We study the problem of estimating Dynamic Discrete Choice (DDC) models, also known as offline Maximum Entropy-Regularized Inverse Reinforcement Learning (offline MaxEnt-IRL) in machine learning. The objective is to recover reward or $Q^*$ functions that govern agent behavior from offline behavior data. In this paper, we propose a globally convergent gradient-based method for solving these problems without the restrictive assumption of linearly parameterized rewards. The novelty of our approach lies in introducing the Empirical Risk Minimization (ERM) based IRL/DDC framework, which circumvents the need for explicit state transition probability estimation in the Bellman equation. Furthermore, our method is compatible with non-parametric estimation techniques such as neural networks. Therefore, the proposed method has the potential to be scaled to high-dimensional, infinite state spaces. A key theoretical insight underlying our approach is that the Bellman residual satisfies the Polyak-Lojasiewicz (PL) condition -- a property that, while weaker than strong convexity, is sufficient to ensure fast global convergence guarantees. Through a series of synthetic experiments, we demonstrate that our approach consistently outperforms benchmark methods and state-of-the-art alternatives.

This paper proposes an online inference method of the stochastic gradient descent (SGD) with a constant learning rate for quantile loss functions with theoretical guarantees. Since the quantile loss function is neither smooth nor strongly convex, we view such SGD iterates as an irreducible and positive recurrent Markov chain. By leveraging this interpretation, we show the existence of a unique asymptotic stationary distribution, regardless of the arbitrarily fixed initialization. To characterize the exact form of this limiting distribution, we derive bounds for its moment generating function and tail probabilities, controlling over the first and second moments of SGD iterates. By these techniques, we prove that the stationary distribution converges to a Gaussian distribution as the constant learning rate $\eta\rightarrow0$. Our findings provide the first central limit theorem (CLT)-type theoretical guarantees for the last iterate of constant learning-rate SGD in non-smooth and non-strongly convex settings. We further propose a recursive algorithm to construct confidence intervals of SGD iterates in an online manner. Numerical studies demonstrate strong finite-sample performance of our proposed quantile estimator and inference method. The theoretical tools in this study are of independent interest to investigate general transition kernels in Markov chains.

--Independent microgrids are crucial for supplying electricity by combining distributed energy resources and loads in scenarios like isolated islands and field combat. Fast and accurate assessments of microgrid vulnerability against intentional attacks or natural disasters are essential for effective risk prevention and design optimization. However, conventional Monte Carlo simulation (MCS) methods are computationally expensive and time-consuming, while existing machine learning-based approaches often lack accuracy and explainability. T o address these challenges, this study proposes a fast and explainable vulnerability assessment framework that integrates MCS with a graph attention network enhanced by self-attention pooling (GA T -S). MCS generates training data, while the GA T - S model learns the structural and electrical characteristics of the microgrid and further assesses its vulnerability intelligently. The GA T -S improves explainability and computational efficiency by dynamically assigning attention weights to critical nodes. Comprehensive experimental evaluations across various micro-grid configurations demonstrate that the proposed framework provides accurate vulnerability assessments, achieving a mean squared error as low as 0.001, real-time responsiveness within 1 second, and delivering explainable results. An independent microgrid, like a battlefield or island mi-crogrid, operates separately from the main grid, supplying electricity to a localized area by integrating distributed energy resources and loads via interconnected buses, transformers, and lines. Assessing the vulnerability of independent micro-grids is essential to ensure its normal power supply capacity against disruptions, particularly in scenarios like deliberate attacks and natural disasters. Chenhui Lin is with the State Key Laboratory of Power Systems, Department of Electrical Engineering, Tsinghua University, Beijing 100084, China.

Large Language Models (LLMs) have fundamentally transformed approaches to Natural Language Processing (NLP) tasks across diverse domains. In healthcare, accurate and cost-efficient text classification is crucial, whether for clinical notes analysis, diagnosis coding, or any other task, and LLMs present promising potential. Text classification has always faced multiple challenges, including manual annotation for training, handling imbalanced data, and developing scalable approaches. With healthcare, additional challenges are added, particularly the critical need to preserve patients' data privacy and the complexity of the medical terminology. Numerous studies have been conducted to leverage LLMs for automated healthcare text classification and contrast the results with existing machine learning-based methods where embedding, annotation, and training are traditionally required. Existing systematic reviews about LLMs either do not specialize in text classification or do not focus on the healthcare domain. This research synthesizes and critically evaluates the current evidence found in the literature regarding the use of LLMs for text classification in a healthcare setting. Major databases (e.g., Google Scholar, Scopus, PubMed, Science Direct) and other resources were queried, which focused on the papers published between 2018 and 2024 within the framework of PRISMA guidelines, which resulted in 65 eligible research articles. These were categorized by text classification type (e.g., binary classification, multi-label classification), application (e.g., clinical decision support, public health and opinion analysis), methodology, type of healthcare text, and metrics used for evaluation and validation. This review reveals the existing gaps in the literature and suggests future research lines that can be investigated and explored.

We investigate off-policy evaluation (OPE), a central and fundamental problem in reinforcement learning (RL), in the challenging setting of Partially Observable Markov Decision Processes (POMDPs) with large observation spaces. Recent works of Uehara et al. (2023a); Zhang & Jiang (2024) developed a model-free framework and identified important coverage assumptions (called belief and outcome coverage) that enable accurate OPE of memoryless policies with polynomial sample complexities, but handling more general target policies that depend on the entire observable history remained an open problem. In this work, we prove information-theoretic hardness for model-free OPE of history-dependent policies in several settings, characterized by additional assumptions imposed on the behavior policy (memoryless vs. history-dependent) and/or the state-revealing property of the POMDP (single-step vs. multi-step revealing). We further show that some hardness can be circumvented by a natural model-based algorithm -- whose analysis has surprisingly eluded the literature despite the algorithm's simplicity -- demonstrating provable separation between model-free and model-based OPE in POMDPs.

