The demand for realistic virtual immersive audio continues to grow, with Head-Related Transfer Functions (HRTFs) playing a key role. HRTFs capture how sound reaches our ears, reflecting unique anatomical features and enhancing spatial perception. It has been shown that personalized HRTFs improve localization accuracy, but their measurement remains time-consuming and requires a noise-free environment. Although machine learning has been shown to reduce the required measurement points and, thus, the measurement time, a controlled environment is still necessary. This paper proposes a method to address this constraint by presenting a novel technique that can upsample sparse, noisy HRTF measurements. The proposed approach combines an HRTF Denoisy U-Net for denoising and an Autoencoding Generative Adversarial Network (AE-GAN) for upsampling from three measurement points. The proposed method achieves a log-spectral distortion (LSD) error of 5.41 dB and a cosine similarity loss of 0.0070, demonstrating the method's effectiveness in HRTF upsampling.

Surgical object detection in laparoscopic videos enables real-time instrument identification for workflow analysis and skills assessment, but training robust models such as You Only Look Once (YOLO) is challenged by limited data, privacy constraints, and inter-institutional variability. Federated learning (FL) enables collaborative training without sharing raw data, yet practical support for modern YOLO pipelines under heterogeneous surgical data remains limited. We present UltraFlwr, an open-source, communication-efficient, and edge-deployable framework that integrates Ultralytics YOLO with the Flower FL platform and supports native Partial Aggregation (PA) of YOLO components (backbone, neck, head). Using two public laparoscopic surgical tool detection datasets, we conduct a systematic empirical study of federated YOLO training under Independent and Identically Distributed (IID) and multiple clinically motivated heterogeneous scenarios, including differences in data curation, video length, and label availability. Results show that standard FL aggregators (e.g., FedAvg) do not consistently match centralized training per client, but reduce inter-client performance variability. Aggregating both backbone and neck components achieves performance comparable to full aggregation with lower communication costs. Also, improving within-client data consistency can benefit FL even when it increases distribution shift across clients. These findings provide practical guidance for deploying federated YOLO-based object detection in heterogeneous surgical environments. UltraFlwr is publicly available at https://github.com/KCL-BMEIS/UltraFlwr.

A fundamental challenge in Visual Autoregressive models is the substantial memory overhead required during inference to store previously generated representations. Despite various attempts to mitigate this issue through compression techniques, prior works have not explicitly formalized the problem of KV-cache compression in this context. In this work, we take the first step in formally defining the KV-cache compression problem for Visual Autoregressive transformers. We then establish a fundamental negative result, proving that any mechanism for sequential visual token generation under attention-based architectures must use at least $Ω(n^2 d)$ memory, when $d = Ω(\log n)$, where $n$ is the number of tokens generated and $d$ is the embedding dimensionality. This result demonstrates that achieving truly sub-quadratic memory usage is impossible without additional structural constraints. Our proof is constructed via a reduction from a computational lower bound problem, leveraging randomized embedding techniques inspired by dimensionality reduction principles. Finally, we discuss how sparsity priors on visual representations can influence memory efficiency, presenting both impossibility results and potential directions for mitigating memory overhead.

Self-supervised learning (SSL) with Vision Transformers (ViT) has shown immense potential in medical image analysis. However, the quadratic complexity ($\mathcal{O}(N^2)$) of standard self-attention poses a severe barrier for high-resolution biomedical tasks, effectively excluding resource-constrained research labs from utilizing state-of-the-art models. To address this computational bottleneck without sacrificing diagnostic accuracy, we propose \textbf{MIRAM}, a Multi-scale Masked Autoencoder that leverages a \textbf{hybrid-attention mechanism}. Our architecture uniquely decouples semantic learning from detail reconstruction using a dual-decoder design: a standard transformer decoder captures global semantics at low resolution, while a linear-complexity decoder (comparing Linformer, Performer, and Nyströmformer) handles the computationally expensive high-resolution reconstruction. This reduces the complexity of the upscaling stage from quadratic to linear ($\mathcal{O}(N)$), enabling high-fidelity training on consumer-grade GPUs. We validate our approach on the CBIS-DDSM mammography dataset. Remarkably, our \textbf{Nyströmformer-based variant} achieves a classification accuracy of \textbf{61.0\%}, outperforming both standard MAE (58.9\%) and MoCo-v3 (60.2\%) while requiring significantly less memory. These results demonstrate that hybrid-attention architectures can democratize high-resolution medical AI, making powerful SSL accessible to researchers with limited hardware resources.

