Semantic segmentation serves as a cornerstone of scene understanding in autonomous driving but continues to face significant challenges under complex conditions such as occlusion. Light field and LiDAR modalities provide complementary visual and spatial cues that are beneficial for robust perception; however, their effective integration is hindered by limited viewpoint diversity and inherent modality discrepancies. To address these challenges, the first multimodal semantic segmentation dataset integrating light field data and point cloud data is proposed. Based on this dataset, we proposed a multi-modal light field point-cloud fusion segmentation network(Mlpfseg), incorporating feature completion and depth perception to segment both camera images and LiDAR point clouds simultaneously. The feature completion module addresses the density mismatch between point clouds and image pixels by performing differential reconstruction of point-cloud feature maps, enhancing the fusion of these modalities. The depth perception module improves the segmentation of occluded objects by reinforcing attention scores for better occlusion awareness. Our method outperforms image-only segmentation by 1.71 Mean Intersection over Union(mIoU) and point cloud-only segmentation by 2.38 mIoU, demonstrating its effectiveness.

As programmers write code, they often edit and retry multiple times, creating rich "interaction traces" that reveal how they approach coding tasks and provide clues about their level of skill development. For novice programmers in particular, these traces reflect the diverse reasoning processes they employ to code, such as exploratory behavior to understand how a programming concept works, re-strategizing in response to bugs, and personalizing stylistic choices. In this work, we explore what can be learned from training language models on such reasoning traces: not just about code, but about coders, and particularly students learning to program. We introduce a dataset of over 3.8 million programming reasoning traces from users of Pencil Code, a free online educational platform used by students to learn simple programming concepts. Compared to models trained only on final programs or synthetically-generated traces, we find that models trained on real traces are stronger at modeling diverse student behavior. Through both behavioral and probing analyses, we also find that many properties of code traces, such as goal backtracking or number of comments, can be predicted from learned representations of the students who write them. Building on this result, we show that we can help students recover from mistakes by steering code generation models to identify a sequence of edits that will results in more correct code while remaining close to the original student's style. Together, our results suggest that many properties of code are properties of individual students and that training on edit traces can lead to models that are more steerable, more predictive of student behavior while programming, and better at generating programs in their final states. Code and data is available at https://github.com/meghabyte/pencilcode-public

Power transforms are popular parametric methods for making data more Gaussian-like, and are widely used as preprocessing steps in statistical analysis and machine learning. However, we find that direct implementations of power transforms suffer from severe numerical instabilities, which can lead to incorrect results or even crashes. In this paper, we provide a comprehensive analysis of the sources of these instabilities and propose effective remedies. We further extend power transforms to the federated learning setting, addressing both numerical and distributional challenges that arise in this context. Experiments on real-world datasets demonstrate that our methods are both effective and robust, substantially improving stability compared to existing approaches.

Distilling long-form reasoning from teacher models into smaller students requires selecting which candidate solutions to train on. Recent work argues that one should select responses the student model assigns highest probability, i.e., favoring solutions ``natural'' to the student. However, we find that this approach works within a single teacher but fails when scaling to long reasoning traces from multiple diverse teachers. We identify a key cause: this approach scores entire solutions, but students generalize by recombining familiar reasoning steps, not by memorizing complete solutions. Full-trajectory scoring optimizes the wrong target; it rewards global fluency while the transferable signal lies in local step transitions. We propose Local Average Log Probability (LALP), which scores each reasoning step using only a small window of preceding context, measuring whether each step is justified by its immediate premises rather than whether the full response looks natural to the student. LALP enables two practical use cases: selecting the best teacher before fine-tuning and curating training data from diverse teacher pools. Across math, coding, and science reasoning tasks, LALP consistently improves accuracy when selecting the most natural solutions by a large margin.

