Machine learning models in the wild have been shown to be vulnerable to Trojan attacks during training. Although many detection mechanisms have been proposed, strong adaptive attackers have been shown to be effective against them. In this paper, we aim to answer the questions considering an intelligent and adaptive adversary: (i) What is the minimal amount of instances required to be Trojaned by a strong attacker? and (ii) Is it possible for such an attacker to bypass strong detection mechanisms? We provide an analytical characterization of adversarial capability and strategic interactions between the adversary and detection mechanism that take place in such models. We characterize adversary capability in terms of the fraction of the input dataset that can be embedded with a Trojan trigger. We show that the loss function has a submodular structure, which leads to the design of computationally efficient algorithms to determine this fraction with provable bounds on optimality. We propose a Submodular Trojan algorithm to determine the minimal fraction of samples to inject a Trojan trigger. To evade detection of the Trojaned model, we model strategic interactions between the adversary and Trojan detection mechanism as a two-player game. We show that the adversary wins the game with probability one, thus bypassing detection. We establish this by proving that output probability distributions of a Trojan model and a clean model are identical when following the Min-Max (MM) Trojan algorithm. We perform extensive evaluations of our algorithms on MNIST, CIFAR-10, and EuroSAT datasets. The results show that (i) with Submodular Trojan algorithm, the adversary needs to embed a Trojan trigger into a very small fraction of samples to achieve high accuracy on both Trojan and clean samples, and (ii) the MM Trojan algorithm yields a trained Trojan model that evades detection with probability 1.

We study policy optimization problems for deterministic Markov decision processes (MDPs) with metric state and action spaces, which we refer to as Metric Policy Optimization Problems (MPOPs). Our goal is to establish theoretical results on the well-posedness of MPOPs that can characterize practically relevant continuous control systems. To do so, we define a special class of MPOPs called Compactly Restrictable MPOPs (CR-MPOPs), which are flexible enough to capture the complex behavior of robotic systems but specific enough to admit solutions using dynamic programming methods such as value iteration. We show how to arrive at CR-MPOPs using forward-invariance. We further show that our theoretical results on CR-MPOPs can be used to characterize feedback linearizable control affine systems.

Security-constrained unit commitment (SCUC) is solved for power system day-ahead generation scheduling, which is a large-scale mixed-integer linear programming problem and is very computationally intensive. Model reduction of SCUC may bring significant time savings. In this work, a novel approach is proposed to effectively utilize machine learning (ML) to reduce the problem size of SCUC. An ML model using logistic regression (LR) algorithm is proposed and trained with historical nodal demand profiles and the respective commitment schedules. The ML outputs are processed and analyzed to reduce variables and constraints in SCUC. The proposed approach is validated on several standard test systems including IEEE 24-bus system, IEEE 73-bus system, IEEE 118-bus system, synthetic South Carolina 500-bus system and Polish 2383-bus system. Simulation results demonstrate that the use of the prediction from the proposed LR model in SCUC model reduction can substantially reduce the computing time while maintaining solution quality.

More and more investors and machine learning models rely on social media (e.g., Twitter and Reddit) to gather real-time information and sentiment to predict stock price movements. Although text-based models are known to be vulnerable to adversarial attacks, whether stock prediction models have similar vulnerability is underexplored. In this paper, we experiment with a variety of adversarial attack configurations to fool three stock prediction victim models. We address the task of adversarial generation by solving combinatorial optimization problems with semantics and budget constraints. Our results show that the proposed attack method can achieve consistent success rates and cause significant monetary loss in trading simulation by simply concatenating a Figure 1: An example of word-replacement adversarial perturbed but semantically similar tweet.

A promising approach to useful computational quantum advantage is to use variational quantum algorithms for optimization problems. Crucial for the performance of these algorithms is to ensure that the algorithm converges with high probability to a near-optimal solution in a small time. In Barkoutsos et al. [Quantum 4, 256 (2020)], an alternative class of objective functions, called conditional value at risk (CVaR), was introduced and it was shown that they perform better than standard objective functions. Here we extend that work by introducing an evolving objective function, which we call ascending-CVaR and that can be used for any optimization problem. We test our proposed objective function in an emulation environment, using as case studies three different optimization problems: MaxCut, number partitioning, and portfolio optimization. We examine multiple instances of different sizes and analyze the performance using the variational quantum eigensolver with hardware-efficient ansatz and the quantum approximate optimization algorithm. We show that ascending-CVaR in all cases performs better than standard objective functions or the constant CVaR of Barkoutsos et al. [Quantum 4, 256 (2020)] and that it can be used as a heuristic for avoiding suboptimal minima. Our proposal achieves higher overlap with the ideal state in all problems, whether we consider easy or hard instances---on average, it gives up to ten times greater overlap at portfolio optimization and number partitioning, while it gives an 80% improvement at MaxCut. In the hard instances we consider, for the number partitioning problem, standard objective functions fail to find the correct solution in almost all cases, CVaR finds the correct solution at 60% of the cases, while ascending-CVaR finds the correct solution in 95% of the cases.