Generative modeling is widely regarded as one of the most essential problems in today's AI community, with text-to-image generation having gained unprecedented real-world impacts. Among various approaches, diffusion models have achieved remarkable success and have become the de facto solution for text-to-image generation. However, despite their impressive performance, these models exhibit fundamental limitations in adhering to numerical constraints in user instructions, frequently generating images with an incorrect number of objects. While several prior works have mentioned this issue, a comprehensive and rigorous evaluation of this limitation remains lacking. To address this gap, we introduce T2ICountBench, a novel benchmark designed to rigorously evaluate the counting ability of state-of-the-art text-to-image diffusion models. Our benchmark encompasses a diverse set of generative models, including both open-source and private systems. It explicitly isolates counting performance from other capabilities, provides structured difficulty levels, and incorporates human evaluations to ensure high reliability. Extensive evaluations with T2ICountBench reveal that all state-of-the-art diffusion models fail to generate the correct number of objects, with accuracy dropping significantly as the number of objects increases. Additionally, an exploratory study on prompt refinement demonstrates that such simple interventions generally do not improve counting accuracy. Our findings highlight the inherent challenges in numerical understanding within diffusion models and point to promising directions for future improvements.

Retrieval Augmented Generation (RAG) has gained popularity as a method for conveniently incorporating novel facts that were not seen during the pre-training stage in Large Language Model (LLM)-based Natural Language Generation (NLG) systems. However, LLMs are known to encode significant levels of unfair social biases. The modulation of these biases by RAG in NLG systems is not well understood. In this paper, we systematically study the relationship between the different components of a RAG system and the social biases presented in the text generated across three languages (i.e. English, Japanese and Chinese) and four social bias types (i.e. gender, race, age and religion). Specifically, using the Bias Question Answering (BBQ) benchmark datasets, we evaluate the social biases in RAG responses from document collections with varying levels of stereotypical biases, employing multiple LLMs used as generators. We find that the biases in document collections are often amplified in the generated responses, even when the generating LLM exhibits a low-level of bias. Our findings raise concerns about the use of RAG as a technique for injecting novel facts into NLG systems and call for careful evaluation of potential social biases in RAG applications before their real-world deployment.

The expressive power and computational complexity of deep visual generative models, such as flow-based and autoregressive (AR) models, have gained considerable interest for their wide-ranging applications in generative tasks. However, the theoretical characterization of their expressiveness through the lens of circuit complexity remains underexplored, particularly for the state-of-the-art architecture like FlowAR proposed by [Ren et al., 2024], which integrates flow-based and autoregressive mechanisms. This gap limits our understanding of their inherent computational limits and practical efficiency. In this study, we address this gap by analyzing the circuit complexity of the FlowAR architecture. We demonstrate that when the largest feature map produced by the FlowAR model has dimensions $n \times n \times c$, the FlowAR model is simulable by a family of threshold circuits $\mathsf{TC}^0$, which have constant depth $O(1)$ and polynomial width $\mathrm{poly}(n)$. This is the first study to rigorously highlight the limitations in the expressive power of FlowAR models. Furthermore, we identify the conditions under which the FlowAR model computations can achieve almost quadratic time. To validate our theoretical findings, we present efficient model variant constructions based on low-rank approximations that align with the derived criteria. Our work provides a foundation for future comparisons with other generative paradigms and guides the development of more efficient and expressive implementations.