Imaging inverse problems aim to recover high-dimensional signals from undersampled, noisy measurements, a fundamentally ill-posed task with infinite solutions in the null-space of the sensing operator. To resolve this ambiguity, prior information is typically incorporated through handcrafted regularizers or learned models that constrain the solution space. However, these priors typically ignore the task-specific structure of that null-space. In this work, we propose Non-Linear Projections of the Null-Space (NPN), a novel class of regularization that, instead of enforcing structural constraints in the image domain, promotes solutions that lie in a low-dimensional projection of the sensing matrix's null-space with a neural network. Our approach has two key advantages: (1) Interpretability: by focusing on the structure of the null-space, we design sensing-matrix-specific priors that capture information orthogonal to the signal components that are fundamentally blind to the sensing process. (2) Flexibility: NPN is adaptable to various inverse problems, compatible with existing reconstruction frameworks, and complementary to conventional image-domain priors. We provide theoretical guarantees on convergence and reconstruction accuracy when used within plug-and-play methods. Empirical results across diverse sensing matrices demonstrate that NPN priors consistently enhance reconstruction fidelity in various imaging inverse problems, such as compressive sensing, deblurring, super-resolution, computed tomography, and magnetic resonance imaging, with plug-and-play methods, unrolling networks, deep image prior, and diffusion models.

Robotic food scooping is a critical manipulation skill for food preparation and service robots. However, existing robot learning algorithms, especially learn-from-demonstration methods, still struggle to handle diverse and dynamic food states, which often results in spillage and reduced reliability. In this work, we introduce GRITS: A Spillage-Aware Guided Diffusion Policy for Robot Food Scooping Tasks. This framework leverages guided diffusion policy to minimize food spillage during scooping and to ensure reliable transfer of food items from the initial to the target location. Specifically, we design a spillage predictor that estimates the probability of spillage given current observation and action rollout. The predictor is trained on a simulated dataset with food spillage scenarios, constructed from four primitive shapes (spheres, cubes, cones, and cylinders) with varied physical properties such as mass, friction, and particle size. At inference time, the predictor serves as a differentiable guidance signal, steering the diffusion sampling process toward safer trajectories while preserving task success. We validate GRITS on a real-world robotic food scooping platform. GRITS is trained on six food categories and evaluated on ten unseen categories with different shapes and quantities. GRITS achieves an 82% task success rate and a 4% spillage rate, reducing spillage by over 40% compared to baselines without guidance, thereby demonstrating its effectiveness. More details are available on our project website: https://hcis-lab.github.io/GRITS/.

Choroidal nevi are common benign pigmented lesions in the eye, with a small risk of transforming into melanoma. Early detection is critical to improving survival rates, but misdiagnosis or delayed diagnosis can lead to poor outcomes. Despite advancements in AI-based image analysis, diagnosing choroidal nevi in colour fundus images remains challenging, particularly for clinicians without specialized expertise. Existing datasets often suffer from low resolution and inconsistent labelling, limiting the effectiveness of segmentation models. This paper addresses the challenge of achieving precise segmentation of fundus lesions, a critical step toward developing robust diagnostic tools. While deep learning models like U-Net have demonstrated effectiveness, their accuracy heavily depends on the quality and quantity of annotated data. Previous mathematical/clustering segmentation methods, though accurate, required extensive human input, making them impractical for medical applications. This paper proposes a novel approach that combines mathematical/clustering segmentation models with insights from U-Net, leveraging the strengths of both methods. This hybrid model improves accuracy, reduces the need for large-scale training data, and achieves significant performance gains on high-resolution fundus images. The proposed model achieves a Dice coefficient of 89.7% and an IoU of 80.01% on 1024*1024 fundus images, outperforming the Attention U-Net model, which achieved 51.3% and 34.2%, respectively. It also demonstrated better generalizability on external datasets. This work forms a part of a broader effort to develop a decision support system for choroidal nevus diagnosis, with potential applications in automated lesion annotation to enhance the speed and accuracy of diagnosis and monitoring.

