Valerie Chen

I'm a first year PhD student in the Machine Learning Department at Carnegie Mellon University advised by Ameet Talwalkar. My PhD is supported by the NSF Graduate Research Fellowship.

I completed my BS in Computer Science at Yale University, where I previously worked with Zhong Shao and Abhinav Gupta. I have also spent time at IBM Research and the Naval Research Laboratory.

I am interested in principled algorithmic development of interpretable machine learning for human-AI value alignment as well as in the issues of fairness and accountability of AI systems.

Email / Google Scholar / LinkedIn / Github

profile photo
Recent News

- Graduated from Yale and awarded the Henry Prentiss Becton Prize, the highest award in SEAS.

- Awarded 2020 NSF Graduate Research Fellowship

- 1 paper accepted to 2020 ICRA

Research
Ask Your Humans: Using Human Instructions to Improve Generalization in Reinforcement Learning
Valerie Chen, Kenny Marino, Abhinav Gupta
Robotics Institute Summer Scholar Poster Presentation, 2019
Submitted to NeurIPS, 2020

We introduce a dataset of human demonstrations in a crafting-based grid world. Our model consists of a high-level language generator and low-level policy, conditioned on language. We find that human demonstrations are required to solve the most complex steps. We also find that incorporating natural language is critical in allowing the model to generalize to unseen tasks in a zero-shot approach and to learn quickly from a few demonstrations. Our approach also gives our trained agent interpretable behaviors because it is able to generate a sequence of high-level descriptions of its actions.

Task-Aware Novelty Detection for Visual-based Deep Learning in Autonomous Systems
Valerie Chen, Man-Ki Yoon, Zhong Shao
Accepted to ICRA, 2020

I extended my prior approach to an end-to-end method for novelty detection, which simultaneously trains the visual prediction module along with the novelty detection network. The method is robust not only to out-of-distribution inputs, but also to adversarial attacks on the training images.

Explainable Deep Learning for Visual-based Safety-critical Systems
Valerie Chen, Man-Ki Yoon, Zhong Shao
DSN DSML Workshop, 2019

We present a method for detecting out of distribution novel scenarios in vision based autonomous driving via analysis of network saliency. We train an autoencoder to learn a representation of the generated saliency maps from the training distribution to detect anomalies in test images.

Video-Text Compliance: Activity Verification based on Natural Language Instructions
Mayoore Jaiswal, Frank Liu, Anupama Jagannathan, Anne Gattiker, Inseok Hwang, Jinho Lee, Matt Tong, Sahil Dureja, Soham Shah, Peter Hofstee, Valerie Chen, Suvadip Paul, Rogerio Feris
Spotlight Paper at ICCV Closing the Loop Between Vision and Language Workshop, 2019
Oral Presentation at ICCV Large Scale Holistic Video Understanding Workshop
, 2019

We define a new multi-modal compliance problem and ComplianceNet, a novel end-to-end trainable network to solve the video-text compliance task.

Secure Computation for Machine Learning With SPDZ
Valerie Chen, Valerio Pastro, Mariana Raykova
NeurIPS PPML Workshop, 2018
1st Place at ACM Student Research Competition at Grace Hopper Conference, 2018

We investigate the efficiency of the SPDZ framework in the context of ML algorithms and demonstrate that the SPDZ framework outperforms these previous implementations while providing stronger security.


Integrating a formal requirements modeling simulator and an autonomy software simulator to validate the behavior of unmanned vehicles
Elizabeth Leonard, Constance Heitmeyer, Valerie Chen
Spring Simulation Multi-conference, 2015

We introduce the integration of the SCR requirements simulator with the eBotworks 3D simulator for autonomy software, illustrating the utility of the combined simulation framework by applying it to validate the requirements of an unmanned ground vehicle.

Teaching
CPSC 201 Intro CS, Fall 2017
CPSC 470 Artificial Intelligence, Spring 2019 and 2020
Computer Science Department Peer Mentor, 2019-2020
Extra Curricular
Lead student organizer of 1st AI, Ethics, and Society @ Yale workshop
Student volunteer at Computation and Society at Yale
Former president of the Yale Women's Leadership Initiative, and conference director of the 10th Annual Women Empowering Women Conference
Former co-president of SheCode, an outreach organization to teach New Haven middle and high school girls to code