Recent advancements in large language models (LLMs) have shown remarkable potential in automating machine learning tasks. However, existing LLM-based agents often struggle with low-diversity and suboptimal code generation. While recent work has introduced Monte Carlo Tree Search (MCTS) to address these issues, limitations persist in the quality and diversity of thoughts generated, as well as in the scalar value feedback mechanisms used for node selection. In this study, we introduce Introspective Monte Carlo Tree Search (I-MCTS), a novel approach that iteratively expands tree nodes through an introspective process that meticulously analyzes solutions and results from parent and sibling nodes. This facilitates a continuous refinement of the node in the search tree, thereby enhancing the overall decision-making process. Furthermore, we integrate a Large Language Model (LLM)-based value model to facilitate direct evaluation of each node's solution prior to conducting comprehensive computational rollouts. A hybrid rewarding mechanism is implemented to seamlessly transition the Q-value from LLM-estimated scores to actual performance scores. This allows higher-quality nodes to be traversed earlier. Applied to the various ML tasks, our approach demonstrates a 4% absolute improvement in performance compared to the strong open-source AutoML agents, showcasing its effectiveness in enhancing agentic AutoML systems. Resource available at https://github.com/jokieleung/I-MCTS

Large Language Models (LLMs) have achieved significant advances in natural language processing, yet their potential for high-stake political decision-making remains largely unexplored. This paper addresses the gap by focusing on the application of LLMs to the United Nations (UN) decision-making process, where the stakes are particularly high and political decisions can have far-reaching consequences. We introduce a novel dataset comprising publicly available UN Security Council (UNSC) records from 1994 to 2024, including draft resolutions, voting records, and diplomatic speeches. Using this dataset, we propose the United Nations Benchmark (UNBench), the first comprehensive benchmark designed to evaluate LLMs across four interconnected political science tasks: co-penholder judgment, representative voting simulation, draft adoption prediction, and representative statement generation. These tasks span the three stages of the UN decision-making process--drafting, voting, and discussing--and aim to assess LLMs' ability to understand and simulate political dynamics. Our experimental analysis demonstrates the potential and challenges of applying LLMs in this domain, providing insights into their strengths and limitations in political science. This work contributes to the growing intersection of AI and political science, opening new avenues for research and practical applications in global governance. The UNBench Repository can be accessed at: https://github.com/yueqingliang1/UNBench.

Explainable Artificial Intelligence (XAI) is essential for the transparency and clinical adoption of Clinical Decision Support Systems (CDSS). However, the real-world effectiveness of existing XAI methods remains limited and is inconsistently evaluated. This study conducts a systematic PRISMA-guided survey of 31 human-centered evaluations (HCE) of XAI applied to CDSS, classifying them by XAI methodology, evaluation design, and adoption barrier. Our findings reveal that most existing studies employ post-hoc, model-agnostic approaches such as SHAP and Grad-CAM, typically assessed through small-scale clinician studies. The results show that over 80% of the studies adopt post-hoc, model-agnostic approaches such as SHAP and Grad-CAM, and that clinician sample sizes remain below 25 participants. The findings indicate that explanations generally improve clinician trust and diagnostic confidence, but frequently increase cognitive load and exhibit misalignment with domain reasoning processes. To bridge these gaps, we propose a stakeholder-centric evaluation framework that integrates socio-technical principles and human-computer interaction to guide the future development of clinically viable and trustworthy XAI-based CDSS.

The goal of inductive logic programming (ILP) is to find a set of logical rules that generalises training examples and background knowledge. We introduce an ILP approach that identifies pointless rules. A rule is pointless if it contains a redundant literal or cannot discriminate against negative examples. We show that ignoring pointless rules allows an ILP system to soundly prune the hypothesis space. Our experiments on multiple domains, including visual reasoning and game playing, show that our approach can reduce learning times by 99% whilst maintaining predictive accuracies.

