Click-Through Rate (CTR) prediction, a core task in recommendation systems, estimates user click likelihood using historical behavioral data. Modeling user behavior sequences as text to leverage Language Models (LMs) for this task has gained traction, owing to LMs' strong semantic understanding and contextual modeling capabilities. However, a critical structural gap exists: user behavior sequences consist of discrete actions connected by semantically empty separators, differing fundamentally from the coherent natural language in LM pre-training. This mismatch causes semantic fragmentation, where LM attention scatters across irrelevant tokens instead of focusing on meaningful behavior boundaries and inter-behavior relationships, degrading prediction performance. To address this, we propose $\textit{CTR-Sink}$, a novel framework introducing behavior-level attention sinks tailored for recommendation scenarios. Inspired by attention sink theory, it constructs attention focus sinks and dynamically regulates attention aggregation via external information. Specifically, we insert sink tokens between consecutive behaviors, incorporating recommendation-specific signals such as temporal distance to serve as stable attention sinks. To enhance generality, we design a two-stage training strategy that explicitly guides LM attention toward sink tokens and a attention sink mechanism that amplifies inter-sink dependencies to better capture behavioral correlations. Experiments on one industrial dataset and two open-source datasets (MovieLens, Kuairec), alongside visualization results, validate the method's effectiveness across scenarios.

Increasing concerns for data privacy and other difficulties associated with retrieving source data for model training have created the need for source-free transfer learning, in which one only has access to pre-trained models instead of data from the original source domains. This setting introduces many challenges, as many existing transfer learning methods typically rely on access to source data, which limits their direct applicability to scenarios where source data is unavailable. Further, practical concerns make it more difficult, for instance efficiently selecting models for transfer without information on source data, and transferring without full access to the source models. So motivated, we propose a model recycling framework for parameter-efficient training of models that identifies subsets of related source models to reuse in both white-box and black-box settings. Consequently, our framework makes it possible for Model as a Service (MaaS) providers to build libraries of efficient pre-trained models, thus creating an opportunity for multi-source data-free supervised transfer learning.

The general trace reconstruction problem seeks to recover an original sequence from its noisy copies independently corrupted by insertions, deletions, and substitutions. This problem arises in applications such as DNA data storage, a promising storage medium due to its high information density and longevity. However, errors introduced during DNA synthesis, storage, and sequencing require correction through algorithms and codes, with trace reconstruction often used as part of data retrieval. In this work, we propose TReconLM, a decoder-only transformer that solves trace reconstruction as a next-token prediction task. TReconLM outperforms state-of-the-art trace reconstruction algorithms, including prior deep-learning approaches, recovering a substantially higher fraction of sequences without error. We pretrain on synthetic data generated from a simple error model and fine-tune on real-world data to adapt to technology-specific error patterns. Code is available at https://github.com/MLI-lab/TReconLM.

The rapid expansion of chemistry literature poses significant challenges for researchers seeking to efficiently access domain-specific knowledge. To support advancements in chemistry-focused natural language processing (NLP), we present ChemQuests, a curated dataset of 952 high-quality question-answer (QA) pairs derived from 155 ChemRxiv \cite{chemrxivWebsite} papers across 17 subfields of chemistry. Each QA pair is explicitly linked to its source text segment to ensure traceability and contextual accuracy. ChemQuests was constructed using an automated pipeline that combines optical character recognition (OCR), QA generation using GPT-4o, and fuzzy-search verification. The dataset emphasizes conceptual, mechanistic, applied, and synthetic or experimental questions, enabling applications in retrieval-based QA systems, search engine development, and fine-tuning of domain-adapted large language models. We analyze the dataset's structure, coverage, and limitations, and outline future directions for expansion and expert validation. ChemQuests provides a foundational resource for chemistry NLP research, education, and tool development.

