This paper considers the problem of learning safe policies in the context of reinforcement learning (RL). In particular, we consider the notion of probabilistic safety. This is, we aim to design policies that maintain the state of the system in a safe set with high probability. This notion differs from cumulative constraints often considered in the literature. The challenge of working with probabilistic safety is the lack of expressions for their gradients. Indeed, policy optimization algorithms rely on gradients of the objective function and the constraints. To the best of our knowledge, this work is the first one providing such explicit gradient expressions for probabilistic constraints. It is worth noting that the gradient of this family of constraints can be applied to various policy-based algorithms. We demonstrate empirically that it is possible to handle probabilistic constraints in a continuous navigation problem.

Trajectory generation in confined environment is crucial for wide adoption of intelligent robot manipulators. In this paper, we propose a novel motion planning approach for redundant robot arms that uses a hybrid optimization framework to search for optimal trajectories in both the configuration space and null space, generating high-quality trajectories that satisfy task constraints and collision avoidance constraints, while also avoiding local optima for incremental planners. Our approach is evaluated in an onsite polishing scenario with various robot and workpiece configurations, demonstrating significant improvements in trajectory quality compared to existing methods. The proposed approach has the potential for broad applications in industrial tasks involving redundant robot arms.

Backdoors pose a serious threat to machine learning, as they can compromise the integrity of security-critical systems, such as self-driving cars. While different defenses have been proposed to address this threat, they all rely on the assumption that the hardware on which the learning models are executed during inference is trusted. In this paper, we challenge this assumption and introduce a backdoor attack that completely resides within a common hardware accelerator for machine learning. Outside of the accelerator, neither the learning model nor the software is manipulated, so that current defenses fail. To make this attack practical, we overcome two challenges: First, as memory on a hardware accelerator is severely limited, we introduce the concept of a minimal backdoor that deviates as little as possible from the original model and is activated by replacing a few model parameters only. Second, we develop a configurable hardware trojan that can be provisioned with the backdoor and performs a replacement only when the specific target model is processed. We demonstrate the practical feasibility of our attack by implanting our hardware trojan into the Xilinx Vitis AI DPU, a commercial machine-learning accelerator. We configure the trojan with a minimal backdoor for a traffic-sign recognition system. The backdoor replaces only 30 (0.069%) model parameters, yet it reliably manipulates the recognition once the input contains a backdoor trigger. Our attack expands the hardware circuit of the accelerator by 0.24% and induces no run-time overhead, rendering a detection hardly possible. Given the complex and highly distributed manufacturing process of current hardware, our work points to a new threat in machine learning that is inaccessible to current security mechanisms and calls for hardware to be manufactured only in fully trusted environments.

Evolutionary multi-objective algorithms have successfully been used in the context of Pareto optimization where a given constraint is relaxed into an additional objective. In this paper, we explore the use of 3-objective formulations for problems with chance constraints. Our formulation trades off the expected cost and variance of the stochastic component as well as the given deterministic constraint. We point out benefits that this 3-objective formulation has compared to a bi-objective one recently investigated for chance constraints with Normally distributed stochastic components. Our analysis shows that the 3-objective formulation allows to compute all required trade-offs using 1-bit flips only, when dealing with a deterministic cardinality constraint. Furthermore, we carry out experimental investigations for the chance constrained dominating set problem and show the benefit for this classical NP-hard problem.

Traveling Salesman Problem (TSP), as a classic routing optimization problem originally arising in the domain of transportation and logistics, has become a critical task in broader domains, such as manufacturing and biology. Recently, Deep Reinforcement Learning (DRL) has been increasingly employed to solve TSP due to its high inference efficiency. Nevertheless, most of existing end-to-end DRL algorithms only perform well on small TSP instances and can hardly generalize to large scale because of the drastically soaring memory consumption and computation time along with the enlarging problem scale. In this paper, we propose a novel end-to-end DRL approach, referred to as Pointerformer, based on multi-pointer Transformer. Particularly, Pointerformer adopts both reversible residual network in the encoder and multi-pointer network in the decoder to effectively contain memory consumption of the encoder-decoder architecture. To further improve the performance of TSP solutions, Pointerformer employs both a feature augmentation method to explore the symmetries of TSP at both training and inference stages as well as an enhanced context embedding approach to include more comprehensive context information in the query. Extensive experiments on a randomly generated benchmark and a public benchmark have shown that, while achieving comparative results on most small-scale TSP instances as SOTA DRL approaches do, Pointerformer can also well generalize to large-scale TSPs.

Bilevel optimization problems, which are problems where two optimization problems are nested, have more and more applications in machine learning. In many practical cases, the upper and the lower objectives correspond to empirical risk minimization problems and therefore have a sum structure. In this context, we propose a bilevel extension of the celebrated SARAH algorithm. We demonstrate that the algorithm requires $\mathcal{O}((n+m)^{\frac12}\varepsilon^{-1})$ gradient computations to achieve $\varepsilon$-stationarity with $n+m$ the total number of samples, which improves over all previous bilevel algorithms. Moreover, we provide a lower bound on the number of oracle calls required to get an approximate stationary point of the objective function of the bilevel problem. This lower bound is attained by our algorithm, which is therefore optimal in terms of sample complexity.

