Sponsors
- Amazon Robotics
- Bulger Partners
Judges
- Rob Gurwitz, Former CTO and SVP of Technology & Architecture, Bulger Partners
- Robert W. Martin, Web Development Manager & Accessibility Lead, State of Rhode Island Department of Administration
- Ugur Cetintemel, Professor and Department Chair, Brown University Computer Science
- James Tompkin, Assistant Professor, Brown University Computer Science
Entrants
-
ReNeg for Behaviorally Cloning Autonomous Vehicles – by Jacob Beck
We propose a method allows for better and faster autonomous vehicle learning: regression with the full range of positive AND negative behavior. -
Increasing Retention Rates of Undergraduates in STEM – by Victoria Chávez
This research investigates and addresses the disproportionate underrepresentation of minorities in STEM concentrations and suggests different actionable steps that can be taken at various levels of the university. -
STOP (Stackelberg Trained Optimal Policies) Car Bullying – by Matt Cooper
We employ an adaptive policy computed from the Stackelberg Equilibrium of a two-player game that enables autonomous cars to safely assert themselves in scenarios requiring social negotiation with a human driver. -
Improving Shape Deformation in Unsupervised Image-to-Image Translation – by Aaron Gokaslan
We develop a network that morphs objects in the image to the required domain with minimal background and scene feature distortion. -
Criteria Sliders Web – by Mae Heitmann
Criteria Sliders organizes images based on arbitrary ranking criteria from the user in a useful way. -
ShareAid: The Personalized Photo Sharing Assistant – by Philip Hinch
ShareAid investigates if a machine can automatically learn and imitate the photo-sharing process by learning on a user’s previous social media behavior. -
Inferring the Intentions of Learning Agents – by Vince Kubala
This work addresses the problem of inferring the goals of an intelligent agent who is learning in a known way. It generalizes the standard inverse reinforcement learning problem by assuming instead that the agent may learn in a known way and makes decisions in a known, albeit not necessarily optimal, way. -
Dash – by Luke Murray
Dash’s vision is to be a context-aware, note taking application for individuals and small workgroups that facilitates summarizing and digesting information through a streamlined yet highly customizable UI. -
MS Lesion Segmentation with Cascaded 3D Convolutional Neural Network – by Charles Njoroge
I present a reimplementation of automated multiple sclerosis lesion segmentation with a cascaded 3d convolutional neural network which utilizes a cascaded 3d convolutional neural network to effectively segment the MICCAI2008 public dataset. -
Deep Abstract Q-Networks – by Melrose Roderick
We examine the problem of learning/planning on high-dimensional reinforcement learning domains with long horizons & sparse rewards. -
Virtual Reality Teleoperation Task Feasibility – by Eric Rosen
This work poses the question of what robot tasks are feasible under a commercially VR teleoperation system like ROS Reality. -
Driving as Fast as Possible: End to End Learning Obstacle Detection and Avoidance – by Josh Roy
In this project, we hope to evaluate the performance of an end-to-end neural network to perform this avoidance and navigation using a small, two-wheeled, robot which the neural network will learn to control. -
Applying Rademacher-Like Bounds to Combinatorial Samples and Function Selection – by Clayton Sanford
We applied our framework (a variant of the Rademacher Complexity) and Rademacher complexity to incorporate bounds into learning algorithms in the audio domain. -
ProSecCo: Progressive Sequence Mining with Convergence Guarantees – by Sacha Servan-Schreiber
This work extends existing data mining techniques into a progressive algorithm that keeps interactivity by outputting incremental results very frequently. -
Rewind: Automatically Generated Location-based Videos for Revisiting Personal Memories – by Neilly Tan
Rewind is a system that automatically recreates a digital memory of a past trip by taking historical geolocation data and generating a stitched-together cinematic-style video made of streetside images from where a user has been. -
An Optical MIDI Interpreter for Acoustic Guitar – by Sam Title
The optical pickup explored in this paper was a viable design that can serve as an alternative to hexaphonic guitar pickups for detecting note events. -
DBPal: An End-to-end Neural Natural Language Interface for Databases – by Nathaniel Weir
DBPal is an intelligent system that makes databases more accessible and expressive requiring extensive background knowledge. -
SleepCoacher 2.0: A Personalized Automated Self-Experimentation System for Sleep Recommendations – by Jina Yoon
SleepCoacher is an Android app that helps people figure out how to achieve sleep perfection by coaching them through sleep "experiments" and analyzing their data.