Mixed Reality (MR) interfaces have been extensively explored for controlling mobile robots, but there is limited research on their application to managing teams of robots. This paper presents HORUS: Holistic Operational Reality for Unified Systems, a Mixed Reality interface offering a comprehensive set of tools for managing multiple mobile robots simultaneously. HORUS enables operators to monitor individual robot statuses, visualize sensor data projected in real time, and assign tasks to single robots, subsets of the team, or the entire group, all from a Mini-Map (Ground Station). The interface also provides different teleoperation modes: a mini-map mode that allows teleoperation while observing the robot model and its transform on the mini-map, and a semi-immersive mode that offers a flat, screen-like view in either single or stereo view (3D). We conducted a user study in which participants used HORUS to manage a team of mobile robots tasked with finding clues in an environment, simulating search and rescue tasks. This study compared HORUS's full-team management capabilities with individual robot teleoperation. The experiments validated the versatility and effectiveness of HORUS in multi-robot coordination, demonstrating its potential to advance human-robot collaboration in dynamic, team-based environments.

Multimodal Large Language Models (MLLMs) have achieved remarkable success in instruction-following tasks by integrating pretrained visual encoders with large language models (LLMs). However, existing approaches often struggle with fine-grained visual grounding due to semantic entanglement in visual patch representations, where individual patches blend multiple distinct visual elements, making it difficult for models to focus on instruction-relevant details. To address this challenge, we propose MoDA (Modulation Adapter), a lightweight module that enhances visual grounding through instruction-guided channel-wise modulation. Unlike token-level methods such as Q-Former that perform additive feature selection, MoDA operates at the channel level through multiplicative modulation on already-aligned features, enabling fine-grained control over which embedding dimensions are relevant for each instruction. Following the standard LLaVA training protocol, MoDA applies cross-attention between language instructions and pre-aligned visual features, generating dynamic modulation masks without architectural modifications or additional supervision. We evaluate MoDA across 12 benchmarks spanning visual question answering, vision-centric reasoning, and hallucination detection, including recent 2024 benchmarks (MMVP, CV-Bench, MMStar, RealWorldQA), on three distinct MLLM architectures: LLaVA-1.5, LLaVA-MoRE (2025), and Qwen3-VL (2025). MoDA delivers consistent gains across all three families, with +12.0 points on MMVP for the LLaVA-1.5 family and +4.8 points on ScienceQA for the LLaVA-MoRE family, and +4.9 ScienceQA, +4.1 RealWorldQA, and +3.8 GQA on Qwen3-VL, confirming that the gains generalize beyond CLIP-based encoders with minimal overhead (<1% FLOPs). Code is available at https://github.com/waybarrios/MoDA.

Ranking systems influence decision-making in high-stakes domains like health, education, and employment, where they can have substantial economic and social impacts. This makes the integration of safety mechanisms essential. One such mechanism is abstention, which enables algorithmic decision-making systems to defer uncertain or low-confidence decisions to human experts. While abstention has been predominantly explored in the context of classification tasks, its application to other machine learning paradigms remains underexplored. In this paper, we introduce a novel method for abstention in pairwise learning-to-rank tasks. Our approach is based on thresholding the ranker's conditional risk: the system abstains from making a decision when the estimated risk exceeds a predefined threshold. Our contributions are threefold: a theoretical characterization of the optimal abstention strategy, a model-agnostic, plug-in algorithm for constructing abstaining ranking models, and a comprehensive empirical evaluation across multiple datasets, demonstrating the effectiveness of our approach.

We study the problem of assigning operations in a dataflow graph to devices to minimize execution time in a work-conserving system, with emphasis on complex machine learning workloads. Prior learning-based methods often struggle due to three key limitations: (1) reliance on bulk-synchronous systems like TensorFlow, which under-utilize devices due to barrier synchronization; (2) lack of awareness of the scheduling mechanism of underlying systems when designing learning-based methods; and (3) exclusive dependence on reinforcement learning, ignoring the structure of effective heuristics designed by experts. In this paper, we propose Doppler, a three-stage framework for training dual-policy networks consisting of 1) a $\mathsf{SEL}$ policy for selecting operations and 2) a $\mathsf{PLC}$ policy for placing chosen operations on devices. Our experiments show that Doppler outperforms all baseline methods across tasks by reducing system execution time and additionally demonstrates sampling efficiency by reducing per-episode training time.

