arXiv subjects

John Dickerson

Publications and source records attributed to John Dickerson.

22 records · Page 2Linked to original sources

arXiv2020

Forming Diverse Teams from Sequentially Arriving People

Collaborative work often benefits from having teams or organizations with heterogeneous members. In this paper, we present a method to form such diverse teams from people arriving sequentially over time. We define a monotone submodular objective function that combines the diversity and quality of a team and propose an algorithm to maximize the objective while satisfying multiple constraints. This allows us to balance both how diverse the team is and how well it can perform the task at hand. Using crowd experiments, we show that, in practice, the algorithm leads to large gains in team diversity. Using simulations, we show how to quantify the additional cost of forming diverse teams and how to address the problem of simultaneously maximizing diversity for several attributes (e.g., country of origin, gender). Our method has applications in collaborative work ranging from team formation, the assignment of workers to teams in crowdsourcing, and reviewer allocation to journal papers arriving sequentially. Our code is publicly accessible for further research.

cs.AI↗

arXiv2019

Universal Adversarial Training

Standard adversarial attacks change the predicted class label of a selected image by adding specially tailored small perturbations to its pixels. In contrast, a universal perturbation is an update that can be added to any image in a broad class of images, while still changing the predicted class label. We study the efficient generation of universal adversarial perturbations, and also efficient methods for hardening networks to these attacks. We propose a simple optimization-based universal attack that reduces the top-1 accuracy of various network architectures on ImageNet to less than 20%, while learning the universal perturbation 13X faster than the standard method. To defend against these perturbations, we propose universal adversarial training, which models the problem of robust classifier generation as a two-player min-max game, and produces robust models with only 2X the cost of natural training. We also propose a simultaneous stochastic gradient method that is almost free of extra computation, which allows us to do universal adversarial training on ImageNet.

cs.CV↗

arXiv2019

Adversarial Training for Free!

Adversarial training, in which a network is trained on adversarial examples, is one of the few defenses against adversarial attacks that withstands strong attacks. Unfortunately, the high cost of generating strong adversarial examples makes standard adversarial training impractical on large-scale problems like ImageNet. We present an algorithm that eliminates the overhead cost of generating adversarial examples by recycling the gradient information computed when updating model parameters. Our "free" adversarial training algorithm achieves comparable robustness to PGD adversarial training on the CIFAR-10 and CIFAR-100 datasets at negligible additional cost compared to natural training, and can be 7 to 30 times faster than other strong adversarial training methods. Using a single workstation with 4 P100 GPUs and 2 days of runtime, we can train a robust model for the large-scale ImageNet classification task that maintains 40% accuracy against PGD attacks. The code is available at https://github.com/ashafahi/free_adv_train.

cs.LG↗

arXiv2019

The Muffin Problem

You have $m$ muffins and $s$ students. You want to divide the muffins into pieces and give the shares to students such that every student has $\frac{m}{s}$ muffins. Find a divide-and-distribute protocol that maximizes the minimum piece. Let $f(m,s)$ be the minimum piece in the optimal protocol. We prove that $f(m,s)$ exists, is rational, and finding it is computable (though possibly difficult). We show that $f(m,s)$ can be derived from $f(s,m)$; hence we need only consider $m\ge s$. We give a function $FC(m,s)$ such that, for $m\ge s+1$, $f(m,s)\le FC(m,s)$. It is often the case that $f(m,s)=FC(m,s)$. More formally, for all $s$, for all but a finite number of $m$, $f(m,s)=FC(m,s)$. This leads to a nice formula for $f(m,s)$, though there are exceptions to it. We give a formula $INT(m,s)$, which has 6 parts, such that for many of the exceptional $m$, $f(m,s)=INT(m,s)<FC(m,s)$. This works for most of the exceptional $m$ where ceil${2m/s}\ge 4$. There are still some exceptional $m$ with ceil${2m/s}=3$ (if its $\le 2$ then the problem is trivial). For these cases we have a way to {\it generate theorems}. For $1\le d\le 7$ we have generated formulas for $f(s+d,s)$. We do not have a theorem here but we do have a methodology which leads to, for some of the $m$, a value $BM(m,s)$ such that often $f(m,s)=BM(m,s)<INT(m,s)<FC(m,s)$. So far it seems like, for $m\ge s$, $f(m,s) = \min\{FC(m,s), INT(m,s), BM(m,s) \}$, though we have not prove this. For $1\le s\le 50$ and $s\le m\le 60$ we have obtained $f(m,s)$ for all but 20 values.

math.CO↗