Ruth Fong

I am a teaching faculty member in the Department of Computer Science at Princeton University, where I teach introductory CS courses and conduct research in computer vision and machine learning with a focus on explainable AI. At Princeton, I lead the Looking Glass Lab as well as collaborate with Professor Olga Russakovsky and the Visual AI Lab.

I completed my PhD in the Visual Geometry Group at the University of Oxford, where I was advised by Andrea Vedaldi and funded by the Rhodes Trust and Open Philanthropy. Also at Oxford, I earned a Masters in Neuroscience, where I worked with Rafal Bogacz, Ben Willmore, and Nicol Harper. I received a Bachelors in Computer Science at Harvard University, where I worked with worked with David Cox and Walter Scheirer.

Email  /  CV  /  Bio  /  Google Scholar  /  GitHub

Hello! 👋

I am excited to work with Princeton students! If you are a...

  • Princeton undergraduate student interested in doing an IW, a senior thesis, and/or research with me, please first read my preferred list of IW topics. If one of the topics interests you, please email me your (1) CV, (2) transcript, and (3) a short description of what you might be interested in working on (I also have a few project ideas I can suggest). I recommend you first (or currently) take COS429 (or at least COS324) before working with me. I am particularly excited about helping students from underrepresented groups have a positive and successful first research experience.
  • Princeton graduate student interested in working or collaborating with me, please reach out via email.
  • non-Princeton student (including prospective applicants) interested in working with me, there's no need to email; I unfortunately am not accepting non-Princeton students at this time.
News 🗞
  • Along with my co-instructors, I introduced an open-ended final project to COS126 (i.e. Princeton's intro CS course), here's the online gallery of the amazing projects students created!
  • I am excited to announce that I am joining Princeton's CS department as a teaching faculty member starting July 2021.
  • My PhD thesis on "Understanding Convolutional Neural Networks" can be found here. For those with less experience, all chapters except chapters 3-6 were written with accessibility in mind.
  • We have a new report out on "Toward Trustworthy AI Development: Mechanisms for Supporting Verifiable Claims." webpage | arXiv
  • We just released TorchRay, a PyTorch interpretability library. In the initial release, we focus on attribution and re-implemented popular methods and benchmarks to encourage reproducible research. Resources: tutorial slides | colab notebook
  • Slides for VGG tutorial on interpretability that I gave recently can be found here.
Looking Glass Lab 👤


Creston Brooks Indu Panigrahi Alexis Sursock Frelicia Tucker Devon Ulrich Dhopte Vedant


Nicole Meister Sunnie S.Y. Kim Vikram Ramaswamy Angelina Wang Prof. Olga Russakovsky

[image attribution]

Research 🧪

My research interests are in computer vision, machine learning, deep learning, and explainable AI. I am interested in developing novel techniques for understanding AI models post-hoc, designing new deep learning archiectures that are interpretable-by-design, and introducing paradigms for finding and correcting existing failure points in AI models. Representative papers are highlighted; see Google Scholar for the most updated list of papers.

HIVE image HIVE: Evaluating the Human Interpretability of Visual Explanations
Sunnie S.Y. Kim, Nicole Meister, Vikram V. Ramaswamy, Ruth Fong, Olga Russakovsky
arXiv, 2021 
arXiv | project page | code | bibtex

We introduce HIVE, a novel human evaluation framework for diverse interpretability methods in computer vision, and develop metrics that measure achievement on two desiderata for explanations used to assist human decision making: (1) Explanations should allow users to distinguish between correct and incorrect predictions. (2) Explanations should be understandable to users.

interactive similarity example #2
interactive similarity example #1
Interactive Similarity Overlays
Ruth Fong, Alexander Mordvintsev, Andrea Vedaldi, Chris Olah
VISxAI, 2021 
interactive article | code | bibtex

We introduce a novel interactive visualization that allows machine learning practitioners and researchers to easily observe, explore, and compare how a neural network perceives different image regions.

hierarchical transformations On Compositions of Transformations in Contrastive Self-Supervised Learning
Mandela Patrick*, Yuki M. Asano*, Polina Kuznetsova, Ruth Fong, João F. Henriques, Geoffrey Zweig, and Andrea Vedaldi
ICCV, 2021 
arXiv | code | bibtex

We give transformations the prominence they deserve by introducing a systematic framework suitable for contrastive learning. SOTA video representation learning by learning (in)variances systematically.

describability metric diagram
describability metric diagram
Quantifying Learnability and Describability of Visual Concepts Emerging in Representation Learning
Iro Laina, Ruth Fong, and Andrea Vedaldi
NeurIPS, 2020
arxiv | supp | bibtex

We introduce two novel human evaluation metrics for quantifying for evaluating the interpretability of clusters discovered via self-supervised methods. We also outline how to partially approximate one of the metrics using a group captioning model.

Debiasing Convolutional Neural Networks via Meta Orthogonalization
Kurtis Evan David, Qiang Liu, and Ruth Fong
NeurIPS Workshop on Algorithmic Fairness through the Lens of Causality and Interpretability (AFCI), 2020
arxiv | supp | poster | bibtex

We introduce a novel paradigm for debiasing CNNs by encouraging salient concept vectors to orthogonal to class vectors in the activation space of an intermediate CNN layer (e.g., orthogonalizing gender and oven concepts in conv5).