Graphical User Interface (GUI) agents powered by Multimodal Large Language Models (MLLMs) are increasingly deployed yet vulnerable to Environmental Injection Attacks (EIAs).However, current red-teaming methods are hindered by prohibitive computational costs and limited adaptability. A fundamental question remains unaddressed: does the bottleneck of attack success lie in visual perception or semantic understanding? Through controlled experiments, we observe that semantic deception, rather than visual appearance, serves as the primary determinant of attack success. Based on this insight, we introduce EVA, an evolutionary framework that evolves adversarial payloads exclusively within the semantic dimension. EVA employs a discovery-deployment framework to mine linguistic vulnerability patterns and distill them into generalizable rules. Experimental results across five representative victim agents demonstrate that EVA achieves up to 85\% attack success rate, evolving benign seeds into successful attacks within only 1.18 to 1.71 iterations. This rapid convergence uncovers a dense semantic attack space in the model's latent representation, unveiling a critical alignment paradox: the instruction-following capabilities reinforced by alignment training render agents inherently susceptible to authoritative, semantically deceptive environmental cues.

In Continual Learning (CL), using a Pre-Trained Model (PTM) as the feature extractor has become a popular practice. Accompanied by analytic classifiers, the PTM-based methods have achieved state-of-the-art performance in CL, in pursuit of the non-forgetting goal. Meanwhile, actively forgetting specific knowledge acquired during the CL phase is also essential in most service construction paradigms, for example, Mobile Crowd Sensing (MCS), where mobile edge nodes continuously collect sensory data and demand not only non-forgetting adaptation but also specific knowledge forgetting for privacy preservation. Thus, a unique problem, called Continual Unlearning (CU), arises when the forgetting requests show sequentially in CL. However, existing unlearning methods focus on single-shot joint forgetting and prove highly inadequate when applied to CU, including (1) violating the historical data privacy in CL and (2) vulnerably being overwhelmed or degraded with adversarially frequent requests. To handle the challenges of CU, we propose a gradient-free approach, called Analytic Continual Unlearning (ACU), for efficient and exact forgetting with historical data privacy preservation in PTM-based CL. In response to each unlearning request, our ACU recursively derives the analytical (i.e., closed-form) solutions via least squares in an interpretable manner. By meticulous design, our ACU is compatible with both sample-level and class-level unlearning requests. The theoretical and experimental evaluations validate our ACU's superiority in unlearning effectiveness, model fidelity, and system efficiency.

Post-training LLMs with RLHF and preference optimization methods (e.g., DPO, IPO) has greatly improved alignment, yet these approaches assume a single objective. In reality, humans express multiple, often conflicting objectives, such as helpfulness and harmlessness, with no natural scalarization. We study the multi-objective preference alignment problem, where a policy must balance several objectives simultaneously. We propose Multi-Objective Preference Optimization (MOPO), a constrained KL-regularized framework that maximizes a primary objective while enforcing lower bounds on secondary objectives via tunable safety thresholds. MOPO operates directly on pairwise preferences without point-wise rewards, and admits simple closed-form iterative updates. Empirically, MOPO recovers Pareto-optimal policies on synthetic benchmarks and, when fine-tuned on human-preference data, yields multi-billion parameter models that achieve higher rewards and Pareto-dominate baselines, with stable and robust optimization dynamics.