The da Vinci Research Kit (dVRK, also known as dVRK Classic) is an open-source teleoperated surgical robotic system whose hardware is obtained from the first generation da Vinci Surgical System (Intuitive, Sunnyvale, CA, USA). The dVRK has greatly facilitated research in robot-assisted surgery over the past decade and helped researchers address multiple major challenges in this domain. Recently, the dVRK-Si system, a new version of the dVRK which uses mechanical components from the da Vinci Si Surgical System, became available to the community. The major difference between the first generation da Vinci and the da Vinci Si is in the structural upgrade of the Patient Side Manipulator (PSM). Because of this upgrade, the gravity of the dVRK-Si PSM can no longer be ignored as in the dVRK Classic. The high gravity offset may lead to relatively low control accuracy and longer response time. In addition, although substantial progress has been made in addressing the dynamic model identification problem for the dVRK Classic, further research is required on model-based control for the dVRK-Si, due to differences in mechanical components and the demand for enhanced control performance. To address these problems, in this work, we present (1) a novel full kinematic model of the dVRK-Si PSM, and (2) a gravity compensation approach based on the dynamic model identification.

This paper presents a method for text simplification based on two neural architectures: a neural machine translation (NMT) model and a fine-tuned large language model (LLaMA). Given the scarcity of existing resources for Estonian, a new dataset was created by combining manually translated corpora with GPT-4.0-generated simplifications. OpenNMT was selected as a representative NMT-based system, while LLaMA was fine-tuned on the constructed dataset. Evaluation shows LLaMA outperforms OpenNMT in grammaticality, readability, and meaning preservation. These results underscore the effectiveness of large language models for text simplification in low-resource language settings. The complete dataset, fine-tuning scripts, and evaluation pipeline are provided in a publicly accessible supplementary package to support reproducibility and adaptation to other languages.

Vision-language models (VLMs) deliver strong zero-shot recognition but frequently inherit social biases from their training data. We systematically disentangle three design factors -- model size, training-data scale, and training-data source -- by comparing CLIP and OpenCLIP, two models that share an identical contrastive objective yet differ in encoder width and in the image-text corpora on which they are pre-trained (400M proprietary pairs vs. 400M/2B LAION). Across balanced face-analysis benchmarks, enlarging the encoder reduces gender skew in CLIP but amplifies both gender and racial skew in OpenCLIP; increasing the LAION corpus from 400M to 2B further increases OpenCLIP bias. At matched model and data budgets, substituting proprietary data with LAION improves gender fairness while increasing racial skew, underscoring data source as the primary driver of bias patterns. We also evaluate three post-hoc, test-time debiasing strategies -- Bias Prompts, Prompt Array, and SANER. Debiasing reduces but does not eliminate harm, and its effectiveness is source- and size-dependent: Bias Prompts most effectively reduce gender skew in CLIP at smaller model sizes, whereas Prompt Array and SANER more reliably reduce racial skew in OpenCLIP; scaling LAION reconfigures which method is most fair. Taken together, these findings challenge the assumption that bigger models or datasets are automatically fairer and foreground training data source as the key determinant of both bias and mitigation efficacy. We release code and evaluation scripts to enable transparent, reproducible auditing of future VLMs.

Humans possess a remarkable ability to acquire knowledge efficiently and apply it across diverse modalities through a coherent and shared understanding of the world. Inspired by this cognitive capability, we introduce a concept-centric multi-modality learning framework built around a modality-agnostic concept space that captures structured, abstract knowledge, alongside a set of modality-specific projection models that map raw inputs onto this shared space. The concept space is decoupled from any specific modality and serves as a repository of universally applicable knowledge. Once learned, the knowledge embedded in the concept space enables more efficient adaptation to new modalities, as projection models can align with existing conceptual representations rather than learning from scratch. This efficiency is empirically validated in our experiments, where the proposed framework exhibits faster convergence compared to baseline models. In addition, the framework's modular design supports seamless integration of new modalities, since projection models are trained independently yet produce unified outputs within the shared concept space. We evaluate the framework on two representative downstream tasks. While the focus is not on task-specific optimization, the framework attains comparable results with a smaller training footprint, no task-specific fine-tuning, and inference performed entirely within a shared space of learned concepts that offers interpretability. These findings point toward a promising direction for developing learning systems that operate in a manner more consistent with human cognitive processes.