Real-world multi-hop QA is naturally linked with ambiguity, where a single query can trigger multiple reasoning paths that require independent resolution. Since ambiguity can occur at any stage, models must navigate layered uncertainty throughout the entire reasoning chain. Despite its prevalence in real-world user queries, previous benchmarks have primarily focused on single-hop ambiguity, leaving the complex interaction between multi-step inference and layered ambiguity underexplored. In this paper, we introduce MARCH, a benchmark for their intersection, with 2,209 multi-hop ambiguous questions curated via multi-LLM verification and validated by human annotation with strong agreement. Our experiments reveal that even state-of-the-art models struggle with MARCH, confirming that combining ambiguity resolution with multi-step reasoning is a significant challenge. To address this, we propose CLARION, a two-stage agentic framework that explicitly decouples ambiguity planning from evidence-driven reasoning, significantly outperforms existing approaches, and paves the way for robust reasoning systems.

Guidance provides a simple and effective framework for posterior sampling by steering the generation process towards the desired distribution. When modeling discrete data, existing approaches mostly focus on guidance with the first-order approximation to improve the sampling efficiency. However, such an approximation is inappropriate in discrete state spaces since the approximation error could be large. A novel guidance framework for discrete data is proposed to address this problem: we derive the exact transition rate for the desired distribution given a learned discrete flow matching model, leading to guidance that only requires a single forward pass in each sampling step, significantly improving efficiency. This unified novel framework is general enough, encompassing existing guidance methods as special cases, and it can also be seamlessly applied to the masked diffusion model. We demonstrate the effectiveness of our proposed guidance on energy-guided simulations and preference alignment on text-to-image generation and multimodal understanding tasks. The code is available at https://github.com/WanZhengyan/Discrete-Guidance-Matching.

Reward is critical to the evaluation and training of large language models (LLMs). However, existing rule-based or model-based reward methods struggle to generalize to GUI agents, where access to ground-truth trajectories or application databases is often unavailable, and static trajectory-based LLM-as-a-Judge approaches suffer from limited accuracy. To address these challenges, we propose ProRe, a proactive reward system that leverages a general-purpose reasoner and domain-specific evaluator agents (actors). The reasoner schedules targeted state probing tasks, which the evaluator agents then execute by actively interacting with the environment to collect additional observations. This enables the reasoner to assign more accurate and verifiable rewards to GUI agents. Empirical results on over 3K trajectories demonstrate that ProRe improves reward accuracy and F1 score by up to 5.3\% and 19.4\%, respectively. Furthermore, integrating ProRe with state-of-the-art policy agents yields a success rate improvement of up to 22.4\%. The source code is available at https://github.com/V-Droid-Agent/ProRe.

Tool-augmented large language models (LLMs) are usually trained with supervised imitation or coarse-grained reinforcement learning that optimizes single tool calls. Current self-reflection practices rely on heuristic prompts or one-way reasoning: the model is urged to 'think more' instead of learning error diagnosis and repair. This is fragile in multi-turn interactions; after a failure the model often repeats the same mistake. We propose structured reflection, which turns the path from error to repair into an explicit, controllable, and trainable action. The agent produces a short yet precise reflection: it diagnoses the failure using evidence from the previous step and then proposes a correct, executable follow-up call. For training we combine DAPO and GSPO objectives with a reward scheme tailored to tool use, optimizing the stepwise strategy Reflect, then Call, then Final. To evaluate, we introduce Tool-Reflection-Bench, a lightweight benchmark that programmatically checks structural validity, executability, parameter correctness, and result consistency. Tasks are built as mini trajectories of erroneous call, reflection, and corrected call, with disjoint train and test splits. Experiments on BFCL v3 and Tool-Reflection-Bench show large gains in multi-turn tool-call success and error recovery, and a reduction of redundant calls. These results indicate that making reflection explicit and optimizing it directly improves the reliability of tool interaction and offers a reproducible path for agents to learn from failure.