Work-Related Musculoskeletal Disorders continue to be a major challenge in industrial environments, leading to reduced workforce participation, increased healthcare costs, and long-term disability. This study introduces a human-sensitive robotic system aimed at reintegrating individuals with a history of musculoskeletal disorders into standard job roles, while simultaneously optimizing ergonomic conditions for the broader workforce. This research leverages reinforcement learning (RL) to develop a human-aware control strategy for collaborative robots, focusing on optimizing ergonomic conditions and preventing pain during task execution. Two RL approaches, Q-Learning and Deep Q-Network (DQN), were implemented and tested to personalize control strategies based on individual user characteristics. Although experimental results revealed a simulation-to-real gap, a fine-tuning phase successfully adapted the policies to real-world conditions. DQN outperformed Q-Learning by completing tasks faster while maintaining zero pain risk and safe ergonomic levels, achieving on average 38% shorter task completion times across all tested anthropometries. The structured testing protocol confirmed the system's adaptability to diverse human anthropometries, underscoring the potential of RL-driven cobots to enable safer, more inclusive workplaces.

Finding deletion-correcting codes of maximum size has been an open problem for over 70 years, even for a single deletion. We adapt FunSearch, a large language model (LLM)-guided evolutionary search, to discover functions that construct deletion-correcting codes at short code lengths. For a single deletion, our search finds a function that we prove constructs the conjectured-optimal Varshamov-Tenengolts code. For multiple deletions and quaternary edit codes, the discovered functions improve on prior explicit, search-based, and neural constructions but remain empirical heuristics without new theoretical insights. We study design choices for LLM-guided evolutionary search and find that, for our problem, compute is better allocated to sampling more functions than to longer reasoning traces per function, and that co-evolving natural language descriptions with code hurts search quality. We propose deduplicating logically identical functions during evolution, which we find critical for search diversity. Our results demonstrate the potential of LLM-guided evolutionary search for information theory and code design and represent the first application of such methods for constructing error-correcting codes. However, in our current formulation, evaluating a function scales exponentially with code length, limiting the approach to short codes.

We consider multi-robot systems under recurring tasks formalized as linear temporal logic (LTL) specifications. To solve the planning problem efficiently, we propose a bottom-up approach combining offline plan synthesis with online coordination, dynamically adjusting plans via real-time communication. To address action delays, we introduce a synchronization mechanism ensuring coordinated task execution, leading to a multi-agent coordination and synchronization framework that is adaptable to a wide range of multi-robot applications. The software package is developed in Python and ROS2 for broad deployment. We validate our findings through lab experiments involving nine robots showing enhanced adaptability compared to previous methods. Additionally, we conduct simulations with up to ninety agents to demonstrate the reduced computational complexity and the scalability features of our work.

Machine unlearning, the process of selectively removing data from trained models, is increasingly crucial for addressing privacy concerns and knowledge gaps post-deployment. Despite this importance, existing approaches are often heuristic and lack formal guarantees. In this paper, we analyze the fundamental utility, time, and space complexity trade-offs of approximate unlearning, providing rigorous certification analogous to differential privacy. For in-distribution forget data -- data similar to the retain set -- we show that a surprisingly simple and general procedure, empirical risk minimization with output perturbation, achieves tight unlearning-utility-complexity trade-offs, addressing a previous theoretical gap on the separation from unlearning "for free" via differential privacy, which inherently facilitates the removal of such data. However, such techniques fail with out-of-distribution forget data -- data significantly different from the retain set -- where unlearning time complexity can exceed that of retraining, even for a single sample. To address this, we propose a new robust and noisy gradient descent variant that provably amortizes unlearning time complexity without compromising utility.

The KV cache in large language models is a dominant factor in memory usage, limiting their broader applicability. Quantizing the cache to lower bit widths is an effective way to reduce computational costs; however, previous methods struggle with quantizing key vectors due to outliers, resulting in excessive overhead. We propose a novel quantization approach called PolarQuant, which efficiently addresses the outlier challenge. We observe that outliers typically appear in only one of two dimensions, which are rotated together by a specific angle when rotary position embeddings are applied. When represented as two-dimensional vectors, these dimensions exhibit well-structured patterns, with radii and angles smoothly distributed in polar coordinates. This alleviates the challenge of outliers on per-channel quantization, making them well-suited for quantization. Thus, PolarQuant divides key vectors into groups of two-dimensional sub-vectors, encoding them as the corresponding quantized radius and the polar angle, rather than quantizing original key vectors directly. PolarQuant achieves the superior efficiency in KV cache quantization and accelerates the decoding process by turning the query-key inner product into a table lookup, all while maintaining the downstream performance of full-precision models.