Simultaneous Localization and Mapping (SLAM) systems typically assume static, distant illumination; however, many real-world scenarios, such as endoscopy, subterranean robotics, and search & rescue in collapsed environments, require agents to operate with a co-located light and camera in the absence of external lighting. In such cases, dynamic near-field lighting introduces strong, view-dependent shading that significantly degrades SLAM performance. We introduce Near-Field Lighting Bundle Adjustment Loss (NFL-BA) which explicitly models near-field lighting as a part of Bundle Adjustment loss and enables better performance for scenes captured with dynamic lighting. NFL-BA can be integrated into neural rendering-based SLAM systems with implicit or explicit scene representations. Our evaluations mainly focus on endoscopy procedure where SLAM can enable autonomous navigation, guidance to unsurveyed regions, blindspot detections, and 3D visualizations, which can significantly improve patient outcomes and endoscopy experience for both physicians and patients. Replacing Photometric Bundle Adjustment loss of SLAM systems with NFL-BA leads to significant improvement in camera tracking, 37% for MonoGS and 14% for EndoGS, and leads to state-of-the-art camera tracking and mapping performance on the C3VD colonoscopy dataset. Further evaluation on indoor scenes captured with phone camera with flashlight turned on, also demonstrate significant improvement in SLAM performance due to NFL-BA. See results at https://asdunnbe.github.io/NFL-BA/

Extracting interpretable equations from observational datasets to describe complex natural phenomena is one of the core goals of artificial intelligence. This field is known as symbolic regression (SR). In recent years, Transformer-based paradigms have become a new trend in SR, addressing the well-known problem of inefficient search. However, the modal heterogeneity between datasets and equations often hinders the convergence and generalization of these models. In this paper, we propose ViSymRe, a Vision Symbolic Regression framework, to explore the positive role of visual modality in enhancing the performance of Transformer-based SR paradigms. To overcome the challenge where the visual SR model is untrainable in high-dimensional scenarios, we present Multi-View Random Slicing (MVRS). By projecting multivariate equations into 2-D space using random affine transformations, MVRS avoids common defects in high-dimensional visualization, such as variable degradation, non-linear interaction missing, and exponentially increasing sampling complexity, enabling ViSymRe to be trained with low computational costs. To support dataset-only deployment of ViSymRe, we design a dual-vision pipeline architecture based on generative techniques, which reconstructs visual features directly from the datasets via an auxiliary Visual Decoder and automatically suppresses the attention weights of reconstruction noise through a proposed Biased Cross-Attention feature fusion module, ensuring that subsequent processes are not affected by noisy modalities. Ablation studies demonstrate the positive contribution of visual modality to improving model convergence level and enhancing various SR metrics. Furthermore, evaluation results on mainstream benchmarks indicate that ViSymRe achieves competitive performance compared to baselines, particularly in low-complexity and rapid-inference scenarios.

Crack detection is a critical task in structural health monitoring, aimed at assessing the structural integrity of bridges, buildings, and roads to prevent potential failures. Vision-based crack detection has become the mainstream approach due to its ease of implementation and effectiveness. Fusing infrared (IR) channels with red, green and blue (RGB) channels can enhance feature representation and thus improve crack detection. However, IR and RGB channels often differ in resolution. To align them, higher-resolution RGB images typically need to be downsampled to match the IR image resolution, which leads to the loss of fine details. Moreover, crack detection performance is restricted by the limited receptive fields and high computational complexity of traditional image segmentation networks. Inspired by the recently proposed Mamba neural architecture, this study introduces a two-stage paradigm called MSCrackMamba, which leverages Vision Mamba along with a super-resolution network to address these challenges. Specifically, to align IR and RGB channels, we first apply super-resolution to IR channels to match the resolution of RGB channels for data fusion. Vision Mamba is then adopted as the backbone network, while UperNet is employed as the decoder for crack detection. Our approach is validated on the large-scale Crack Detection dataset Crack900, demonstrating an improvement of 3.55% in mIoU compared to the best-performing baseline methods.