Enabling robotic assistants to navigate complex environments and locate objects described in free-form language is a critical capability for real-world deployment. While foundation models, particularly Vision-Language Models (VLMs), offer powerful semantic understanding, effectively adapting their web-scale knowledge for embodied decision-making remains a key challenge. We present FiLM-Nav (Fine-tuned Language Model for Navigation), an approach that directly fine-tunes pre-trained VLM as the navigation policy. In contrast to methods that use foundation models primarily in a zero-shot manner or for map annotation, FiLM-Nav learns to select the next best exploration frontier by conditioning directly on raw visual trajectory history and the navigation goal. Leveraging targeted simulated embodied experience allows the VLM to ground its powerful pre-trained representations in the specific dynamics and visual patterns relevant to goal-driven navigation. Critically, fine-tuning on a diverse data mixture combining ObjectNav, OVON, ImageNav, and an auxiliary spatial reasoning task proves essential for achieving robustness and broad generalization. FiLM-Nav sets a new state-of-the-art in both SPL and success rate on HM3D ObjectNav among open-vocabulary methods, and sets a state-of-the-art SPL on the challenging HM3D-OVON benchmark, demonstrating strong generalization to unseen object categories. Our work validates that directly fine-tuning VLMs on diverse simulated embodied data is a highly effective pathway towards generalizable and efficient semantic navigation capabilities.

Sparse-view computed tomography (CT) reconstruction is fundamentally challenging due to undersampling, leading to an ill-posed inverse problem. Traditional iterative methods incorporate handcrafted or learned priors to regularize the solution but struggle to capture the complex structures present in medical images. In contrast, diffusion models (DMs) have recently emerged as powerful generative priors that can accurately model complex image distributions. In this work, we introduce Diffusion Consensus Equilibrium (DICE), a framework that integrates a two-agent consensus equilibrium into the sampling process of a DM. DICE alternates between: (i) a data-consistency agent, implemented through a proximal operator enforcing measurement consistency, and (ii) a prior agent, realized by a DM performing a clean image estimation at each sampling step. By balancing these two complementary agents iteratively, DICE effectively combines strong generative prior capabilities with measurement consistency. Experimental results show that DICE significantly outperforms state-of-the-art baselines in reconstructing high-quality CT images under uniform and non-uniform sparse-view settings of 15, 30, and 60 views (out of a total of 180), demonstrating both its effectiveness and robustness.

Autonomous vehicles must navigate dynamically uncertain environments while balancing safety and efficiency. This challenge is exacerbated by unpredictable human-driven vehicle (HV) behaviors and perception inaccuracies, necessitating planners that adapt to evolving uncertainties while maintaining safe trajectories. Overly conservative planning degrades driving efficiency, while deterministic methods risk failure in unexpected scenarios. To address these issues, we propose a real-time contingency trajectory optimization framework. Our method employs event-triggered online learning of HV control-intent sets to dynamically quantify multimodal HV uncertainties and incrementally refine their forward reachable sets (FRSs). Crucially, we enforce invariant safety through FRS-based barrier constraints that ensure safety without reliance on accurate trajectory prediction. These constraints are seamlessly embedded in contingency trajectory optimization and solved efficiently through consensus alternating direction method of multipliers (ADMM). The system continuously adapts to HV behavioral uncertainties, preserving feasibility and safety without excessive conservatism. High-fidelity simulations on highway and urban scenarios, along with a series of real-world experiments, demonstrate significant improvements in driving efficiency and passenger comfort while maintaining safety under uncertainty. The project page is available at https://pathetiue.github.io/frscp.github.io/.

Shortcuts such as APIs and deep-links have emerged as efficient complements to flexible GUI operations, fostering a promising hybrid paradigm for MLLM-based mobile automation. However, systematic evaluation of GUI-shortcut hybrid agents remains largely underexplored. To bridge this gap, we introduce MAS-Bench, a benchmark that pioneers the evaluation of GUI-shortcut hybrid agents with a specific focus on the mobile domain. Beyond merely using predefined shortcuts, MAS-Bench assesses an agent's capability to autonomously generate shortcuts by discovering and creating reusable, low-cost workflows. It features 139 complex tasks across 11 real-world applications, a knowledge base of 88 predefined shortcuts (APIs, deep-links, RPA scripts), and 9 evaluation metrics. Experiments demonstrate that hybrid agents achieve up to 68.3% success rate and 39% greater execution efficiency than GUI-only counterparts. Furthermore, our evaluation framework effectively reveals the quality gap between predefined and agent-generated shortcuts, validating its capability to assess shortcut generation methods. MAS-Bench addresses the lack of systematic benchmarks for GUI-shortcut hybrid mobile agents, providing a foundational platform for future advancements in creating more efficient and robust intelligent agents. Project page: https://pengxiang-zhao.github.io/MAS-Bench.