This article presents an error-state Linear Quadratic Regulator (LQR) formulation for robust trajectory tracking in quadrotor Unmanned Aerial Vehicles (UAVs). The proposed approach leverages error-state dynamics and employs exponential coordinates to represent orientation errors, enabling a linearized system representation for real-time control. The control strategy integrates an LQR-based full-state feedback controller for trajectory tracking, combined with a cascaded bodyrate controller to handle actuator dynamics. Detailed derivations of the error-state dynamics, the linearization process, and the controller design are provided, highlighting the applicability of the method for precise and stable quadrotor control in dynamic environments.

Modern machine learning pipelines leverage large amounts of public data, making it infeasible to guarantee data quality and leaving models open to poisoning and backdoor attacks. Provably bounding model behavior under such attacks remains an open problem. In this work, we address this challenge by developing the first framework providing provable guarantees on the behavior of models trained with potentially manipulated data without modifying the model or learning algorithm. In particular, our framework certifies robustness against untargeted and targeted poisoning, as well as backdoor attacks, for bounded and unbounded manipulations of the training inputs and labels. Our method leverages convex relaxations to over-approximate the set of all possible parameter updates for a given poisoning threat model, allowing us to bound the set of all reachable parameters for any gradient-based learning algorithm. Given this set of parameters, we provide bounds on worst-case behavior, including model performance and backdoor success rate. We demonstrate our approach on multiple real-world datasets from applications including energy consumption, medical imaging, and autonomous driving.

Many scientific and engineering problems involve observing a common phenomenon through multiple heterogeneous sensors or measurement modalities. Such observations typically contain both information shared across sensors, reflecting the underlying system, and sensor-specific or extraneous components arising from measurement processes or environmental effects. Disentangling these contributions is essential when sensor-independent observations are unavailable. We propose a neural autoencoder framework that explicitly separates shared and sensor-specific latent variables from multi-sensor data. The architecture enforces geometric independence between latent components through structural constraints and orthogonality-based regularization, yielding interpretable and disentangled representations. Building on this representation, we then introduce a latent-space generative methodology in which generative models are tuned/"restricted" on selected disentangled latent subspaces; we then constructively combine disentangled observed latent variables to conditionally synthesize new samples via trained decoders. This enables consistent data generation with prescribed shared (or sensor-specific) characteristics. It also supports cross-sensor inference by consistently sampling distributions over plausible measurements in unobserved modalities. We demonstrate the approach on several computational examples, showing effective disentanglement, targeted data generation, and modality imputation in heterogeneous sensing settings.

In this paper, we address the challenging problem of long-term 3D human motion generation. Specifically, we aim to generate a long sequence of smoothly connected actions from a stream of multiple sentences (i.e., paragraph). Previous long-term motion generating approaches were mostly based on recurrent methods, using previously generated motion chunks as input for the next step. However, this approach has two drawbacks: 1) it relies on sequential datasets, which are expensive; 2) these methods yield unrealistic gaps between motions generated at each step. To address these issues, we introduce simple yet effective T2LM, a continuous long-term generation framework that can be trained without sequential data. T2LM comprises two components: a 1D-convolutional VQVAE, trained to compress motion to sequences of latent vectors, and a Transformer-based Text Encoder that predicts a latent sequence given an input text. At inference, a sequence of sentences is translated into a continuous stream of latent vectors. This is then decoded into a motion by the VQVAE decoder; the use of 1D convolutions with a local temporal receptive field avoids temporal inconsistencies between training and generated sequences. This simple constraint on the VQ-VAE allows it to be trained with short sequences only and produces smoother transitions. T2LM outperforms prior long-term generation models while overcoming the constraint of requiring sequential data; it is also competitive with SOTA single-action generation models.

