We study the well-motivated problem of online distribution shift in which the data arrive in batches and the distribution of each batch can change arbitrarily over time. Since the shifts can be large or small, abrupt or gradual, the length of the relevant historical data to learn from may vary over time, which poses a major challenge in designing algorithms that can automatically adapt to the best ``attention span'' while remaining computationally efficient. We propose a meta-algorithm that takes any network architecture and any Online Learner (OL) algorithm as input and produces a new algorithm which provably enhances the performance of the given OL under non-stationarity. Our algorithm is efficient (it requires maintaining only $O(\log(T))$ OL instances) and adaptive (it automatically chooses OL instances with the ideal ``attention'' length at every timestamp). Experiments on various real-world datasets across text and image modalities show that our method consistently improves the accuracy of user specified OL algorithms for classification tasks. Key novel algorithmic ingredients include a \emph{multi-resolution instance} design inspired by wavelet theory and a cross-validation-through-time technique. Both could be of independent interest.

Reinforcement learning (RL) agents have shown remarkable performances in various environments, where they can discover effective policies directly from sensory inputs. However, these agents often exploit spurious correlations in the training data, resulting in brittle behaviours that fail to generalize to new or slightly modified environments. To address this, we introduce the Causal Object-centric Model Extraction Tool (COMET), a novel algorithm designed to learn the exact interpretable causal world models (CWMs). COMET first extracts object-centric state descriptions from observations and identifies the environment's internal states related to the depicted objects' properties. Using symbolic regression, it models object-centric transitions and derives causal relationships governing object dynamics. COMET further incorporates large language models (LLMs) for semantic inference, annotating causal variables to enhance interpretability. By leveraging these capabilities, COMET constructs CWMs that align with the true causal structure of the environment, enabling agents to focus on task-relevant features. The extracted CWMs mitigate the danger of shortcuts, permitting the development of RL systems capable of better planning and decision-making across dynamic scenarios. Our results, validated in Atari environments such as Pong and Freeway, demonstrate the accuracy and robustness of COMET, highlighting its potential to bridge the gap between object-centric reasoning and causal inference in reinforcement learning.

--Concept drift and extreme verification latency pose significant challenges in data stream learning, particularly when dealing with recurring concept changes in dynamic environments. This work introduces a novel method based on the Growing Neural Gas (GNG) algorithm, designed to effectively handle abrupt recurrent drifts while adapting to incrementally evolving data distributions (incremental drifts). Leveraging the self-organizing and topological adaptability of GNG, the proposed approach maintains a compact yet informative memory structure, allowing it to efficiently store and retrieve knowledge of past or recurring concepts, even under conditions of delayed or sparse stream supervision. Our experiments highlight the superiority of our approach over existing data stream learning methods designed to cope with incremental non-stationarities and verification latency, demonstrating its ability to quickly adapt to new drifts, robustly manage recurring patterns, and maintain high predictive accuracy with a minimal memory footprint. Unlike other techniques that fail to leverage recurring knowledge, our proposed approach is proven to be a robust and efficient online learning solution for unsupervised drifting data flows. Data stream learning has become increasingly relevant in a variety of real-world applications, ranging from fraud detection and stock market analysis to personalized recommendations and industrial process monitoring [1]. These systems rely on continuous real-time processing of data streams to make predictions or decisions. Unlike static datasets, data streams are often characterized by their unbounded, high-speed nature, which necessitates models that can operate incrementally, efficiently, and with minimal reliance on labeled data. Ensuring that such models remain accurate and adaptive over time is crucial for maintaining the performance of systems operating in dynamic environments [2]-[4].

Charts are ubiquitous, as people often use them to analyze data, answer questions, and discover critical insights. However, performing complex analytical tasks with charts requires significant perceptual and cognitive effort. Chart Question Answering (CQA) systems automate this process by enabling models to interpret and reason with visual representations of data. However, existing benchmarks like ChartQA lack real-world diversity and have recently shown performance saturation with modern large vision-language models (LVLMs). To address these limitations, we introduce ChartQAPro, a new benchmark that includes 1,341 charts from 157 diverse sources, spanning various chart types, including infographics and dashboards, and featuring 1,948 questions in various types, such as multiple-choice, conversational, hypothetical, and unanswerable questions, to better reflect real-world challenges. Our evaluations with 21 models show a substantial performance drop for LVLMs on ChartQAPro; e.g., Claude Sonnet 3.5 scores 90.5% on ChartQA but only 55.81% on ChartQAPro, underscoring the complexity of chart reasoning. We complement our findings with detailed error analyses and ablation studies, identifying key challenges and opportunities for advancing LVLMs in chart understanding and reasoning. We release ChartQAPro at https://github.com/vis-nlp/ChartQAPro.

In this paper, we introduce a non-crossing quantile (NQ) network for conditional distribution learning. By leveraging non-negative activation functions, the NQ network ensures that the learned distributions remain monotonic, effectively addressing the issue of quantile crossing. Furthermore, the NQ network-based deep distributional learning framework is highly adaptable, applicable to a wide range of applications, from classical non-parametric quantile regression to more advanced tasks such as causal effect estimation and distributional reinforcement learning (RL). We also develop a comprehensive theoretical foundation for the deep NQ estimator and its application to distributional RL, providing an in-depth analysis that demonstrates its effectiveness across these domains. Our experimental results further highlight the robustness and versatility of the NQ network.