Federated fine-tuning has emerged as a promising approach for adapting foundation models (FMs) to diverse downstream tasks in edge environments. In Internet of Vehicles (IoV) systems, enabling efficient and low-latency multi-task adaptation is particularly challenging due to client mobility, heterogeneous resources, and intermittent connectivity. This paper proposes a hierarchical federated fine-tuning framework that coordinates roadside units (RSUs) and vehicles to support resource-aware and mobility-resilient learning across dynamic IoV scenarios. Leveraging Low-Rank Adaptation (LoRA), we introduce a decentralized, energy-aware rank adaptation mechanism formulated as a constrained multi-armed bandit problem. A novel UCB-DUAL algorithm is developed to enable adaptive exploration under per-task energy budgets, achieving provable sublinear regret. To evaluate our method, we construct a large-scale IoV simulator based on real-world trajectories, capturing dynamic participation, RSU handoffs, and communication variability. Extensive experiments show that our approach achieves the best accuracy-efficiency trade-off among all baselines, reducing latency by over 24\% and improving average accuracy by more than 2.5\%.

Effective tool use is essential for large language models (LLMs) to interact with their environment. However, progress is limited by the lack of efficient reinforcement learning (RL) frameworks specifically designed for tool use, due to challenges in constructing stable training environments and designing verifiable reward mechanisms. To address this, we propose an automated environment construction pipeline, incorporating scenario decomposition, document generation, function integration, complexity scaling, and localized deployment. This enables the creation of high-quality training environments that provide detailed and measurable feedback without relying on external tools. Additionally, we introduce a verifiable reward mechanism that evaluates both the precision of tool use and the completeness of task execution. When combined with trajectory data collected from the constructed environments, this mechanism integrates seamlessly with standard RL algorithms to facilitate feedback-driven model training. Experiments on LLMs of varying scales demonstrate that our approach significantly enhances the models' tool-use performance without degrading their general capabilities. Our analysis suggests that these gains result from improved context understanding and reasoning, driven by updates to the lower-layer MLP parameters in models. Code and data are available at https://github.com/bytedance/FTRL.

Reinforcement Learning with Verifiable Reward (RLVR) has significantly advanced the complex reasoning abilities of Large Language Models (LLMs). However, it struggles to break through the inherent capability boundaries of the base LLM, due to its essentially on-policy strategy coupled with LLM's immense action space and sparse reward. Critically, RLVR can lead to the capability boundary collapse, narrowing the LLM's problem-solving scope. To address this problem, we propose RL-PLUS, a novel hybrid-policy optimization approach for LLMs that synergizes internal exploitation with external data to achieve stronger reasoning capabilities and surpass the boundaries of base models. RL-PLUS integrates two core components, i.e., Multiple Importance Sampling to address distributional mismatch from external data, and Exploration-Based Advantage Function to guide the model towards high-value, unexplored reasoning paths. We provide both theoretical analysis and extensive experiments to demonstrate the superiority and generalizability of our approach. Compared with existing RLVR methods, RL-PLUS achieves 1) state-of-the-art performance on six math reasoning benchmarks; 2) superior performance on six out-of-distribution reasoning tasks; 3) consistent and significant gains across diverse model families, with average relative improvements up to 69.2\%. Moreover, the analysis of Pass@k curves indicates that RL-PLUS effectively resolves the capability boundary collapse problem.