Recent advances have shifted the paradigm of tabular learning toward tabular foundation models, yet their accuracy relies on a heavy inference cost that scales poorly with context size. Deep neural networks remain a highly competitive and more efficient modeling paradigm when equipped with well-designed architectures; however, identifying such architectures in a data-adaptive and budget-aware manner remains challenging. We propose pTNAS, the first progressive neural architecture search (NAS) approach tailored for tabular data, which enables fast identification of a viable architecture and continuously improves its search performance as more budget becomes available. pTNAS adopts a filter-and-refine optimization strategy that combines efficient training-free and effective training-based architecture evaluation. In the filtering phase, we introduce pTProxy, a novel zero-cost proxy specifically designed for tabular networks that jointly captures architectural trainability and expressivity, enabling fast filtering of large architecture search spaces. In the refinement phase, pTNAS employs a fixed-budget scheduling algorithm to accurately identify the best-performing architecture from a small set of promising candidates. We further propose a budget-aware coordinator to optimize budget allocation holistically. Experiments show that pTNAS reduces the time to reach the globally best architecture by up to 82.75 X compared with other NAS approaches, achieves the best average predictive rank, and improves end-to-end efficiency by up to 4.78 X compared with TabPFN.

We introduce Tune without Validation (Twin), a simple and effective pipeline for tuning learning rate and weight decay of homogeneous classifiers without validation sets, eliminating the need to hold out data and avoiding the two-step process. Twin leverages the margin-maximization dynamics of homogeneous networks and an empirical scaling law that links training and test losses across hyper-parameter configurations. This mathematical modeling yields a regime-dependent, validation-free selection rule: in the non-separable regime, training loss is monotonic in test loss and therefore predictive of generalization, whereas in the separable regime, the parameters' norm becomes a reliable indicator of generalization due to margin maximization. Across 37 dataset-architecture configurations for image classification, we demonstrate that Twin achieves a mean absolute error of 1.28% compared to an Oracle baseline that selects HPs using test accuracy. We demonstrate Twin's benefits in scenarios where validation data is scarce, such as small-data regimes, or difficult and costly to collect, as in medical imaging. Code available at https://github.com/lorenzobrigato/twin.

A common way to analyze learning of statistical models is to consider operations in the models parameter space, however this becomes challenging when there is no one-to-one mapping between the parameter space and the underlying statistical model space. Such ``singular models'' occur frequently and exhibit a characteristic decrease in convergence speed of learning trajectories due to attractor behaviors. In this work, we consider a relative reparameterization technique of the parameter space, which yields a general method for extracting regular sub-models from singular models. On the example of Gaussian Mixture Models and Neural Networks we theoretically and numerically analyze the convergence rate for Gradient Descent under both parameterizations. Analyzing second-order methods and explicit properties of the Fisher Information Matrix we distinguish between differences in convergence behavior arising from algorithmic and intrinsic information-geometric aspects.

Hyperspectral imaging provides rich spectral information for quantitative remote sensing, yet hyperspectral sensors remain costly and thus unavailable in many UAV deployments. Spectral super-resolution (SSR) seeks to reconstruct hyperspectral images (HSIs) from multispectral images (MSIs). Most existing SSR methods assume a fixed and known spectral response function (SRF) and are therefore limited to single-sensor settings. In practical cross-sensor scenarios, the spectral degradation from HSI to MSI is unknown and varies with sensor characteristics and scene content, which renders HSI reconstruction ill-posed. This paper proposes a physics-guided deep unfolding network, termed PGU-Net, to address blind cross-sensor SSR by jointly estimating the HSI and a learnable spectral transformation function (STF). PGU-Net unrolls an alternating optimization procedure into an end-to-end trainable architecture with stages, where each stage sequentially updates the HSI and the STF. Both modules combine learnable proximal networks with differentiable closed-form solvers, enabling physical interpretability while retaining strong representation capacity. Experiments on benchmark datasets (CAVE and NTIRE 2022) with multiple SRFs demonstrate accurate recovery of the STF (degradation operator) and improved reconstruction performance over state-of-the-art SSR methods. Furthermore, evaluations on a real UAV cross-sensor dataset (Headwall Nano HSI and DJI P4 Multispectral MSI) verify the effectiveness and robustness of PGU-Net under truly blind conditions, and suggest that the estimated STF may exhibit land-cover-related differences.

