Collecting the most informative data from a large dataset distributed over a network is a fundamental problem in many fields, including control, signal processing and machine learning. In this paper, we establish a connection between selecting the most informative data and finding the top-$k$ elements of a multiset. The top-$k$ selection in a network can be formulated as a distributed nonsmooth convex optimization problem known as quantile estimation. Unfortunately, the lack of smoothness in the local objective functions leads to extremely slow convergence and poor scalability with respect to the network size. To overcome the deficiency, we propose an accelerated method that employs smoothing techniques. Leveraging the piecewise linearity of the local objective functions in quantile estimation, we characterize the iteration complexity required to achieve top-$k$ selection, a challenging task due to the lack of strong convexity. Several numerical results are provided to validate the effectiveness of the algorithm and the correctness of the theory.

In this work, we address the problem of control synthesis for a homogeneous team of robots given a global temporal logic specification and formal user preferences for relaxation in case of infeasibility. The relaxation preferences are represented as a Weighted Finite-state Edit System and are used to compute a relaxed specification automaton that captures all allowable relaxations of the mission specification and their costs. For synthesis, we introduce a Mixed Integer Linear Programming (MILP) formulation that combines the motion of the team of robots with the relaxed specification automaton. Our approach combines automata-based and MILP-based methods and leverages the strengths of both approaches while avoiding their shortcomings. Specifically, the relaxed specification automaton explicitly accounts for the progress towards satisfaction, and the MILP-based optimization approach avoids the state-space explosion associated with explicit product-automata construction, thereby efficiently solving the problem. The case studies highlight the efficiency of the proposed approach.

The maximal coding rate reduction (MCR$^2$) objective for learning structured and compact deep representations is drawing increasing attention, especially after its recent usage in the derivation of fully explainable and highly effective deep network architectures. However, it lacks a complete theoretical justification: only the properties of its global optima are known, and its global landscape has not been studied. In this work, we give a complete characterization of the properties of all its local and global optima, as well as other types of critical points. Specifically, we show that each (local or global) maximizer of the MCR$^2$ problem corresponds to a low-dimensional, discriminative, and diverse representation, and furthermore, each critical point of the objective is either a local maximizer or a strict saddle point. Such a favorable landscape makes MCR$^2$ a natural choice of objective for learning diverse and discriminative representations via first-order optimization methods. To validate our theoretical findings, we conduct extensive experiments on both synthetic and real data sets.

Simulating a single trajectory of a dynamical system under some state-dependent policy is a core bottleneck in policy optimization algorithms. The many inherently serial policy evaluations that must be performed in a single simulation constitute the bulk of this bottleneck. To wit, in applying policy optimization to supply chain optimization (SCO) problems, simulating a single month of a supply chain can take several hours. We present an iterative algorithm for policy simulation, which we dub Picard Iteration. This scheme carefully assigns policy evaluation tasks to independent processes. Within an iteration, a single process evaluates the policy only on its assigned tasks while assuming a certain 'cached' evaluation for other tasks; the cache is updated at the end of the iteration. Implemented on GPUs, this scheme admits batched evaluation of the policy on a single trajectory. We prove that the structure afforded by many SCO problems allows convergence in a small number of iterations, independent of the horizon. We demonstrate practical speedups of 400x on large-scale SCO problems even with a single GPU, and also demonstrate practical efficacy in other RL environments.

Training Large Language Models (LLMs) relies heavily on distributed implementations, employing multiple GPUs to compute stochastic gradients on model replicas in parallel. However, synchronizing gradients in data parallel settings induces a communication overhead increasing with the number of distributed workers, which can impede the efficiency gains of parallelization. To address this challenge, optimization algorithms reducing inter-worker communication have emerged, such as local optimization methods used in Federated Learning. While effective in minimizing communication overhead, these methods incur significant memory costs, hindering scalability: in addition to extra momentum variables, if communications are only allowed between multiple local optimization steps, then the optimizer's states cannot be sharded among workers. In response, we propose ACcumulate while COmmunicate (ACCO), a memory-efficient optimization algorithm tailored for distributed training of LLMs. ACCO allows to shard optimizer states across workers, overlaps gradient computations and communications to conceal communication costs, and accommodates heterogeneous hardware. Our method relies on a novel technique to mitigate the one-step delay inherent in parallel execution of gradient computations and communications, eliminating the need for warmup steps and aligning with the training dynamics of standard distributed optimization while converging faster in terms of wall-clock time. We demonstrate the effectiveness of ACCO on several LLMs training and fine-tuning tasks.

