Visual textures -- spatially homogeneous image regions containing repeated elements (e.g. a field of grass, the bark of a tree) -- are ubiquitous in visual scenes and provide important cues for recognizing and analyzing materials, surface geometry and objects. A number of existing texture models extract essential statistics from a single texture image, and can then generate high-quality samples that are visually similar to the original by matching these statistics. However, their statistics are either hand-designed or based on a network pretrained for another purpose (e.g., object recognition). Here, we develop the first principled method for unsupervised learning of a set of statistics that are used to constrain a maximum entropy probability model for texture. We leverage methods developed for generative diffusion models to derive training and sampling procedures, and compare these to the traditional method of sampling via matching the statistics. Despite the compactness of our trained model (512 statistics), it generates texture images whose quality is as good as or better than the current state-of-the-art model (~177k statistics), both visually and by quantitative perceptual scoring using DreamSim. Finally, we show that images generated from parameters lying along a line between the parameter vectors corresponding to two different reference textures are homogeneous textures with features that smoothly transition between those of the two references.
An AI agent may keep a successful check in context after an edit has invalidated its evidence. We give a semantics for specifying when a system may rely on a message and act on it. Reliability checks an asserted claim against accepted domain knowledge, sources, and messages in the selected context. Justified reliance additionally checks current evidence, authority, and the proposed world change. Relational component semantics and event traces represent nondeterministic outputs, iteration, and external changes. We characterise when merging claims preserves derivable conclusion classes, then construct the coarsest permitted quotient whose rules preserve the exact statuses of designated concrete claims. Construction is polynomial for unary rules; deciding even one permitted merge is coNP-complete for acyclic binary rules. This applies completeness and strong-preservation methods to the execution of quotiented rules. We also relate transition safeguards to runtime enforcement. Relative to a supplied set of possible worlds, a sound guard accepts an observation only when every matching world passes the checks. Exact acceptance requires observations that distinguish opposite-decision worlds; choosing sufficient services is NP-complete on explicit finite tables. A complete finite passport-renewal instance, Lean theorem encodings, Z3 instance checks, and an OCaml harness exercise the semantics. Domain knowledge, interpretation, source licenses, and policy remain application inputs.
We propose an instantiation of flow matching that relies on a time-independent velocity field---an autonomous flow---to exactly map between two distributions when the target is supported on a sufficiently lower-dimensional data manifold. We also show that the associated one-step generative map is constant in the direction of the autonomous velocity field and therefore satisfies a simple conservation equation, which motivates learning the map directly from samples. These autonomous flows and maps give a dynamical meaning to the flux constraint of Beckmann's transportation problem, and we show that, under suitable conditions, the construction extends beyond flow matching to other probability fluxes satisfying the same constraint. The resulting framework includes, for instance, a closed-form Coulomb transport closely related to Poisson Flow and a flow-matching-consistent version of Equilibrium Matching. We illustrate these distinctions and demonstrate the effectiveness of the autonomous flow and the one-step map on ImageNet $256\times256$.
Reinforcement learning (RL) has become central to improving large language models (LLMs) on complex reasoning tasks, yet RL post-training is largely studied in isolation from the pretraining that precedes it. As a result, two basic questions remain open: (1) how do pretraining choices (model size, data) shape the returns to RL compute, and (2) what does RL actually do to the model? These questions are difficult to study in the standard LLM setting: pretraining corpora are vast and uncontrolled, making it hard to attribute behaviors to pretraining versus RL, and systematic compute sweeps across both stages are prohibitively expensive. To address these challenges, we use chess as a controlled testbed for studying reasoning across the full pretraining-to-post-training pipeline. We follow the standard LLM training pipeline by pretraining language models from 5M to 1B parameters on human chess games, supervised fine-tuning on synthetic reasoning traces, and running RL on chess puzzles with verifiable rewards. Using this framework, we find that the post-RL performance at given RL compute level is well-predicted from the pretraining loss, and slope of the RL reward curves improves approximately linearly with the pretraining tokens. Beyond scaling, we find that RL does not simply sharpen the SFT policy: on easy puzzles it amplifies correct moves the SFT policy already preferred, while on hard puzzles it surfaces correct moves that were nearly absent under SFT. We further test whether our findings transfer beyond chess by training a 1B language model on math-domain text, where the same predictive pattern emerges: longer-pretrained checkpoints reach higher post-RL performance and improve faster under RL. In sum, we provide a quantitative account of the pretraining-to-RL interface and a controlled testbed for studying the science of reasoning across the full pretraining-to-post-training pipeline.