Recent 3D feed-forward models, such as the Visual Geometry Grounded Transformer (VGGT), have shown strong capability in inferring 3D attributes of static scenes. However, since they are typically trained on static datasets, these models often struggle in real-world scenarios involving complex dynamic elements, such as moving humans or deformable objects like umbrellas. To address this limitation, we introduce PAGE-4D, a feedforward model that extends VGGT to dynamic scenes, enabling camera pose estimation, depth prediction and point cloud reconstruction - all without post-processing. A central challenge in multitask 4D reconstruction is the inherent conflict between tasks: accurate camera pose estimation requires suppressing dynamic regions, while geometry reconstruction requires modeling them. To resolve this tension, we propose a dynamics aware aggregator that disentangles static and dynamic information by predicting a dynamics-aware mask - suppressing motion cues for pose estimation while amplifying them for geometry reconstruction. Extensive experiments show that PAGE-4D consistently outperforms the original VGGT in dynamic scenarios, achieving superior results in camera pose estimation, monocular and video depth estimation, and dense point map reconstruction. Necessary code and additional demos are available at Link: https://page4d.github.io/, including both the training-and-inference masking variant and the training-only masking variant (= VGGT architecture at inference). Keywords: VGGT-4D, 4D Perception, Dynamic Scene Reconstruction.

Reinforcement learning (RL) has shown promise in enhancing the general Chain-of-Thought (CoT) reasoning capabilities of multimodal large language models (MLLMs). However, when applied to improve general CoT reasoning, existing RL frameworks often struggle to generalize beyond the training distribution. To address this, we propose NoisyGRPO, a systematic multimodal RL framework that introduces controllable noise into visual inputs for enhanced exploration and explicitly models the advantage estimation process via a Bayesian framework. Specifically, NoisyGRPO improves RL training by: (1) Noise-Injected Exploration Policy: Perturbing visual inputs with Gaussian noise to encourage exploration across a wider range of visual scenarios; and (2) Bayesian Advantage Estimation: Formulating advantage estimation as a principled Bayesian inference problem, where the injected noise level serves as a prior and the observed trajectory reward as the likelihood. This Bayesian modeling fuses both sources of information to compute a robust posterior estimate of trajectory advantage, effectively guiding MLLMs to prefer visually grounded trajectories over noisy ones. Experiments on standard CoT quality, general capability, and hallucination benchmarks demonstrate that NoisyGRPO substantially improves generalization and robustness, especially in RL settings with small-scale MLLMs such as Qwen2.5-VL 3B. The project page is available at https://artanic30.github.io/project_pages/NoisyGRPO/.

The increasing availability of high-resolution satellite imagery, together with advances in deep learning, creates new opportunities for forest monitoring workflows. Two central challenges in this domain are pixel-level change detection and semantic change interpretation, particularly for complex forest dynamics. While large language models (LLMs) are increasingly adopted for data exploration, their integration with vision-language models (VLMs) for remote sensing image change interpretation (RSICI) remains underexplored, especially beyond urban environments. This paper introduces Forest-Chat, an LLM-driven agent for forest change analysis, enabling natural language querying across multiple RSICI tasks, including change detection and captioning, object counting, deforestation characterisation, and change reasoning. Forest-Chat builds upon a multi-level change interpretation (MCI) vision-language backbone with LLM-based orchestration, incorporating zero-shot change detection via AnyChange and multimodal LLM-based zero-shot change captioning and refinement. To support adaptation and evaluation in forest environments, we introduce the Forest-Change dataset, comprising bi-temporal satellite imagery, pixel-level change masks, and semantic change captions via human annotation and rule-based methods. Forest-Chat achieves mIoU and BLEU-4 scores of 67.10% and 40.17% on Forest-Change, and 88.13% and 34.41% on LEVIR-MCI-Trees, a tree-focused subset of LEVIR-MCI. In a zero-shot capacity, it achieves 60.15% and 34.00% on Forest-Change, and 47.32% and 18.23% on LEVIR-MCI-Trees. Further experiments demonstrate the value of caption refinement for injecting geographic domain knowledge into supervised captions, and the system's limited label domain transfer onto JL1-CD-Trees. These findings demonstrate that interactive, LLM-driven systems can support accessible and interpretable forest change analysis.