Subgroup-discovery methods allow users to obtain simple descriptions of interesting regions in a dataset. Using constraints in subgroup discovery can enhance interpretability even further. In this article, we focus on two types of constraints: First, we limit the number of features used in subgroup descriptions, making the latter sparse. Second, we propose the novel optimization problem of finding alternative subgroup descriptions, which cover a similar set of data objects as a given subgroup but use different features. We describe how to integrate both constraint types into heuristic subgroup-discovery methods. Further, we propose a novel Satisfiability Modulo Theories (SMT) formulation of subgroup discovery as a white-box optimization problem, which allows solver-based search for subgroups and is open to a variety of constraint types. Additionally, we prove that both constraint types lead to an NP-hard optimization problem. Finally, we employ 27 binary-classification datasets to compare heuristic and solver-based search for unconstrained and constrained subgroup discovery. We observe that heuristic search methods often yield high-quality subgroups within a short runtime, also in scenarios with constraints.

The diffusion model has shown success in generating high-quality and diverse solutions to trajectory optimization problems. However, diffusion models with neural networks inevitably make prediction errors, which leads to constraint violations such as unmet goals or collisions. This paper presents a novel constraint-aware diffusion model for trajectory optimization. We introduce a novel hybrid loss function for training that minimizes the constraint violation of diffusion samples compared to the groundtruth while recovering the original data distribution. Our model is demonstrated on tabletop manipulation and two-car reach-avoid problems, outperforming traditional diffusion models in minimizing constraint violations while generating samples close to locally optimal solutions.

Efficient resource utilization and perfect user experience usually conflict with each other in cloud computing platforms. Great efforts have been invested in increasing resource utilization but trying not to affect users' experience for cloud computing platforms. In order to better utilize the remaining pieces of computing resources spread over the whole platform, deferrable jobs are provided with a discounted price to users. For this type of deferrable jobs, users are allowed to submit jobs that will run for a specific uninterrupted duration in a flexible range of time in the future with a great discount. With these deferrable jobs to be scheduled under the remaining capacity after deploying those on-demand jobs, it remains a challenge to achieve high resource utilization and meanwhile shorten the waiting time for users as much as possible in an online manner. In this paper, we propose an online deferrable job scheduling method called \textit{Online Scheduling for DEferrable jobs in Cloud} (\OSDEC{}), where a deep reinforcement learning model is adopted to learn the scheduling policy, and several auxiliary tasks are utilized to provide better state representations and improve the performance of the model. With the integrated reinforcement learning framework, the proposed method can well plan the deployment schedule and achieve a short waiting time for users while maintaining a high resource utilization for the platform. The proposed method is validated on a public dataset and shows superior performance.

Dynamical System (DS)-based closed-loop control is a simple and effective way to generate reactive motion policies that well generalize to the robotic workspace, while retaining stability guarantees. Lately the formalism has been expanded in order to handle arbitrary geometry curved spaces, namely manifolds, beyond the standard flat Euclidean space. Despite the many different ways proposed to handle DS on manifolds, it is still unclear how to apply such structures on real robotic systems. In this preliminary study, we propose a way to combine modern optimal control techniques with a geometry-based formulation of DS. The advantage of such approach is two fold. First, it yields a torque-based control for compliant and adaptive motions; second, it generates dynamical systems consistent with the controlled system's dynamics. The salient point of the approach is that the complexity of designing a proper constrained-based optimal control problem, to ensure that dynamics move on a manifold while avoiding obstacles or self-collisions, is "outsourced" to the geometric DS. Constraints are implicitly embedded into the structure of the space in which the DS evolves. The optimal control, on the other hand, provides a torque-based control interface, and ensures dynamical consistency of the generated output. The whole can be achieved with minimal computational overhead since most of the computational complexity is delegated to the closed-form geometric DS.

Graph Neural Networks (GNNs) have been widely used in node classification tasks, such as advertising recommendation [15], social network anomaly detection [34], etc. However, these GNN models typically assume that the training and test graph data are drawn from the same distribution, which does not always hold in practice. In real-world graph data, sample selection bias [8, 12] as well as graph construction techniques [27, 43] often brings distribution shifts between training nodes and test nodes. For instance, In WebKB [26] datasets, web pages (nodes) and categories (labels) are heavily affected by the university they originate from, leading to distribution shifts among nodes drawn from different universities. Therefore, in order to enhance the practical validity of GNNs, it is of paramount importance to deal with distribution shifts on graph data. To address the distribution shift problem in node classification, recent works [18, 36, 32, 37, 23] borrow the idea of invariant learning methods from the literature of out-of-distribution (OOD) generalization and adopt them on graph-structured data. Invariant learning [1, 19] stems from the causal inference literature, and now becomes one of the key approaches to solving OOD problems on graphs. The core concept is to identify invariant features with stable prediction mechanisms across different environments, thereby mitigating performance degradation under distribution shifts. And most of the works in this line directly apply existing invariant learning algorithms to graph-level classification tasks (major) [18, 32, 23, 41] and node classification tasks (minor) [36, 38].