Generative models have rapidly evolved to generate realistic outputs. However, their synthetic outputs increasingly challenge the clear distinction between natural and AI-generated content, necessitating robust watermarking techniques. Watermarks are typically expected to preserve the integrity of the target image, withstand removal attempts, and prevent unauthorized replication onto unrelated images. To address this need, recent methods embed persistent watermarks into images produced by diffusion models using the initial noise. Yet, to do so, they either distort the distribution of generated images or rely on searching through a long dictionary of used keys for detection. In this paper, we propose a novel watermarking method that embeds semantic information about the generated image directly into the watermark, enabling a distortion-free watermark that can be verified without requiring a database of key patterns. Instead, the key pattern can be inferred from the semantic embedding of the image using locality-sensitive hashing. Furthermore, conditioning the watermark detection on the original image content improves robustness against forgery attacks. To demonstrate that, we consider two largely overlooked attack strategies: (i) an attacker extracting the initial noise and generating a novel image with the same pattern; (ii) an attacker inserting an unrelated (potentially harmful) object into a watermarked image, possibly while preserving the watermark. We empirically validate our method's increased robustness to these attacks. Taken together, our results suggest that content-aware watermarks can mitigate risks arising from image-generative models.

As the quality of image generators continues to improve, deepfakes become a topic of considerable societal debate. Image watermarking allows responsible model owners to detect and label their AI-generated content, which can mitigate the harm. Yet, current state-of-the-art methods in image watermarking remain vulnerable to forgery and removal attacks. This vulnerability occurs in part because watermarks distort the distribution of generated images, unintentionally revealing information about the watermarking techniques. In this work, we first demonstrate a distortion-free watermarking method for images, based on a diffusion model's initial noise. However, detecting the watermark requires comparing the initial noise reconstructed for an image to all previously used initial noises. To mitigate these issues, we propose a two-stage watermarking framework for efficient detection. During generation, we augment the initial noise with generated Fourier patterns to embed information about the group of initial noises we used. For detection, we (i) retrieve the relevant group of noises, and (ii) search within the given group for an initial noise that might match our image. This watermarking approach achieves state-of-the-art robustness to forgery and removal against a large battery of attacks. The project code is available at https://github.com/Kasraarabi/Hidden-in-the-Noise.

Originally introduced in game theory, Shapley values have emerged as a central tool in explainable machine learning, where they are used to attribute model predictions to specific input features. However, computing Shapley values exactly is expensive: for a model with $n$ features, $O(2^n)$ model evaluations are necessary. To address this issue, approximation algorithms are widely used. One of the most popular is the Kernel SHAP algorithm, which is model agnostic and remarkably effective in practice. However, to the best of our knowledge, Kernel SHAP has no strong non-asymptotic complexity guarantees. We address this issue by introducing *Leverage SHAP*, a light-weight modification of Kernel SHAP that provides provably accurate Shapley value estimates with just $O(n\log n)$ model evaluations. Our approach takes advantage of a connection between Shapley value estimation and agnostic active learning by employing *leverage score sampling*, a powerful regression tool. Beyond theoretical guarantees, we find that Leverage SHAP achieves an approximately 50% reduction in error compared to the highly optimized implementation of Kernel SHAP in the widely used SHAP library [Lundberg & Lee, 2017].

With origins in game-theory, probabilistic values like Shapley values, Banzhaf values, and semi-values have emerged as a central tool in explainable AI. They are used for feature attribution, data attribution, data valuation, and more. Since all of these values require exponential time to compute exactly, research has focused on efficient approximation methods using two techniques: Monte Carlo sampling and linear regression formulations. In this work, we present a new way of combining both of these techniques. Our approach is more flexible than prior algorithms, allowing for linear regression to be replaced with any function family whose probabilistic values can be computed efficiently. This allows us to harness the accuracy of tree-based models like XGBoost, while still producing unbiased estimates. From experiments across eight datasets, we find that our methods give state-of-the-art performance for estimating probabilistic values. For Shapley values, the error of our methods is up to $6\times$ lower than Permutation SHAP (the most popular Monte Carlo method), $2.75\times$ lower than Kernel SHAP (the most popular linear regression method), and $1.75\times$ lower than Leverage SHAP (the prior state-of-the-art Shapley value estimator). For more general probabilistic values, we can obtain error up to $60\times$ lower than prior work.

The open streets initiative "opens" streets to pedestrians and bicyclists by closing them to cars and trucks. The initiative, adopted by many cities across North America, increases community space in urban environments. But could open streets also make cities safer and less congested? We study this question by framing the choice of which streets to open as a reinforcement learning problem. In order to simulate the impact of opening streets, we first compare models for predicting vehicle collisions given network and temporal data. We find that a recurrent graph neural network, leveraging the graph structure and the short-term temporal dependence of the data, gives the best predictive performance. Then, with the ability to simulate collisions and traffic, we frame a reinforcement learning problem to find which streets to open. We compare the streets in the open streets initiative to those proposed by a Q-learning algorithm. We find that the streets proposed by the Q-learning algorithm have reliably better outcomes, while streets already selected by the open streets initiative have similar outcomes to randomly selected streets. We present our work as a step toward principally choosing which streets to open for safer and less congested cities.