Federated Learning (FL) facilitates decentralized model training while preserving data privacy. However, achieving both robust generalization and effective personalization simultaneously in heterogeneous (non-IID) environments remains a formidable challenge. Furthermore, the widespread adoption of proprietary Foundation Models (FMs) introduces a critical requirement for dual privacy: (a) protecting sensitive client data and (b) securing the server's valuable intellectual property. This mandates strictly black-box access to the FM. To address these multifaceted challenges, we introduce FedOT, a novel FL framework optimized for black-box FMs. FedOT employs a shared global task-dependent classifier while facilitating local adaptation through client-specific orthogonal transformations applied externally to the FM embeddings. This architecture inherently guarantees that the FM's internal parameters remain inaccessible and unmodified. By enforcing orthogonality, FedOT effectively mitigates gradient conflicts across diverse clients, which is theoretically bounded, preserves the semantic integrity of the FM representations, and achieves robust performance under significant data heterogeneity. The synergy of global and local parameters optimally balances generalization and personalization, markedly outperforming baseline FL methods across diverse benchmarks. Extensive empirical analysis, including rigorous multi-seed validation and scalability assessments, substantiates the robustness, efficiency, and superior performance of FedOT.

Ensuring trustworthiness in machine learning (ML) systems is crucial as they become increasingly embedded in high-stakes domains. This paper advocates for integrating causal methods into machine learning to navigate the trade-offs among key principles of trustworthy ML, including fairness, privacy, robustness, accuracy, and explainability. While these objectives should ideally be satisfied simultaneously, they are often addressed in isolation, leading to conflicts and suboptimal solutions. Drawing on existing applications of causality in ML that successfully align goals such as fairness and accuracy or privacy and robustness, this paper argues that a causal approach is essential for balancing multiple competing objectives in both trustworthy ML and foundation models. Beyond highlighting these trade-offs, we examine how causality can be practically integrated into ML and foundation models, offering solutions to enhance their reliability and interpretability. Finally, we discuss the challenges, limitations, and opportunities in adopting causal frameworks, paving the way for more accountable and ethically sound AI systems.

Many density estimation techniques for 3D human motion prediction require a significant amount of inference time, often exceeding the duration of the predicted time horizon. To address the need for faster density estimation for 3D human motion prediction, we introduce a novel flow-based method for human motion prediction called CacheFlow. Unlike previous conditional generative models that suffer from poor time efficiency, CacheFlow takes advantage of an unconditional flow-based generative model that transforms a Gaussian mixture into the density of future motions. The results of the computation of the flow-based generative model can be precomputed and cached. Then, for conditional prediction, we seek a mapping from historical trajectories to samples in the Gaussian mixture. This mapping can be done by a much more lightweight model, thus saving significant computation overhead compared to a typical conditional flow model. In such a two-stage fashion and by caching results from the slow flow model computation, we build our CacheFlow without loss of prediction accuracy and model expressiveness. This inference process is completed in approximately one millisecond, making it 4 times faster than previous VAE methods and 30 times faster than previous diffusion-based methods on standard benchmarks such as Human3.6M and AMASS datasets. Furthermore, our method demonstrates improved density estimation accuracy and comparable prediction accuracy to a SOTA method on Human3.6M. Our code and models are available at https://github.com/meaten/CacheFlow.