider public and amplify the impact they will be realized as free Android and Apple Apps and presented at teaching conferences and venues. This project funded by the Neukom Institute will be realized with the help of the DALI Lab.

Psychological and Brain Sciences

Katherine Nautiyal

Neural Basis of Impulsivity


Katherine Nautiyal

Exerting self-control is a key component in many daily behaviors such as making healthy eating choices, saving for retirement, and waiting for the light to cross the street.  A limiting factor in successful self-control is impulsivity, namely reduced delay-of-gratification and lower response inhibition.  Our understanding of the biological basis of impulsivity is limited, and my research aims to contribute to a better understanding of the neural circuits that control impulsive behavior.  My lab primarily uses behavioral neuroscience techniques to manipulate and probe the neural circuits underlying impulsive behavior in mice, in order to understand how serotonin modulates self-control.  We also use in vivo calcium imaging approaches that allow large-scale visualization of neural activity at the single cell level, during freely-moving behavior. This generates large datasets of the activity of hundreds of neurons across many days, and the concurrent behavioral repertoire in tasks of impulsivity.

In this project funded by the Neukom CompX grant, we will apply computational methods to our study of the neural basis of self-control behavior to investigate how impulsive behavior is encoded at the single cell level. We will build logistic regression classifiers to discriminate behavioral states of impulsivity based on calcium imaging events in order to understand if how this encoding changes over days and in pathological states of impulsivity.  Overall, these models  will help us understand what features of the neural code contribute to self-control.

Earth Sciences

Marisa Palucis

Seeing through the forest: LiDAR drones


Marisa Palucis

Airborne Light Detection and Ranging (or LiDAR) has become a valuable tool for a wide range of research applications in geomorphology, ecology, archaeology, and other fields. This is partially because LiDAR sensors can penetrate tree canopy and other vegetation, such that the technology is uniquely capable of documenting topography and other subtle surface features that are obscured in most other forms of aerial and satellite imagery. Yet, collection of conventional aircraft-acquired LiDAR remains prohibitively expensive, severely limiting applications and impacts of this potentially transformative technology. In recent years, drones are also revolutionizing researchers' abilities to collect near-surface aerial imagery, however, drone-acquired LiDAR has remained out of reach to most scientists due to the high cost and mediocre performance of miniaturized LiDAR sensors. In the past year, the cost of drone-optimized LiDAR sensors has come down considerably, while consumer drone technologies have steadily improved, making the collection of low-cost, high-resolution LiDAR data feasible for the first time.

This funding will allow for the purchase of a drone-optimized LiDAR system that will be shared by faculty in EARS and Anthropology, and will be employed on several major projects over the next 18 months. One project is focused on understanding sediment transport processes in cold and icy environments as an analog for geomorphic processes on Mars, while another will document the response and recovery of fluvial systems to Tropical Storm Irene. The drone will also be used to explore for evidence of ancient Puebloan-period settlements and agricultural landscapes in Mesa Verde (Colorado) and document the layout and architecture of prehistoric Hawaii's largest town in Captain Cook National Monument. When not deployed in the field, this drone will be incorporated into Dartmouth's Digital Archeology (Anth 50.03) and Remote Sensing (EARS 65) courses.

Psychological and Brain Sciences

Peter Tse

Decoding Octopus Electroencephalography (EEG)*


Peter Tse

The focus of Tse's work in Psychophysics and Cognitive Neuroscience is mid- and high-level human vision. In the domain of mid-level vision his group has worked on deciphering the rapid form-motion computations that go into the construction of subsequent visual experience. For example, in the case of the apparent motion of lights jumping back and forth on top of a police car, brief, static flashes of light, toggled back and forth, give rise to the experience of analog motion jumping back and forth. We experience this 'mistake' even though, cognitively, we know that no lightbulbs are actually jumping around. Tse argues that this illusion of apparent motion occurs because a stage of inference is central to the computation of motion vectors, such that objects in one scene are matched to what is presumed to be themselves in the next scene, on the basis of assumptions about how objects in fact move about the world. What is experienced is the motion that is inferred to have taken place given these initial and end states. In Tse's view, following Gestalt Psychology, consciousness is constructed on the basis of numerous such interacting operations. These preconscious and unconscious operations may add information, such as illusory contours, surfaces and volumes, that are not present in the detected image. Consciousness is therefore an internal construction, but one that differs from hallucination in that it has evolved to be  veridical representation of what is happening. This construction is not passive. In fact, we can volitionally alter our consciousness by, for example, by choosing to look next at this versus that, or to attend to this versus that, for our own reasons.

       Since 2016, Tse has begun to study octopus cognition. This began with this epiphany: "I asked myself 'what is the most alien nervous system on the planet capable of sophisticated visual processing?' The answer was the octopus, with its two brains, one bilateral and the other shaped like a ring around it. If we could understand how an 'alien' brain and our mammalian brain solved similar complex computational problems, whether at the algorithmic or neural level, we could learn a lot about potential universals of computation in brains." Regarding the "alien" nature of the octopus brain, note that our last common ancestor was likely the sub-1mm long, bilateral seaworm 'Urbilateria' that is hypothesized to have lived in the sands of the warm shallow seas of the pre-Cambrian, perhaps similar to the modern 'living fossil' seaworm platynereis dumerlii. Given the shared genetic inheritance from this ancestor found in vertebrates, arthropods and cephalopods, Urbilateria likely had photosensitive simple eyes and centralized ganglia for processing sensory input. Since our divergence, the lineages that led to the octopus and to us have convergently evolved almost identical camera-like eyes. In addition, both lineages have convergently evolved the capacity to process visual input up to a level of 3-D shape recognition; We know this must be true of the octopus because many species can mimic the shapes of coral on their skin, or even mimic whole animals. How deep does octopus and vertebrate convergent evolution go? What kinds of cognition are they capable of using their 500 million neurons, which, incidentally is approximately as many as in the cortex of a bear or lion? The new Dartmouth Octopus Cognition Lab hopes to answer such questions experimentally.

Psychological and Brain Sciences & Tuck

Thalia Wheatley, Adam Kleinbaum, & Adrienne Wood

Predicting future social network position from the interactions of strangers: An applied computer vision project


Thalia Wheatley


Adam Kleinbaum


Adrienne Wood

Social connectedness has important implications for a variety of domains of life. Being well-integrated in a social network is positively associated with academic and professional success; people who build broad and diverse networks enjoy myriad advantages in their lives and careers. However, little is known about the critical early stages of network formation. What does successful networking look like at a behavioral level? Studying how social connections form will yield insights into what kinds of behavioral interventions might help people avoid social isolation and become better-connected.

We aim to uncover observable behaviors that predict later social connectedness. With the help of the CompX grant, we will develop machine learning-based computer vision tools to track the movements of people attending a professional social mixer. Attendees’ movement and affiliation patterns will then be used to predict their positions in the social network several months later.

The contributions of the proposed research and development are two-fold: first, we will gain important insights into the behaviors that put people on a path to social integration versus isolation. This work may also inform interventions that could be implemented in any newly-formed community, from first-year undergraduates to working professionals to residential care facilities. Second, the award will support the development of video-based movement tracking software and pipeline that will be made freely available on Github and other open source software platforms, enabling other researchers to conduct naturalistic, large-scale behavioral studies.