The high costs and risks involved in extensive environment interactions hinder the practical application of current online safe reinforcement learning (RL) methods. While offline safe RL addresses this by learning policies from static datasets, the performance therein is usually limited due to reliance on data quality and challenges with out-of-distribution (OOD) actions. Inspired by recent successes in offline-to-online (O2O) RL, it is crucial to explore whether offline safe RL can be leveraged to facilitate faster and safer online policy learning, a direction that has yet to be fully investigated. To fill this gap, we first demonstrate that naively applying existing O2O algorithms from standard RL would not work well in the safe RL setting due to two unique challenges: \emph{erroneous Q-estimations}, resulted from offline-online objective mismatch and offline cost sparsity, and \emph{Lagrangian mismatch}, resulted from difficulties in aligning Lagrange multipliers between offline and online policies. To address these challenges, we introduce \textbf{Marvel}, a novel framework for O2O safe RL, comprising two key components that work in concert: \emph{Value Pre-Alignment} to align the Q-functions with the underlying truth before online learning, and \emph{Adaptive PID Control} to effectively adjust the Lagrange multipliers during online finetuning. Extensive experiments demonstrate that Marvel significantly outperforms existing baselines in both reward maximization and safety constraint satisfaction. By introducing the first policy-finetuning based framework for O2O safe RL, which is compatible with many offline and online safe RL methods, our work has the great potential to advance the field towards more efficient and practical safe RL solutions.

We study the problem of robot navigation in dense and interactive crowds with static constraints such as corridors and furniture. Previous methods fail to consider all types of spatial and temporal interactions among agents and obstacles, leading to unsafe and inefficient robot paths. In this article, we leverage a graph-based representation of crowded and constrained scenarios and propose a structured framework to learn robot navigation policies with deep reinforcement learning. We first split the representations of different inputs and propose a heterogeneous spatio-temporal graph to model distinct interactions among humans, robots, and obstacles. Based on the heterogeneous spatio-temporal graph, we propose HEIGHT, a novel navigation policy network architecture with different components to capture heterogeneous interactions through space and time. HEIGHT utilizes attention mechanisms to prioritize important interactions and a recurrent network to track changes in the dynamic scene over time, encouraging the robot to avoid collisions adaptively. Through extensive simulation and real-world experiments, we demonstrate that HEIGHT outperforms state-of-the-art baselines in terms of success, navigation time, and generalization to domain shifts in challenging navigation scenarios. More information is available at https://sites.google.com/view/crowdnav-height/home.

Previous text-to-SQL datasets and systems have primarily focused on user questions with clear intentions that can be answered. However, real user questions can often be ambiguous with multiple interpretations or unanswerable due to a lack of relevant data. In this work, we construct a practical conversational text-to-SQL dataset called PRACTIQ, consisting of ambiguous and unanswerable questions inspired by real-world user questions. We first identified four categories of ambiguous questions and four categories of unanswerable questions by studying existing text-to-SQL datasets. Then, we generate conversations with four turns: the initial user question, an assistant response seeking clarification, the user's clarification, and the assistant's clarified SQL response with the natural language explanation of the execution results. For some ambiguous queries, we also directly generate helpful SQL responses, that consider multiple aspects of ambiguity, instead of requesting user clarification. To benchmark the performance on ambiguous, unanswerable, and answerable questions, we implemented large language model (LLM)-based baselines using various LLMs. Our approach involves two steps: question category classification and clarification SQL prediction. Our experiments reveal that state-of-the-art systems struggle to handle ambiguous and unanswerable questions effectively. We will release our code for data generation and experiments on GitHub.

Large Language Models (LLMs) exhibit exceptional proficiency in handling extensive context windows in natural language. Nevertheless, the quadratic scaling of attention computation relative to sequence length creates substantial efficiency bottlenecks, necessitating the development of I/O-optimized algorithms. In this work, we conduct a systematic examination of the I/O complexity inherent in attention mechanisms, with a specific emphasis on the backward pass under both small and large cache settings. By leveraging the red-blue pebble game framework, we derive tight bounds for I/O complexity across the full spectrum of cache sizes. We validate that FlashAttention, one of the current industry standards, achieves optimality in the large-cache scenario for both forward and backward passes. Conversely, for small-cache environments, we introduce a novel algorithm that outperforms contemporary methods and successfully attains theoretical tight bounds. Furthermore, we expand our investigation to include sparse attention by establishing granular lower bounds for both forward and backward passes across all cache configurations. Ultimately, our results solidify the theoretical framework regarding I/O complexity in attention mechanisms, providing critical guidance for the development of efficient LLM training and inference systems.