Moment restrictions identify a parameter of interest by restricting the expectation value of so-called moment functions, which depend on the parameter and random variables representing the underlying noisy data generating process. Important problems in causal inference, economics, and generally robust machine learning can be cast in this form [Newey, 1993, Ai and Chen, 2003, Bennett and Kallus, 2020b, Dikkala et al., 2020]. Particularly challenging are problems formulated as conditional moment restrictions (CMR), which constrain the conditional expectation of the moment function. Such problems appear, e.g., in instrumental variable (IV) regression [Newey and Powell, 2003, Angrist and Pischke, 2008], where the expectation of the residual of the prediction conditioned on so-called instruments is restricted to be zero. Other applications are policy learning [Bennett and Kallus, 2020a] and off-policy evaluation in reinforcement learning [Kallus and Uehara, 2020, Bennett et al., 2021, Chen et al., 2021] and double/debiased machine learning [Chernozhukov et al., 2016, 2017, 2018]. As conditional moment restrictions are difficult to handle directly, a common approach is to transform them into an infinite number of corresponding unconditional moment restrictions Bierens [1982]. Generalizing the corresponding estimation methods from the finite dimensional case to the infinite case is an active area of research [Carrasco and Florens, 2000, Carrasco et al., 2007, Chaussé, 2012, Carrasco and Kotchoni, 2017, Muandet et al., 2020, Bennett and Kallus, 2020b, Zhang et al., 2021]. One of the most popular approaches to learning with moment restrictions is Hansen's celebrated generalized method of moments (GMM) [Hansen, 1982]. In order to improve the small sample properties of GMM estimators, alternative methods have been proposed and are generally known as generalized empirical likelihood (GEL) estimators [Smith, 1997, 2005, Newey and Smith, 2004].

Abstract-- Safety guarantees in motion planning for autonomous driving typically involve certifying the trajectory to be collision-free under any motion of the uncontrollable participants in the environment, such as the human-driven vehicles on the road. As a result they usually employ a conservative bound on the behavior of such participants, such as reachability analysis. We point out that planning trajectories to rigorously avoid the entirety of the reachable regions is unnecessary and too restrictive, because observing the environment in the future will allow us to prune away most of them; disregarding this ability to react to future updates could prohibit solutions to scenarios that are easily navigated by human drivers. We propose to account for the autonomous vehicle's reactions to future environment changes by a novel safety framework, Comprehensive Reactive Safety. Validated in simulations in several urban driving scenarios such as unprotected left turns and lane merging, the resulting planning algorithm called Reactive ILQR demonstrates strong negotiation capabilities and better safety at the same time.

We introduce Uniform Manifold Approximation with Two-phase Optimization (UMATO), a dimensionality reduction (DR) technique that improves UMAP to capture the global structure of high-dimensional data more accurately. In UMATO, optimization is divided into two phases so that the resulting embeddings can depict the global structure reliably while preserving the local structure with sufficient accuracy. In the first phase, hub points are identified and projected to construct a skeletal layout for the global structure. In the second phase, the remaining points are added to the embedding preserving the regional characteristics of local areas. Through quantitative experiments, we found that UMATO (1) outperformed widely used DR techniques in preserving the global structure while (2) producing competitive accuracy in representing the local structure. We also verified that UMATO is preferable in terms of robustness over diverse initialization methods, number of epochs, and subsampling techniques.

Contrast maximization (CMax) is a framework that provides state-of-the-art results on several event-based computer vision tasks, such as ego-motion or optical flow estimation. However, it may suffer from a problem called event collapse, which is an undesired solution where events are warped into too few pixels. As prior works have largely ignored the issue or proposed workarounds, it is imperative to analyze this phenomenon in detail. Our work demonstrates event collapse in its simplest form and proposes collapse metrics by using first principles of space-time deformation based on differential geometry and physics. We experimentally show on publicly available datasets that the proposed metrics mitigate event collapse and do not harm well-posed warps. To the best of our knowledge, regularizers based on the proposed metrics are the only effective solution against event collapse in the experimental settings considered, compared with other methods. We hope that this work inspires further research to tackle more complex warp models.

In the past decade, with the development of big data technology, an increasing amount of patient information has been stored as electronic health records (EHRs). Leveraging these data, various doctor recommendation systems have been proposed. Typically, such studies process the EHR data in a flat-structured manner, where each encounter was treated as an unordered set of features. Nevertheless, the heterogeneous structured information such as service sequence stored in claims shall not be ignored. This paper presents a doctor recommendation system with time embedding to reconstruct the potential connections between patients and doctors using heterogeneous graph attention network. Besides, to address the privacy issue of patient data sharing crossing hospitals, a federated decentralized learning method based on a minimization optimization model is also proposed. The graph-based recommendation system has been validated on a EHR dataset. Compared to baseline models, the proposed method improves the AUC by up to 6.2%. And our proposed federated-based algorithm not only yields the fictitious fusion center's performance but also enjoys a convergence rate of O(1/T).