This paper proposes a quantum computing-based algorithm to solve the single image super-resolution (SISR) problem. One of the well-known classical approaches for SISR relies on the well-established patch-wise sparse modeling of the problem. Yet, this field's current state of affairs is that deep neural networks (DNNs) have demonstrated far superior results than traditional approaches. Nevertheless, quantum computing is expected to become increasingly prominent for machine learning problems soon. As a result, in this work, we take the privilege to perform an early exploration of applying a quantum computing algorithm to this important image enhancement problem, i.e., SISR. Among the two paradigms of quantum computing, namely universal gate quantum computing and adiabatic quantum computing (AQC), the latter has been successfully applied to practical computer vision problems, in which quantum parallelism has been exploited to solve combinatorial optimization efficiently. This work demonstrates formulating quantum SISR as a sparse coding optimization problem, which is solved using quantum annealers accessed via the D-Wave Leap platform. The proposed AQC-based algorithm is demonstrated to achieve improved speed-up over a classical analog while maintaining comparable SISR accuracy.

Safe reinforcement learning (RL) aims to solve an optimal control problem under safety constraints. Existing $\textit{direct}$ safe RL methods use the original constraint throughout the learning process. They either lack theoretical guarantees of the policy during iteration or suffer from infeasibility problems. To address this issue, we propose an $\textit{indirect}$ safe RL method called feasible policy iteration (FPI) that iteratively uses the feasible region of the last policy to constrain the current policy. The feasible region is represented by a feasibility function called constraint decay function (CDF). The core of FPI is a region-wise policy update rule called feasible policy improvement, which maximizes the return under the constraint of the CDF inside the feasible region and minimizes the CDF outside the feasible region. This update rule is always feasible and ensures that the feasible region monotonically expands and the state-value function monotonically increases inside the feasible region. Using the feasible Bellman equation, we prove that FPI converges to the maximum feasible region and the optimal state-value function. Experiments on classic control tasks and Safety Gym show that our algorithms achieve lower constraint violations and comparable or higher performance than the baselines.

Over the years, various benchmark suites have been proposed, in which different global function properties are represented, such The GKLS generator is one of the most used testbeds for benchmarking as multi-modality, separability, ill-conditioning, and various other global optimization algorithms. In this paper, we conduct both types of global structures. In the evolutionary computation community, a computational analysis and the Exploratory Landscape Analysis the two most utilized benchmark sets are the Black-Box (ELA) of the GKLS generator. We utilize both canonically used and Optimization Benchmarking (BBOB) suite [5] which is now part of newly generated classes of GKLS-generated problems and show the COCO platform [6], and the suites that were presented at the their use in benchmarking three state-of-the-art methods (from evolutionary Congress on Evolutionary Computation (CEC) competitions (which and deterministic communities) in dimensions 5 and 10. started in 2005 and continue to this day) [9]. As was shown in [3], We show that the GKLS generator produces "needle in a haystack" the characteristics of the functions used in these two benchmarks type problems that become extremely difficult to optimize in higher are quite different. The CEC benchmarks are constructed by using dimensions. Furthermore, we conduct the ELA on the GKLS generator similar subfunctions, which possibly gives an advantage to methods and then compare it to the ELA of two other widely used that perform well on these fewer subfunctions. It was also found benchmark sets (BBOB and CEC 2014), and discuss the meaningfulness that the CEC functions share more similarities among themselves of the results.

Base placement optimization (BPO) is a fundamental capability for mobile manipulation and has been researched for decades. However, it is still very challenging for some reasons. First, compared with humans, current robots are extremely inflexible, and therefore have higher requirements on the accuracy of base placements (BPs). Second, the BP and task constraints are coupled with each other. The optimal BP depends on the task constraints, and in BP will affect task constraints in turn. More tricky is that some task constraints are flexible and non-deterministic. Third, except for fulfilling tasks, some other performance metrics such as optimal energy consumption and minimal execution time need to be considered, which makes the BPO problem even more complicated. In this paper, a Scale-like disc (SLD) representation of the workspace is used to decouple task constraints and BPs. To evaluate reachability and return optimal working pose over SLDs, a reachability map (RM) is constructed offline. In order to optimize the objectives of coverage, manipulability, and time cost simultaneously, this paper formulates the BPO as a multi-objective optimization problem (MOOP). Among them, the time optimal objective is modeled as a traveling salesman problem (TSP), which is more in line with the actual situation. The evolutionary method is used to solve the MOOP. Besides, to ensure the validity and optimality of the solution, collision detection is performed on the candidate BPs, and solutions from BPO are further fine-tuned according to the specific given task. Finally, the proposed method is used to solve a real-world toilet coverage cleaning task. Experiments show that the optimized BPs can significantly improve the coverage and efficiency of the task.