Contextual Semantic Interpretability
Diego Marcos, Ruth Fong, Sylvain Lobry, Rémi Flamary, Nicolas Courty, and Devis Tuia
ACCV, 2020
arxiv | supp | code | bibtex

We introduce an interpretable-by-design machine vision model that learns to sparse groupings of interpretable concepts and demonstrate the utility of our novel architecture on scenicness prediction.

There and Back Again: Revisiting Backpropagation Saliency Methods
Sylvestre-Alvise Rebuffi*, Ruth Fong*, Xu Ji*, and Andrea Vedaldi
CVPR, 2020
arxiv | code | bibtex

We outline a novel framework that unifies many backpropagation saliency methods. Furthermore, we introduce NormGrad, a saliency method that considers the spatial contribution of the gradients of convolutional weights. We also systematically study the effects of combining saliency maps at different layers. Finally, we introduce a class-sensitivity metric and a meta-learning inspired technique that can be applied to any saliency method to improve class sensitivity.

Toward Trustworthy AI Development: Mechanisms for Supporting Verifiable Claims
Miles Brundage*, Shahar Avin*, Jasmine Wang*, Haydn Belfield*, Gretchen Krueger*, … ,
Ruth Fong, et al.
arXiv, 2020
arxiv | project page | bibtex

This report suggests various steps that different stakeholders can take to make it easier to verify claims made about AI systems and their associated development processes. The authors believe the implementation of such mechanisms can help make progress on one component of the multifaceted problem of ensuring that AI development is conducted in a trustworthy fashion.

Understanding Deep Networks via Extremal Perturbations and Smooth Masks
Ruth Fong*, Mandela Patrick*, and Andrea Vedaldi
ICCV, 2019 (Oral)
arxiv | supp | poster | code (TorchRay) | bibtex

We introduce extremal perturbations, an novel attribution method that highlights "where" a model is "looking." We improve upon Fong and Vedaldi, 2017 by separating out regularization on the size and smoothness of a perturbation mask from the attribution objective of learning a mask that maximally affects a model's output; we also extend our work to intermediate channel representations.

Occlusions for Effective Data Augmentation in Image Classification
Ruth Fong and Andrea Vedaldi
ICCV Workshop on Interpreting and Explaining Visual Artificial Intelligence Models, 2019
paper | bibtex | code (coming soon)

We introduce a simple paradigm based on batch augmentation for leveraging input-level occlusions (both stochastic and saliency-based) to improve ImageNet image classification. We also demonstrate the necessary of batch augmentation and quantify the robustness of different CNN architectures to occlusion via ablation studies.

Net2Vec: Quantifying and Explaining how Concepts are Encoded by Filters in Deep Neural Networks
Ruth Fong and Andrea Vedaldi
CVPR, 2018 (Spotlight)
arxiv | supp | bibtex | code | talk | slides

Investigating how human-interpretable visual concepts (i.e., textures, objects, etc.) are encoded across hidden units of a convolutional neural network (CNN) layer as well as across CNN layers.

Using Human Brain Activity to Guide Machine Learning
Ruth Fong, Walter Scheirer, and David Cox
Scientific Reports, 2018
arxiv | supp | Harvard thesis | bibtex

We introduce a biologically-informed machine learning paradigm for object classification that biases models to better match the learned, internal representations of the visual cortex.

Interpretable Explanations of Black Box Algorithms by Meaningful Perturbation
Ruth Fong and Andrea Vedaldi
ICCV, 2017
arxiv | supp | bibtex | code | book chapter (extended) | chapter bibtex

We developed a theoretical framework for learning "explanations" of black box functions like CNNs as well as saliency methods for identifying "where" a computer vision algorithm is looking.

Theses 🎓

Understanding Convolutional Neural Networks
Ruth Fong (advised by Andrea Vedaldi)
Ph.D. Thesis

This is a "thesis-by-staples", so the novel parts are the non-paper chapters (i.e. all chapters except chapters 3-6), which I wrote with accessibility in mind (e.g., the ideal reader is a motivated undergraduate or graduate student looking to learn more about deep learning and interpretability). The introduction is accessible to a high-school student, and appendices A and B are primers on the relevant math concepts and convolutional neural networks respectively.

Modelling Blind Single Channel Sound Separation Using Predict Neural Networks
Ruth Fong (advised by Ben Willmore and Nicol Harper)
M.Sc. Thesis #2

I developed an unsupervised learning paradigm for sound separation using fully connected and recurrent neural networks to predict the future from past cochleagram data.

Optimizing Deep Brain Stimulation to Dampen Tremor
Ruth Fong (advised by Rafal Bogacz)
M.Sc. Thesis #1
Tutorial | Demo | MATLAB Rayleigh statistics toolbox

I developed a computational oscillator model that modeled the tremor-dampening effects of phasic deep brain stimulation and analyzed it on experimental data.

Leveraging Human Brain Activity to Improve Object Classification
Ruth Fong (advised by David Cox and Walter Scheirer)
A.B. Thesis

Published as Fong et al., Scientific Reports 2018.

Teaching 📝

Princeton COS126: CS: An Interdisciplinary Approach — Fall 2021 & Spring 2022
Oxford Engineering B14: Image and Signal Analysis — Fall 2019
NJ Governor's School: Mathematics in the World — Summer 2015
Harvard CS121: Intro to Theory of Computation — Fall 2014
Harvard CS20: Intro to Discrete Math — Spring 2014
Harvard CS50: Intro to CS I — Fall 2012

This ubiquitous CS researcher website template spawned from here.
Last updated: March 2022