The poultry industry has been driven by broiler chicken production and has grown into the world's largest animal protein sector. Automated detection of chicken carcasses on processing lines is vital for quality control, food safety, and operational efficiency in slaughterhouses and poultry processing plants. However, developing robust deep learning models for tasks like instance segmentation in these fast-paced industrial environments is often hampered by the need for laborious acquisition and annotation of large-scale real-world image datasets. We present the first pipeline generating photo-realistic, automatically labeled synthetic images of chicken carcasses. We also introduce a new benchmark dataset containing 300 annotated real-world images, curated specifically for poultry segmentation research. Using these datasets, this study investigates the efficacy of synthetic data and automatic data annotation to enhance the instance segmentation of chicken carcasses, particularly when real annotated data from the processing line is scarce. A small real dataset with varying proportions of synthetic images was evaluated in prominent instance segmentation models. Results show that synthetic data significantly boosts segmentation performance for chicken carcasses across all models. This research underscores the value of synthetic data augmentation as a viable and effective strategy to mitigate data scarcity, reduce manual annotation efforts, and advance the development of robust AI-driven automated detection systems for chicken carcasses in the poultry processing industry.

Integrating compositional and symbolic properties into current distributional semantic spaces can enhance the interpretability, controllability, compositionality, and generalisation capabilities of Transformer-based auto-regressive language models (LMs). In this survey, we offer a novel perspective on latent space geometry through the lens of compositional semantics, a direction we refer to as \textit{semantic representation learning}. This direction enables a bridge between symbolic and distributional semantics, helping to mitigate the gap between them. We review and compare three mainstream autoencoder architectures-Variational AutoEncoder (VAE), Vector Quantised VAE (VQVAE), and Sparse AutoEncoder (SAE)-and examine the distinctive latent geometries they induce in relation to semantic structure and interpretability.

Accurately solving partial differential equations (PDEs) is essential across many scientific disciplines. However, high-fidelity solvers can be computationally prohibitive, motivating the development of reduced-order models (ROMs). Recently, Latent Space Dynamics Identification (LaSDI) was proposed as a data-driven, non-intrusive ROM framework. LaSDI compresses the training data via an autoencoder and learns user-specified ordinary differential equations (ODEs), governing the latent dynamics, enabling rapid predictions for unseen parameters. While LaSDI has produced effective ROMs for numerous problems, the autoencoder must simultaneously reconstruct the training data and satisfy the imposed latent dynamics, which are often competing objectives that limit accuracy, particularly for complex or high-frequency phenomena. To address this limitation, we propose multi-stage Latent Space Dynamics Identification (mLaSDI). With mLaSDI, we train LaSDI sequentially in stages. After training the initial autoencoder, we train additional decoders which map the latent trajectories to residuals from previous stages. This staged residual learning, combined with periodic activation functions, enables recovery of high-frequency content without sacrificing interpretability of the latent dynamics. We further provide an error decomposition separating autoencoder and latent dynamics contributions, and prove that additional training stages cannot increase the training residual. Numerical experiments on a multiscale oscillating system, unsteady wake flow, and the 1D-1V Vlasov equation demonstrate that mLaSDI achieves significantly lower reconstruction and prediction errors, often by an order of magnitude, while requiring less training time and reduced hyperparameter tuning compared to standard LaSDI.

Learning dynamical systems that respect physical symmetries and constraints remains a fundamental challenge in data-driven modeling. Integrating physical laws with graph neural networks facilitates principled modeling of complex N-body dynamics and yields accurate and permutation-invariant models. However, training graph neural networks with iterative, gradient-descent-based optimization algorithms (e.g., Adam, RMSProp, LBFGS) often leads to slow training, especially for large, complex systems. In comparison to 15 different optimizers, we demonstrate that Hamiltonian Graph Networks (HGN) can be trained 150-600x faster - but with comparable accuracy - by replacing iterative optimization with random feature-based parameter construction. We show robust performance in diverse simulations, including N-body mass-spring and molecular dynamics systems in up to dimensions and 10,000 particles with different geometries, while retaining essential physical invariances with respect to permutation, rotation, and translation. Our proposed approach is benchmarked using a NeurIPS 2022 Datasets and Benchmarks Track publication to further demonstrate its versatility. We reveal that even when trained on minimal 8-node systems, the model can generalize in a zero-shot manner to systems as large as 4096 nodes without retraining. Our work challenges the dominance of iterative gradient-descent-based optimization algorithms for training neural network models for physical systems.