Text-to-image (T2I) models are increasingly used in impactful real-life applications. As such, there is a growing need to audit these models to ensure that they generate desirable, task-appropriate images. However, systematically inspecting the associations between prompts and generated content in a human-understandable way remains challenging. To address this, we propose Concept2Concept, a framework where we characterize conditional distributions of vision language models using interpretable concepts and metrics that can be defined in terms of these concepts. This characterization allows us to use our framework to audit models and prompt-datasets. To demonstrate, we investigate several case studies of conditional distributions of prompts, such as user-defined distributions or empirical, real-world distributions. Lastly, we implement Concept2Concept as an open-source interactive visualization tool to facilitate use by non-technical end-users. A demo is available at https://tinyurl.com/Concept2ConceptDemo.

Information Retrieval (IR) methods aim to identify documents relevant to a query, which have been widely applied in various natural language tasks. However, existing approaches typically consider only the textual content within documents, overlooking the fact that documents can contain multiple modalities, including images and tables. Also, they often segment each long document into multiple discrete passages for embedding, which prevents them from capturing the overall document context and interactions between paragraphs. To address these two challenges, we propose a method that holistically embeds documents interleaved with multiple modalities by leveraging the capability of recent vision-language models that enable the processing and integration of text, images, and tables into a unified format and representation. Moreover, to mitigate the information loss from segmenting documents into passages, instead of representing and retrieving passages individually, we further merge the representations of segmented passages into one single document representation, while we additionally introduce a reranking strategy to decouple and identify the relevant passage within the document if necessary. Then, through extensive experiments on diverse IR scenarios considering both the textual and multimodal queries, we show that our approach substantially outperforms relevant baselines, thanks to the consideration of the multimodal information within documents.

Differentiating generated and human-written content is increasingly difficult. We examine how an incentive to convey humanness and task characteristics shape this human vs AI race across five studies. In Study 1-2 (n=530 and n=610) humans and a large language model (LLM) wrote relationship advice or relationship descriptions, either with or without instructions to sound human. New participants (n=428 and n=408) judged each text's source. Instructions to sound human were only effective for the LLM, reducing the human advantage. Study 3 (n=360 and n=350) showed that these effects persist when writers were instructed to avoid sounding like an LLM. Study 4 (n=219) tested empathy as mechanism of humanness and concluded that LLMs can produce empathy without humanness and humanness without empathy. Finally, computational text analysis (Study 5) indicated that LLMs become more human-like by applying an implicit representation of humanness to mimic stochastic empathy.

Causal discovery from time-series data aims to capture both intra-slice (contemporaneous) and inter-slice (time-lagged) causality between variables within the temporal chain, which is crucial for various scientific disciplines. Compared to causal discovery from non-time-series data, causal discovery from time-series data necessitates more serialized samples with a larger amount of observed time steps. To address the challenges, we propose a novel gradient-based causal discovery approach STIC, which focuses on \textbf{S}hort-\textbf{T}erm \textbf{I}nvariance using \textbf{C}onvolutional neural networks to uncover the causal relationships from time-series data. Specifically, STIC leverages both the short-term time and mechanism invariance of causality within each window observation, which possesses the property of independence, to enhance sample efficiency. Furthermore, we construct two causal convolution kernels, which correspond to the short-term time and mechanism invariance respectively, to estimate the window causal graph. To demonstrate the necessity of convolutional neural networks for causal discovery from time-series data, we theoretically derive the equivalence between convolution and the underlying generative principle of time-series data under the assumption that the additive noise model is identifiable. Experimental evaluations conducted on both synthetic and FMRI benchmark datasets demonstrate that our STIC outperforms baselines significantly and achieves the state-of-the-art performance, particularly when the datasets contain a limited number of observed time steps. Code is available at \url{https://github.com/HITshenrj/STIC}.

