Chad DeChant

Twitter:

PhD candidate, Computer Science Department

Columbia University

MA, Government, Institute for International Law and Politics

Georgetown University

BA, Philosophy

Georgetown University

I am a PhD candidate at Columbia working at the intersection of natural language processing and robotics, advised by Daniel Bauer. My goal is to apply cognitive science research to make AI safe, reliable, and interpretable.

My thesis work is on enabling robots to remember and report on their past actions in natural language through summaries and question answering. I believe that this ability is an essential one both for training robots and other agents to better understand their environments and for ensuring safe and understandable agent behavior.

I have also worked on artificial metacognition (i.e. thinking about thinking, especially knowing what we know), learning word representations in the order a child learns words, improving robotic grasping techniques by learning to 'imagine' parts of objects that are hidden, and detecting crop disease from field imagery.

I created and taught a course for Columbia undergraduates on AI Safety, Ethics, and Policy.

Selected Publications

Episodic Memory Verbalization using Hierarchical Representations of Life-Long Robot Experience Leonard Bärmann, Chad DeChant, Joana Plewnia, Fabian Peller-Konrad, Daniel Bauer, Tamim Asfour, and Alex Waibel arXiv preprint 2024 Paper
In search of the embgram: forming episodic representations in a deep learning model Chad DeChant, Iretiayo Akinola, and Daniel Bauer Cognitive Computational Neuroscience 2024 Paper
Learning to Summarize and Answer Questions about a Virtual Robot's Past Actions Chad DeChant, Iretiayo Akinola, and Daniel Bauer Autonomous Robots Special issue on LLMs and Robotics, November 2023 Paper Code
On the risks and benefits of episodic memory in AI agents Chad DeChant Safe and Trustworthy AI Workshop at ICLP 2023 Paper
Are robots stuck in time? Learning to represent actions in long horizon planning by learning to remember the past Chad DeChant Learning, Perception, and Abstraction for Long-Horizon Planning Workshop at the Conference on Robot Learning 2022 Paper
RoboReport: Summarizing a virtual robot's past actions in natural language Chad DeChant, Iretiayo Akinola, and Daniel Bauer Language and Robot Learning Workshop at the Conference on Robot Learning 2022 (Spotlight talk) Paper
Do you see what I see? Using questions and answers to align representations of robotic actions Chad DeChant, Iretiayo Akinola, and Daniel Bauer Aligning Robot Representations with Humans Workshop at the Conference on Robot Learning 2022 Paper
Toward robots that learn to summarize their actions in natural language: a set of tasks Chad DeChant and Daniel Bauer Conference on Robot Learning 2021 Paper and OpenReview
Learning word representations in a developmentally realistic order Chad DeChant and Daniel Bauer 2021 Paper
Predicting the accuracy of neural networks from final and intermediate layer outputs Chad DeChant, Seungwook Han, and Hod Lipson Identifying and Understanding Deep Learning Phenomena workshop at ICML 2019 Paper
Learning to make accuracy judgments from intermediate and final outputs of a neural network for a question answering task Chad DeChant, Seungwook Han, and Hod Lipson BlackboxNLP at ACL 2019
Automated identification of northern leaf blight-infected maize plants from field imagery using deep learning Chad DeChant, Tyr Wiesner-Hanks, Siyuan Chen, Ethan Stewart, Jason Yosinski, Michael Gore, Rebecca Nelson, and Hod Lipson Phytopathology 2017 Paper
Shape completion enabled robotic grasping Jacob Varley, Chad DeChant, Adam Richardson, Joaquín Ruales, Peter Allen Intelligent Robots and Systems (IROS) 2017 Paper Video

Other Publications

Creating a Tool to Reproducibly Estimate the Ethical Impact of Artificial Intelligence Sara Jordan, Sina Fazelpour, Adriano Koshiyama, Jaky Kueper, Chad DeChant, Brenda Leong, Gary Marchant, Craig Shank Paper 2019
Could AI Drive Transformative Social Progress? What Would This Require? Ted A Parson, Robert Lempert, Ben Armstrong, Evan Crothers, Chad DeChant, Nicholas Novelli Paper 2019
*What can Natural Language Processing learn from the brain? Intended as a quick guide to the brain's processing of language for NLP researchers along with some of my ideas for how such a perspective might influence future NLP research. Originally a paper and annotated bibliography I wrote for a class called Computation and the Brain, it needs more work. Chad DeChant DRAFT Paper 2018
AI, modularity, and the global workspace Suggestions for applying ideas of modularity and the global workspace to AI. Written for a Psychology seminar. Chad DeChant Paper 2017
Autonomous Detection of Plant Disease Symptoms Directly from Aerial Imagery Harvey Wu, Tyr Wiesner-Hanks, Ethan L Stewart, Chad DeChant, Nicholas Kaczmar, Michael A Gore, Rebecca J Nelson, Hod Lipson The Plant Phenome Journal 2019 Paper
Quantitative Phenotyping of Northern Leaf Blight in UAV Images Using Deep Learning Ethan L Stewart, Tyr Wiesner-Hanks, Nicholas Kaczmar, Chad DeChant, Harvey Wu, Hod Lipson, Rebecca J Nelson, Michael A Gore Remote Sensing 2019 Paper
Millimeter-level plant disease detection from aerial photographs via deep learning and crowdsourced data Tyr Wiesner-Hanks, Harvey Wu, Ethan Stewart, Chad DeChant, Nicholas Kaczmar, Hod Lipson, Michael A Gore, Rebecca J Nelson Frontiers in Plant Science 2019 Paper
Image set for deep learning: field images of maize annotated with disease symptoms Tyr Wiesner-Hanks, Ethan Stewart, Nicholas Kaczmar, Chad DeChant, Harvey Wu, Rebecca Nelson, Hod Lipson, and Michael Gore BMC Research Notes 2018 Paper
Automated Weed Detection in Aerial Imagery with Context Delia Bullock, Andrew Mangeni, Tyr Wiesner-Hanks, Chad DeChant, Ethan L Stewart, Nicholas Kaczmar, Rebecca J Nelson, Michael A Gore, Hod Lipson Paper 2019

Chad DeChant

Selected Publications

Other Publications

Teaching

Other News

Volunteering

Resources