As multimodal conversational systems increasingly engage in spoken interaction, their ability to navigate paralinguistic social cues has become a critical bottleneck for natural human-AI communication. However, existing evaluations of machine emotional intelligence assess reasoning exclusively through isolated text or passive acoustic perception, overlooking the complex cross-modal reasoning required for active, multi-turn dialogue. We introduce \textsc{SpeechEQ}, a comprehensive framework designed to evaluate the sociolinguistic reasoning of Speech-Language Models (SLMs). The framework includes a validated dataset of 2,265 dialogues across 15 Emotional Quotient (EQ) subscales grounded in EQ-i 2.0 theory, along with a multi-turn evaluation protocol measured by our proposed Spoken EQ (SEQ) score inspired by human EQ assessments. Experiments show limitations in how both existing Speech Emotion Recognition and end-to-end Speech-Language Models understand and apply paralinguistic cues through speech. While end-to-end architectures outperform cascaded systems, \textsc{SpeechEQ} reveals that current multimodal models remain bottlenecked by a text-reliant ``modality shortcut,'' an alignment-induced ``safety trap,'' and ``contextual amnesia,'' highlighting the barriers to truly emotionally aware AI. Our benchmark can be accessed at https://huggingface.co/datasets/SpeechEQ/SpeechEQ and demo page at https://speecheq.github.io
Large language models are increasingly used to support organizational decisions. Despite their rapid and widespread adoption, users often lack a principled basis for assessing whether to rely on a specific recommendation. We introduce \emph{epistemic warrant}, a decision-level construct that characterizes the scope over which a recommendation is supported. We operationalize this construct through a reliance certificate that uses behavioral evidence from four tiers of tests to assess a recommendation's underlying warrant and assign it to a warrant category. We assess this operationalization using complementary forms of construct-validity evidence, including transformation validity, known-groups tests, independent human consensus, discriminant analyses, and robustness across models and alternative implementations. The results show that the certificate behaves in theoretically expected ways: its transformations instantiate their intended relationships, certificate assignments increase across groups prespecified to represent increasing levels of warrant, and stronger assigned warrant is associated with greater independent human consensus. Certificate assignments also capture information beyond verbalized confidence and are not readily explained by decision difficulty. The main conclusions remain robust across models and operationalizations. These findings support the reliance certificate as an operationalization of epistemic warrant for assessing the scope of support for a particular LLM recommendation when ground truth is unavailable.
Long-form thinking traces can substantially improve the multi-step reasoning performance of large language models (LLMs), but they introduce high inference-time overhead, with latency dominated by sequential decoding. We propose HyperThink, a text-to-parameter approach that amortizes this reasoning computation into a single query-conditioned parameter update: a lightweight hypernetwork reads the question and predicts updates to a small subset of the base LLM's parameters, while a vector-quantized decoder constrains them to a finite set of reusable patterns to improve robustness and transfer. Trained end-to-end on outputs from the base model itself, HyperThink eliminates long thinking traces at test time: after one hypernetwork forward pass, the adapted model generates a concise step-by-step solution and final answer without an intermediate trace, using far fewer tokens while retaining strong reasoning performance. Empirically, HyperThink improves the low-latency region of the accuracy-latency trade-off on mathematical and general reasoning tasks, with its strongest gains in the near-non-thinking regime.
This work addresses the critical lack of precision in state estimation in the Kalman filter for 3D multi-object tracking (MOT) and the ongoing challenge of selecting the appropriate motion model. Existing literature commonly relies on constant motion models for estimating the states of objects, neglecting the complex motion dynamics unique to each object. Consequently, trajectory division and imprecise object localization arise, especially under occlusion conditions. The core of these challenges lies in the limitations of the current Kalman filter formulation, which fails to account for the variability of motion dynamics as objects navigate their environments. This work introduces a novel formulation of the Kalman filter that incorporates motion dynamics, allowing the motion model to adaptively adjust according to changes in the object's movement. The proposed Kalman filter substantially improves state estimation, localization, and trajectory prediction compared to the traditional Kalman filter. This is reflected in tracking performance that surpasses recent benchmarks on the KITTI and Waymo Open Datasets, with margins of 0.56\% and 0.81\% in higher order tracking accuracy (HOTA) and multi-object tracking accuracy (MOTA), respectively. Furthermore, the proposed Kalman filter consistently outperforms the baseline across various detectors. Additionally, it shows an enhanced capability in managing long occlusions compared to the baseline Kalman filter, achieving margins of 1.22\% in higher order tracking accuracy (HOTA) and 1.55\% in multi-object tracking accuracy (MOTA) on the KITTI dataset. The formulation's efficiency is evident, with an additional processing time of only approximately 0.078 ms per frame, ensuring its applicability in real-time applications.