Large Language Model (LLM) agents are reshaping the game industry, by enabling more intelligent and human-preferable characters. Yet, current game benchmarks fall short of practical needs: they lack evaluations of diverse LLM capabilities across various game genres, studies of agentic modules crucial for complex gameplay, and fine-tuning datasets to adapt pre-trained LLMs into gaming agents. To fill these gaps, we present Orak, a benchmark for training and evaluating LLM agents across 12 popular video games spanning all major genres. Using a plug-and-play interface built on Model Context Protocol (MCP), Orak supports systematic and reproducible studies of agentic modules in varied game scenarios. We further release a fine-tuning dataset of expert LLM gameplay trajectories covering multiple genres, turning general LLMs into effective game agents. Orak offers a united evaluation framework, including game leaderboards, LLM battle arenas, and \fix{ablation studies} of input modality, agentic strategies, and fine-tuning effects, establishing a foundation towards versatile gaming agents. Code and datasets are available at https://github.com/krafton-ai/Orak and https://huggingface.co/datasets/KRAFTON/Orak.

Modern learning-based locomotion controllers typically rely on fully trainable deep neural networks with a large number of parameters. This paper studies a different design point for end-to-end control: whether effective quadruped locomotion can be achieved with a drastically reduced trainable parameter space. We present RANDomized POlicy Learning (RANDPOL), a policy learning approach in which the hidden layers of the actor and critic are randomly initialized and fixed, while only the final linear readout is trained. This yields a parameter-efficient controller class that retains nonlinear expressiveness through a fixed random basis while substantially reducing the dimension of the optimization problem. RANDPOL is supported by the mathematical foundation of randomized function approximation, which provides a principled basis for using fixed random nonlinear features as expressive function classes. We evaluate RANDPOL on end-to-end locomotion control for the Unitree Go2 quadruped and compare it with Proximal Policy Optimization (PPO). The results show that RANDPOL attains comparative locomotion performance with far fewer trainable parameters, lower learning-phase computation time per iteration, and a favorable performance-complexity trade-off. We further demonstrate successful zero-shot sim-to-real transfer of the learned RANDPOL controller on the physical Unitree Go2 under user-issued forward-velocity and yaw-rate commands. These results indicate that, for structured robotic control problems, reducing trainable complexity can remain compatible with effective simulated and real-world performance.

This paper introduces LAV (Latent Audio-Visual), a system that integrates EnCodec's neural audio compression with StyleGAN2's generative capabilities to produce visually dynamic outputs driven by pre-recorded audio. Unlike previous works that rely on explicit feature mappings, LAV uses EnCodec embeddings as latent representations, directly transformed into StyleGAN2's style latent space via randomly initialized linear mapping. This approach preserves semantic richness in the transformation, enabling nuanced and semantically coherent audio-visual translations. The framework demonstrates the potential of using pretrained audio compression models for artistic and computational applications.

Instruction following refers to the ability of large language models (LLMs) to generate outputs that satisfy all specified constraints. Existing research has primarily focused on constraint categories, offering limited evaluation dimensions and little guidance for improving instruction-following abilities. To address this gap, we introduce MulDimIF, a multi-dimensional constraint framework encompassing three constraint patterns, four constraint categories, and four difficulty levels. Based on this framework, we design a controllable instruction generation pipeline. Through constraint expansion, conflict detection, and instruction rewriting, we construct 9,106 code-verifiable samples. We evaluate 18 LLMs from six model families and find marked performance differences across constraint settings. For instance, average accuracy decreases from 80.82% at Level I to 36.76% at Level IV. Moreover, training with data generated by our framework significantly improves instruction following without compromising general performance. In-depth analysis indicates that these gains stem largely from parameter updates in attention modules, which strengthen constraint recognition and adherence. Code and data are available in https://github.com/Junjie-Ye/MulDimIF.