Cross-attention has emerged as a cornerstone module in modern artificial intelligence, underpinning critical applications such as retrieval-augmented generation (RAG), system prompting, and guided stable diffusion. However, this is a rising concern about securing the privacy of cross-attention, as the underlying key and value matrices frequently encode sensitive data or private user information. In this work, we introduce a novel data structure designed to enforce differential privacy (DP) for cross-attention mechanisms, accompanied by provable theoretical guarantees. Specifically, letting $n$ denote the input sequence length, $d$ the feature dimension, $R$ the maximum magnitude of query and key matrices, $R_w$ the maximum magnitude of the value matrix, and $r, s, ε_s$ the parameters for polynomial kernel methods, our proposed structure achieves $\widetilde{O}(ndr^2)$ space and initialization complexity, with a query time of $\widetilde{O}(d r^2)$ per token. Moreover, we demonstrate that our mechanism satisfies $(ε, δ)$-DP, incurring an additive error of $\widetilde{O}((1-ε_s)^{-1} n^{-1} ε^{-1} R^{2s} R_w r^2)$ and a relative error of $2ε_s/(1-ε_s)$ with respect to the ground truth. Crucially, our framework maintains robustness against adaptive queries, ensuring security even in adversarial settings. To the best of our knowledge, this constitutes the first approach providing provable differential privacy for cross-attention, establishing a foundation for future privacy-preserving algorithms in large generative models (LGMs).

In open-set semi-supervised learning (OSSL), we consider unlabeled datasets that may contain unknown classes. Existing OSSL methods often use the softmax confidence for classifying data as in-distribution (ID) or out-of-distribution (OOD). Additionally, many works for OSSL rely on ad-hoc thresholds for ID/OOD classification, without considering the statistics of the problem. We propose a new score for ID/OOD classification based on angles in feature space between data and an ID subspace. Moreover, we propose an approach to estimate the conditional distributions of scores given ID or OOD data, enabling probabilistic predictions of data being ID or OOD. These components are put together in a framework for OSSL, termed ProSub, that is experimentally shown to reach SOTA performance on several benchmark problems. Our code is available at https://github.com/walline/prosub.

Mixture of experts is a prediction aggregation method in machine learning that aggregates the predictions of specialized experts. This method often outperforms Bayesian methods despite the Bayesian having stronger inductive guarantees. We argue that this is due to the greater functional capacity of mixture of experts. We prove that in a limiting case of mixture of experts will have greater capacity than equivalent Bayesian methods, which we vouchsafe through experiments on non-limiting cases. Finally, we conclude that mixture of experts is a type of abductive reasoning in the Peircian sense of hypothesis construction.

The presence of spherical distortion in equirectangular projection (ERP) images presents a persistent challenge in dense regression tasks such as surface normal estimation. Although it may appear straightforward to repurpose architectures developed for 360° depth estimation, our empirical findings indicate that such models yield suboptimal performance when applied to surface normal prediction. This is largely attributed to their architectural bias toward capturing global scene layout, which comes at the expense of the fine-grained local geometric cues that are critical for accurate surface orientation estimation. While convolutional neural networks (CNNs) have been employed to mitigate spherical distortion, their fixed receptive fields limit their ability to capture holistic scene structure. Conversely, vision transformers (ViTs) are capable of modeling long-range dependencies via global self-attention, but often fail to preserve high-frequency local detail. To address these limitations, we propose \textit{PanoNormal}, a monocular surface normal estimation architecture for 360° images that integrates the complementary strengths of CNNs and ViTs. In particular, we design a multi-level global self-attention mechanism that explicitly accounts for the spherical feature distribution, enabling our model to recover both global contextual structure and local geometric details. Experimental results demonstrate that our method not only achieves state-of-the-art performance on several benchmark 360° datasets, but also significantly outperforms adapted depth estimation models on the task of surface normal prediction. The code and model are available at https://github.com/huangkun101230/PanoNormal.