Automatic dietary assessment based on food images remains a challenge, requiring precise food detection, segmentation, and classification. Vision-Language Models (VLMs) offer new possibilities by integrating visual and textual reasoning. In this study, we evaluate six state-of-the-art VLMs (ChatGPT, Gemini, Claude, Moondream, DeepSeek, and LLaVA), analyzing their capabilities in food recognition at different levels. For the experimental framework, we introduce the FoodNExTDB, a unique food image database that contains 9,263 expert-labeled images across 10 categories (e.g., "protein source"), 62 subcategories (e.g., "poultry"), and 9 cooking styles (e.g., "grilled"). In total, FoodNExTDB includes 50k nutritional labels generated by seven experts who manually annotated all images in the database. Also, we propose a novel evaluation metric, Expert-Weighted Recall (EWR), that accounts for the inter-annotator variability. Results show that closed-source models outperform open-source ones, achieving over 90% EWR in recognizing food products in images containing a single product. Despite their potential, current VLMs face challenges in fine-grained food recognition, particularly in distinguishing subtle differences in cooking styles and visually similar food items, which limits their reliability for automatic dietary assessment. The FoodNExTDB database is publicly available at https://github.com/AI4Food/FoodNExtDB.

An embodied system must not only model the patterns of the external world but also understand its own motion dynamics. A motion dynamic model is essential for efficient skill acquisition and effective planning. In this work, we introduce the neural motion simulator (MoSim), a world model that predicts the future physical state of an embodied system based on current observations and actions. MoSim achieves state-of-the-art performance in physical state prediction and provides competitive performance across a range of downstream tasks. This works shows that when a world model is accurate enough and performs precise long-horizon predictions, it can facilitate efficient skill acquisition in imagined worlds and even enable zero-shot reinforcement learning. Furthermore, MoSim can transform any model-free reinforcement learning (RL) algorithm into a model-based approach, effectively decoupling physical environment modeling from RL algorithm development. This separation allows for independent advancements in RL algorithms and world modeling, significantly improving sample efficiency and enhancing generalization capabilities. Our findings highlight that world models for motion dynamics is a promising direction for developing more versatile and capable embodied systems.

To survive and thrive in complex environments, humans have evolved sophisticated self-improvement mechanisms through environment exploration, hierarchical abstraction of experiences into reuseable skills, and collaborative construction of an ever-growing skill repertoire. Despite recent advancements, autonomous web agents still lack crucial self-improvement capabilities, struggling with procedural knowledge abstraction, refining skills, and skill composition. In this work, we introduce SkillWeaver, a skill-centric framework enabling agents to self-improve by autonomously synthesizing reusable skills as APIs. Given a new website, the agent autonomously discovers skills, executes them for practice, and distills practice experiences into robust APIs. Iterative exploration continually expands a library of lightweight, plug-and-play APIs, significantly enhancing the agent's capabilities. Experiments on WebArena and real-world websites demonstrate the efficacy of SkillWeaver, achieving relative success rate improvements of 31.8% and 39.8%, respectively. Additionally, APIs synthesized by strong agents substantially enhance weaker agents through transferable skills, yielding improvements of up to 54.3% on WebArena. These results demonstrate the effectiveness of honing diverse website interactions into APIs, which can be seamlessly shared among various web agents.

Ali Goli 1, Mahdieh Alizadeh 1, and Hadi Sadoghi Yazdi 1,2 1 Department of Computer Engineering, Ferdowsi University of Mashhad, Mashhad, Iran 2 Center of Excellence in Soft Computing and Intelligent Information Processing, Ferdowsi University of Mashhad, Mashhad, Iran April 10, 2025 Abstract Manifold learning techniques, such as Locally linear embedding (LLE), are designed to preserve the local neighborhood structures of high-dimensional data during dimensionality reduction. Traditional LLE employs Euclidean distance to define neighborhoods, which can struggle to capture the intrinsic geometric relationships within complex data. A novel approach, Adaptive locally linear embedding(ALLE), is introduced to address this limitation by incorporating a dynamic, data-driven metric that enhances topological preservation. This method redefines the concept of proximity by focusing on topological neighborhood inclusion rather than fixed distances. By adapting the metric based on the local structure of the data, it achieves superior neighborhood preservation, particularly for datasets with complex geometries and high-dimensional structures. Experimental results demonstrate that ALLE significantly improves the alignment between neighborhoods in the input and feature spaces, resulting in more accurate and topologically faithful embeddings. Keywords-- Manifold Learning, Adaptive Locally Linear Embedding, Dimensionality Reduction, Topological Preservation, Complex Geometries, High-Dimensional Data, Topological Neighborhood Inclusion, Intrinsic Geometric Relationships 1 Introduction Locally linear embedding(LLE) is a prominent manifold learning technique designed to reduce the dimensionality of high-dimensional datasets while preserving their intrinsic geometric structure. Proposed by Roweis and Saul, LLE operates through a systematic process that includes identifying the K-nearest neighbors for each data point, calculating reconstruction weights to express each point as a linear combination of its neighbors, and ultimately generating a low-dimensional representation that retains local relationships [14]. However, LLE traditionally relies on fixed distance metrics, such as Euclidean distance, which may inadequately represent complex data distributions and fail to capture nuanced topological relationships. In response to these limitations, we introduce a novel approach termed Adaptive LLE(ALLE), which integrates a flexible, data-driven metric into the LLE framework.

How can we design AI tools that effectively support human decision-making by complementing and enhancing users' reasoning processes? Common recommendation-centric approaches face challenges such as inappropriate reliance or a lack of integration with users' decision-making processes. Here, we explore an alternative interaction model in which the AI outputs build upon users' own decision-making rationales. We compare this approach, which we call ExtendAI, with a recommendation-based AI. Participants in our mixed-methods user study interacted with both AIs as part of an investment decision-making task. We found that the AIs had different impacts, with ExtendAI integrating better into the decision-making process and people's own thinking and leading to slightly better outcomes. RecommendAI was able to provide more novel insights while requiring less cognitive effort. We discuss the implications of these and other findings along with three tensions of AI-assisted decision-making which our study revealed.