In many practical sequential decision-making problems, tracking the state of the environment incurs a sensing/communication/computation cost. In these settings, the agent's interaction with its environment includes the additional component of deciding when to sense the state, in a manner that balances the value associated with optimal (state-specific) actions and the cost of sensing. We formulate this as an expected discounted cost Markov Decision Process (MDP), wherein the agent incurs an additional cost for sensing its next state, but has the option to take actions while remaining `blind' to the system state. We pose this problem as a classical discounted cost MDP with an expanded (countably infinite) state space. While computing the optimal policy for this MDP is intractable in general, we derive lower bounds on the optimal value function, which allow us to bound the suboptimality gap of any policy. We also propose a computationally efficient algorithm SPI, based on policy improvement, which in practice performs close to the optimal policy. Finally, we benchmark against the state-of-the-art via a numerical case study.

To address the challenge of scarce computational resources in genomic modeling, we introduce GERM, a genomic foundation model with strong compression performance and fast adaptability. GERM improves upon models like DNABERT-2 by eliminating outliers that hinder low-rank adaptation and post-training quantization, enhancing both efficiency and robustness. We replace the vanilla attention layer with an outlier-free mechanism inspired by associative memory models. By removing outliers during both pre-training and fine-tuning, this approach accelerates adaptation, reduces computational costs, and enhances quantization robustness within acceptable loss margins. Additionally, we propose GERM-T, a strategy that employs small-step continual learning within the outlier-free framework, leveraging original checkpoints to avoid retraining from scratch. Empirically, GERM improves fine-tuning performance by 37.98% and quantization by 64.34% over the baseline model. It also reduces average kurtosis by 92.14% and maximum infinity norm by 82.77%. Compared to leading methods, GERM consistently delivers superior performance, offering a practical solution for genomic modeling in resource-constrained settings. Code is available at https://github.com/MAGICS-LAB/GERM.

As large language models (LLMs) and LLM-based agents increasingly interact with humans in decision-making contexts, understanding the trust dynamics between humans and AI agents becomes a central concern. While considerable literature studies how humans trust AI agents, it is much less understood how LLM-based agents develop effective trust in humans. LLM-based agents likely rely on some sort of implicit effective trust in trust-related contexts (e.g., evaluating individual loan applications) to assist and affect decision making. Using established behavioral theories, we develop an approach that studies whether LLMs trust depends on the three major trustworthiness dimensions: competence, benevolence and integrity of the human subject. We also study how demographic variables affect effective trust. Across 43,200 simulated experiments, for five popular language models, across five different scenarios we find that LLM trust development shows an overall similarity to human trust development. We find that in most, but not all cases, LLM trust is strongly predicted by trustworthiness, and in some cases also biased by age, religion and gender, especially in financial scenarios. This is particularly true for scenarios common in the literature and for newer models. While the overall patterns align with human-like mechanisms of effective trust formation, different models exhibit variation in how they estimate trust; in some cases, trustworthiness and demographic factors are weak predictors of effective trust. These findings call for a better understanding of AI-to-human trust dynamics and monitoring of biases and trust development patterns to prevent unintended and potentially harmful outcomes in trust-sensitive applications of AI.

Understanding humor-particularly when it involves complex, contradictory narratives that require comparative reasoning-remains a significant challenge for large vision-language models (VLMs). This limitation hinders AI's ability to engage in human-like reasoning and cultural expression. In this paper, we investigate this challenge through an in-depth analysis of comics that juxtapose panels to create humor through contradictions. We introduce the YesBut (V2), a novel benchmark with 1,262 comic images from diverse multilingual and multicultural contexts, featuring comprehensive annotations that capture various aspects of narrative understanding. Using this benchmark, we systematically evaluate a wide range of VLMs through four complementary tasks spanning from surface content comprehension to deep narrative reasoning, with particular emphasis on comparative reasoning between contradictory elements. Our extensive experiments reveal that even the most advanced models significantly underperform compared to humans, with common failures in visual perception, key element identification, comparative analysis and hallucinations. We further investigate text-based training strategies and social knowledge augmentation methods to enhance model performance. Our findings not only highlight critical weaknesses in VLMs' understanding of cultural and creative expressions but also provide pathways toward developing context-aware models capable of deeper narrative understanding though comparative reasoning.