While likelihood-based generative models, particularly diffusion and autoregressive models, have achieved remarkable fidelity in visual generation, the maximum likelihood estimation (MLE) objective inherently suffers from a mode-covering tendency that limits the generation quality under limited model capacity. In this work, we propose Direct Discriminative Optimization (DDO) as a unified framework that bridges likelihood-based generative training and the GAN objective to bypass this fundamental constraint. Our key insight is to parameterize a discriminator implicitly using the likelihood ratio between a learnable target model and a fixed reference model, drawing parallels with the philosophy of Direct Preference Optimization (DPO). Unlike GANs, this parameterization eliminates the need for joint training of generator and discriminator networks, allowing for direct, efficient, and effective finetuning of a well-trained model to its full potential beyond the limits of MLE. DDO can be performed iteratively in a self-play manner for progressive model refinement, with each round requiring less than 1% of pretraining epochs. Our experiments demonstrate the effectiveness of DDO by significantly advancing the previous SOTA diffusion model EDM, reducing FID scores from 1.79/1.58 to new records of 1.30/0.97 on CIFAR-10/ImageNet-64 datasets, and by consistently improving both guidance-free and CFG-enhanced FIDs of visual autoregressive models on ImageNet 256$\times$256.

From a first-principles perspective, it may seem odd that the strongest results in foundation model fine-tuning (FT) are achieved via a relatively complex, two-stage training procedure. Specifically, one first trains a reward model (RM) on some dataset (e.g. human preferences) before using it to provide online feedback as part of a downstream reinforcement learning (RL) procedure, rather than directly optimizing the policy parameters on the dataset via offline maximum likelihood estimation. In fact, from an information-theoretic perspective, we can only lose information via passing through a reward model and cannot create any new information via on-policy sampling. To explain this discrepancy, we scrutinize several hypotheses on the value of RL in FT through both theoretical and empirical lenses. Of the hypotheses considered, we find the most support for the explanation that on problems with a generation-verification gap, the combination of the ease of learning the relatively simple RM (verifier) from the preference data, coupled with the ability of the downstream RL procedure to then filter its search space to the subset of policies (generators) that are optimal for relatively simple verifiers is what leads to the superior performance of online FT.

Satisfiability Modulo Counting (SMC) is a recently proposed general language to reason about problems integrating statistical and symbolic artificial intelligence. An SMC formula is an extended SAT formula in which the truth values of a few Boolean variables are determined by probabilistic inference. Existing approximate solvers optimize surrogate objectives, which lack formal guarantees. Current exact solvers directly integrate SAT solvers and probabilistic inference solvers resulting in slow performance because of many back-and-forth invocations of both solvers. We propose KOCO-SMC, an integrated exact SMC solver that efficiently tracks lower and upper bounds in the probabilistic inference process. It enhances computational efficiency by enabling early estimation of probabilistic inference using only partial variable assignments, whereas existing methods require full variable assignments. In the experiment, we compare KOCO-SMC with currently available approximate and exact SMC solvers on large-scale datasets and real-world applications. Our approach delivers high-quality solutions with high efficiency.

-- Recent advances in large language models (LLMs) have demonstrated their potential as planners in human-robot collaboration (HRC) scenarios, offering a promising alternative to traditional planning methods. LLMs, which can generate structured plans by reasoning over natural language inputs, have the ability to generalize across diverse tasks and adapt to human instructions. This paper investigates the potential of LLMs to facilitate planning in the context of human-robot collaborative tasks, with a focus on their ability to reason from high-level, vague human inputs, and fine-tune plans based on real-time feedback. We propose a novel hybrid framework that combines LLMs with human feedback to create dynamic, context-aware task plans. Our work also highlights how a single, concise prompt can be used for a wide range of tasks and environments, overcoming the limitations of long, detailed structured prompts typically used in prior studies. By integrating user preferences into the planning loop, we ensure that the generated plans are not only effective but aligned with human intentions. Planning is a fundamental aspect of robotics, enabling autonomous agents to generate sequences of actions to achieve specific goals. Traditional planning methods in the context of human-robot collaboration (HRC) and assistive robots can be broadly categorized into two main types: rule-based planners and learning-based planners . Rule-based planners rely on predefined heuristics and symbolic representations, making them interpretable at the expense of not being able to adapt to complex or dynamic environments. In contrast, learning-based planners, particularly those utilizing deep reinforcement learning, learn to generate plans from experience in an adaptive manner.

Markov decision processes (MDPs) are a canonical model to reason about decision making within a stochastic environment. We study a fundamental class of infinite MDPs: one-counter MDPs (OC-MDPs). They extend finite MDPs via an associated counter taking natural values, thus inducing an infinite MDP over the set of configurations (current state and counter value). We consider two characteristic objectives: reaching a target state (state-reachability), and reaching a target state with counter value zero (selective termination). The synthesis problem for the latter is not known to be decidable and connected to major open problems in number theory. Furthermore, even seemingly simple strategies (e.g., memoryless ones) in OC-MDPs might be impossible to build in practice (due to the underlying infinite configuration space): we need finite, and preferably small, representations. To overcome these obstacles, we introduce two natural classes of concisely represented strategies based on a (possibly infinite) partition of counter values in intervals. For both classes, and both objectives, we study the verification problem (does a given strategy ensure a high enough probability for the objective?), and two synthesis problems (does there exist such a strategy?): one where the interval partition is fixed as input, and one where it is only parameterized. We develop a generic approach based on a compression of the induced infinite MDP that yields decidability in all cases, with all complexities within PSPACE.