We study regret minimization in repeated first-price auctions (FPAs), where a bidder observes only the realized outcome after each auction -- win or loss. This setup reflects practical scenarios in online display advertising where the actual value of an impression depends on the difference between two potential outcomes, such as clicks or conversion rates, when the auction is won versus lost. We analyze three outcome models: (1) adversarial outcomes without features, (2) linear potential outcomes with features, and (3) linear treatment effects in features. For each setting, we propose algorithms that jointly estimate private values and optimize bidding strategies, achieving near-optimal regret bounds. Notably, our framework enjoys a unique feature that the treatments are also actively chosen, and hence eliminates the need for the overlap condition commonly required in causal inference.
Currently, state-of-the-art RL methods excel in single-task settings, but they still struggle to generalize across multiple tasks due to catastrophic forgetting challenges, where previously learned tasks are forgotten as new tasks are introduced. This multi-task learning capability is significantly important for generalist agents, where adaptation features are highly required (e.g., autonomous robots). On the other hand, Spiking Neural Networks (SNNs) have emerged as alternative energy-efficient neural network algorithms due to their sparse spike-based operations. Toward this, we propose MTSpark, a novel methodology to enable multi-task RL using spiking networks. Specifically, MTSpark develops a Deep Spiking Q-Network (DSQN) with active dendrites and dueling structure by leveraging task-specific context signals. Specifically, each neuron computes task-dependent activations that dynamically modulate inputs, forming specialized sub-networks for each task. Moreover, this bioplausible network model also benefits from SNNs, enhancing energy efficiency and making the model suitable for hardware implementation. Experimental results show that, our MTSpark effectively learns multiple tasks with higher performance compared to the state-of-the-art. Specifically, MTSpark successfully achieves high score in three Atari games (i.e., Pong: -5.4, Breakout: 0.6, and Enduro: 371.2), reaching human-level performance (i.e., Pong: -3, Breakout: 31, and Enduro: 368), where state-of-the-art struggle to achieve. In addition, our MTSpark also shows better accuracy in image classification tasks than the state-of-the-art. These results highlight the potential of our MTSpark methodology to develop generalist agents that can learn multiple tasks by leveraging both RL and SNN concepts.
We study nonconvex optimization in high dimensions through Langevin dynamics, focusing on the multi-spiked tensor PCA problem. This tensor estimation problem involves recovering $r$ hidden signal vectors (spikes) from noisy Gaussian tensor observations using maximum likelihood estimation. We study the number of samples required for Langevin dynamics to efficiently recover the spikes and determine the necessary separation condition on the signal-to-noise ratios (SNRs) for exact recovery, distinguishing the cases $p \ge 3$ and $p=2$, where $p$ denotes the order of the tensor. In particular, we show that the sample complexity required for recovering the spike associated with the largest SNR matches the well-known algorithmic threshold for the single-spike case, while this threshold degrades when recovering all $r$ spikes. As a key step, we provide a detailed characterization of the trajectory and interactions of low-dimensional projections that capture the high-dimensional dynamics.
We introduce a low-resource safety enhancement method for aligning large language models (LLMs) without the need for supervised fine-tuning (SFT) or reinforcement learning from human feedback (RLHF). Our main idea is to exploit knowledge distillation to extract the alignment information from existing well-aligned LLMs and integrate it into unaligned LLMs in a plug-and-play fashion. Methodology, we employ delta debugging to identify the critical components of knowledge necessary for effective distillation. On the harmful question dataset, our method significantly enhances the average defense success rate by approximately 14.41%, reaching as high as 51.39%, in 17 unaligned pre-trained LLMs, without compromising performance.
In the past several years, the last-iterate convergence of the Stochastic Gradient Descent (SGD) algorithm has triggered people's interest due to its good performance in practice but lack of theoretical understanding. For Lipschitz convex functions, different works have established the optimal $O(\log(1/\delta)\log T/\sqrt{T})$ or $O(\sqrt{\log(1/\delta)/T})$ high-probability convergence rates for the final iterate, where $T$ is the time horizon and $\delta$ is the failure probability. However, to prove these bounds, all the existing works are either limited to compact domains or require almost surely bounded noises. It is natural to ask whether the last iterate of SGD can still guarantee the optimal convergence rate but without these two restrictive assumptions. Besides this important question, there are still lots of theoretical problems lacking an answer. For example, compared with the last-iterate convergence of SGD for non-smooth problems, only few results for smooth optimization have yet been developed. Additionally, the existing results are all limited to a non-composite objective and the standard Euclidean norm. It still remains unclear whether the last-iterate convergence can be provably extended to wider composite optimization and non-Euclidean norms. In this work, to address the issues mentioned above, we revisit the last-iterate convergence of stochastic gradient methods and provide the first unified way to prove the convergence rates both in expectation and in high probability to accommodate general domains, composite objectives, non-Euclidean norms, Lipschitz conditions, smoothness, and (strong) convexity simultaneously. Additionally, we extend our analysis to obtain the last-iterate convergence under heavy-tailed noises.