Estimating the effect of treatments from natural experiments, where treatments are pre-assigned, is an important and well-studied problem. We introduce a novel natural experiment dataset obtained from an early childhood literacy nonprofit. Surprisingly, applying over 20 established estimators to the dataset produces inconsistent results in evaluating the nonprofits efficacy. To address this, we create a benchmark to evaluate estimator accuracy using synthetic outcomes, whose design was guided by domain experts. The benchmark extensively explores performance as real world conditions like sample size, treatment correlation, and propensity score accuracy vary. Based on our benchmark, we observe that the class of doubly robust treatment effect estimators, which are based on simple and intuitive regression adjustment, generally outperform other more complicated estimators by orders of magnitude. To better support our theoretical understanding of doubly robust estimators, we derive a closed form expression for the variance of any such estimator that uses dataset splitting to obtain an unbiased estimate. This expression motivates the design of a new doubly robust estimator that uses a novel loss function when fitting functions for regression adjustment. We release the dataset and benchmark in a Python package; the package is built in a modular way to facilitate new datasets and estimators. https://github.com/rtealwitter/naturalexperiments

Making fair decisions is crucial to ethically implementing machine learning algorithms in social settings. In this work, we consider the celebrated definition of counterfactual fairness. We begin by showing that an algorithm which satisfies counterfactual fairness also satisfies demographic parity, a far simpler fairness constraint. Similarly, we show that all algorithms satisfying demographic parity can be trivially modified to satisfy counterfactual fairness. Together, our results indicate that counterfactual fairness is basically equivalent to demographic parity, which has important implications for the growing body of work on counterfactual fairness. We then validate our theoretical findings empirically, analyzing three existing algorithms for counterfactual fairness against three simple benchmarks. We find that two simple benchmark algorithms outperform all three existing algorithms---in terms of fairness, accuracy, and efficiency---on several data sets. Our analysis leads us to formalize a concrete fairness goal: to preserve the order of individuals within protected groups. We believe transparency around the ordering of individuals within protected groups makes fair algorithms more trustworthy. By design, the two simple benchmark algorithms satisfy this goal while the existing algorithms do not.

The general adversary dual is a powerful tool in quantum computing because it gives a query-optimal bounded-error quantum algorithm for deciding any Boolean function. Unfortunately, the algorithm uses linear qubits in the worst case, and only works if the constraints of the general adversary dual are exactly satisfied. The challenge of improving the algorithm is that it is brittle to arbitrarily small errors since it relies on a reflection over a span of vectors. We overcome this challenge and build a robust dual adversary algorithm that can handle approximately satisfied constraints. As one application of our robust algorithm, we prove that for any Boolean function with polynomially many 1-valued inputs (or in fact a slightly weaker condition) there is a query-optimal algorithm that uses logarithmic qubits. As another application, we prove that numerically derived, approximate solutions to the general adversary dual give a bounded-error quantum algorithm under certain conditions. Further, we show that these conditions empirically hold with reasonable iterations for Boolean functions with small domains. We also develop several tools that may be of independent interest, including a robust approximate spectral gap lemma, a method to compress a general adversary dual solution using the Johnson-Lindenstrauss lemma, and open-source code to find solutions to the general adversary dual.

We consider the problem of solving the Min-Sum Submodular Cover problem using local search. The Min-Sum Submodular Cover problem generalizes the NP-complete Min-Sum Set Cover problem, replacing the input set cover instance with a monotone submodular set function. A simple greedy algorithm achieves an approximation factor of 4, which is tight unless P=NP [Streeter and Golovin, NeurIPS, 2008]. We complement the greedy algorithm with analysis of a local search algorithm. Building on work of Munagala et al. [ICDT, 2005], we show that, using simple initialization, a straightforward local search algorithm achieves a $(4+ε)$-approximate solution in time $O(n^3\log(n/ε))$, provided that the monotone submodular set function is also second-order supermodular. Second-order supermodularity has been shown to hold for a number of submodular functions of practical interest, including functions associated with set cover, matching, and facility location. We present experiments on two special cases of Min-Sum Submodular Cover and find that the local search algorithm can outperform the greedy algorithm on small data sets.

We design quantum algorithms for maximum matching. Working in the query model, in both adjacency matrix and adjacency list settings, we improve on the best known algorithms for general graphs, matching previously obtained results for bipartite graphs. In particular, for a graph with $n$ nodes and $m$ edges, our algorithm makes $O(n^{7/4})$ queries in the matrix model and $O(n^{3/4}(m+n)^{1/2})$ queries in the list model. Our approach combines Gabow's classical maximum matching algorithm [Gabow, Fundamenta Informaticae, '17] with the guessing tree method of Beigi and Taghavi [Beigi and Taghavi, Quantum, '20].

In the board game Ticket to Ride, players race to claim routes and connect cities on a map of the U.S. In this work, we identify winning strategies for and potential improvements to Ticket to Ride by applying probabilistic and graph-theoretic concepts. We find that longer routes are overvalued, presenting a simple winning strategy for opportunistic players. The scoring scheme we propose—based on indicator random variables—prevents exploitation from this strategy and improves the competitive nature of the game. Using a variety of game data visualizations, we also investigate why players who connect particular pairs of cities perform better than others. In addition, we build a statistical model from the effective resistance of the game’s underlying graph structure to suggest how to choose the best pairs of cities.