Assortment optimization has received active explorations in the past few decades due to its practical importance. Despite the extensive literature dealing with optimization algorithms and latent score estimation, uncertainty quantification for the optimal assortment still needs to be explored and is of great practical significance. Instead of estimating and recovering the complete optimal offer set, decision-makers may only be interested in testing whether a given property holds true for the optimal assortment, such as whether they should include several products of interest in the optimal set, or how many categories of products the optimal set should include. This paper proposes a novel inferential framework for testing such properties. We consider the widely adopted multinomial logit (MNL) model, where we assume that each customer will purchase an item within the offered products with a probability proportional to the underlying preference score associated with the product. We reduce inferring a general optimal assortment property to quantifying the uncertainty associated with the sign change point detection of the marginal revenue gaps. We show the asymptotic normality of the marginal revenue gap estimator, and construct a maximum statistic via the gap estimators to detect the sign change point. By approximating the distribution of the maximum statistic with multiplier bootstrap techniques, we propose a valid testing procedure. We also conduct numerical experiments to assess the performance of our method.
In crowdsourcing, quality control is commonly achieved by having workers examine items and vote on their correctness. To minimize the impact of unreliable worker responses, a $\delta$-margin voting process is utilized, where additional votes are solicited until a predetermined threshold $\delta$ for agreement between workers is exceeded. The process is widely adopted but only as a heuristic. Our research presents a modeling approach using absorbing Markov chains to analyze the characteristics of this voting process that matter in crowdsourced processes. We provide closed-form equations for the quality of resulting consensus vote, the expected number of votes required for consensus, the variance of vote requirements, and other distribution moments. Our findings demonstrate how the threshold $\delta$ can be adjusted to achieve quality equivalence across voting processes that employ workers with varying accuracy levels. We also provide efficiency-equalizing payment rates for voting processes with different expected response accuracy levels. Additionally, our model considers items with varying degrees of difficulty and uncertainty about the difficulty of each example. Our simulations, using real-world crowdsourced vote data, validate the effectiveness of our theoretical model in characterizing the consensus aggregation process. The results of our study can be effectively employed in practical crowdsourcing applications.
We study the sequential batch learning problem in linear contextual bandits with finite action sets, where the decision maker is constrained to split incoming individuals into (at most) a fixed number of batches and can only observe outcomes for the individuals within a batch at the batch's end. Compared to both standard online contextual bandits learning or offline policy learning in contexutal bandits, this sequential batch learning problem provides a finer-grained formulation of many personalized sequential decision making problems in practical applications, including medical treatment in clinical trials, product recommendation in e-commerce and adaptive experiment design in crowdsourcing. We study two settings of the problem: one where the contexts are arbitrarily generated and the other where the contexts are \textit{iid} drawn from some distribution. In each setting, we establish a regret lower bound and provide an algorithm, whose regret upper bound nearly matches the lower bound. As an important insight revealed therefrom, in the former setting, we show that the number of batches required to achieve the fully online performance is polynomial in the time horizon, while for the latter setting, a pure-exploitation algorithm with a judicious batch partition scheme achieves the fully online performance even when the number of batches is less than logarithmic in the time horizon. Together, our results provide a near-complete characterization of sequential decision making in linear contextual bandits when batch constraints are present.
Along with fruitful applications of Deep Neural Networks (DNNs) to realistic problems, recently, some empirical studies of DNNs reported a universal phenomenon of Frequency Principle (F-Principle): a DNN tends to learn a target function from low to high frequencies during the training. The F-Principle has been very useful in providing both qualitative and quantitative understandings of DNNs. In this paper, we rigorously investigate the F-Principle for the training dynamics of a general DNN at three stages: initial stage, intermediate stage, and final stage. For each stage, a theorem is provided in terms of proper quantities characterizing the F-Principle. Our results are general in the sense that they work for multilayer networks with general activation functions, population densities of data, and a large class of loss functions. Our work lays a theoretical foundation of the F-Principle for a better understanding of the training process of DNNs.