×
![]()
arXiv:2504.10337v2 Announce Type: replace
Abstract: An AI system can create and maintain knowledge only to the extent that it can verify that knowledge itself. Recent work on long Chain-of-Thought reasoning has demonstrated great potential of LLMs on solving competitive problems, but their verification ability remains to be weak and not sufficiently investigated. In this paper, we propose Heimdall, the long CoT verification LLM that can accurately judge the correctness of solutions. With pure reinforcement learning, we boost the verification accuracy from 62.5% to 94.5% on competitive math problems. By scaling with repeated sampling, the accuracy further increases to 97.5%. Through human evaluation, Heimdall demonstrates impressive generalization capabilities, successfully detecting most issues in challenging math proofs, the type of which is not included during training. Furthermore, we propose Pessimistic Verification to extend the functionality of Heimdall to scaling up the problem solving. It calls Heimdall to judge the solutions from a solver model and based on the pessimistic principle, selects the most likely correct solution with the least uncertainty. Taking DeepSeek-R1-Distill-Qwen-32B as the solver model, Pessimistic Verification improves the solution accuracy on AIME2025 from 54.2% to 70.0% with 16x compute budget and to 83.3% with more compute budget. With the stronger solver Gemini 2.5 Pro, the score reaches 93.0%. Finally, we prototype an automatic knowledge discovery system, a ternary system where one poses questions, another provides solutions, and the third verifies the solutions. Using the data synthesis work NuminaMath for the first two components, Heimdall effectively identifies problematic records within the dataset and reveals that nearly half of the data is flawed, which interestingly aligns with the recent ablation studies from NuminaMath.
arXiv:2503.16743v2 Announce Type: replace
Abstract: We introduce an open-ended test grounded in algorithmic probability that can avoid benchmark contamination in the quantitative evaluation of frontier models in the context of their Artificial General Intelligence (AGI) and Superintelligence (ASI) claims. Unlike other tests, this test does not rely on statistical compression methods (such as GZIP or LZW), which are more closely related to Shannon entropy than to Kolmogorov complexity and are not able to test beyond simple pattern matching. The test challenges aspects of AI, in particular LLMs, related to features of intelligence of fundamental nature such as synthesis and model creation in the context of inverse problems (generating new knowledge from observation). We argue that metrics based on model abstraction and abduction (optimal Bayesian `inference') for predictive `planning' can provide a robust framework for testing intelligence, including natural intelligence (human and animal), narrow AI, AGI, and ASI. We found that LLM model versions tend to be fragile and incremental as a result of memorisation only with progress likely driven by the size of training data. The results were compared with a hybrid neurosymbolic approach that theoretically guarantees universal intelligence based on the principles of algorithmic probability and Kolmogorov complexity. The method outperforms LLMs in a proof-of-concept on short binary sequences. We prove that compression is equivalent and directly proportional to a system's predictive power and vice versa. That is, if a system can better predict it can better compress, and if it can better compress, then it can better predict. Our findings strengthen the suspicion regarding the fundamental limitations of LLMs, exposing them as systems optimised for the perception of mastery over human language.
arXiv:2405.14142v2 Announce Type: replace-cross
Abstract: We introduce a multimodal dataset where users express preferences through images. These images encompass a broad spectrum of visual expressions ranging from landscapes to artistic depictions. Users request recommendations for books or music that evoke similar feelings to those captured in the images, and recommendations are endorsed by the community through upvotes. This dataset supports two recommendation tasks: title generation and multiple-choice selection. Our experiments with large foundation models reveal their limitations in these tasks. Particularly, vision-language models show no significant advantage over language-only counterparts that use descriptions, which we hypothesize is due to underutilized visual capabilities. To better harness these abilities, we propose the chain-of-imagery prompting, which results in notable improvements. We release our code and datasets.
arXiv:2504.11511v1 Announce Type: cross
Abstract: The rise of reinforcement learning (RL) in critical real-world applications demands a fundamental rethinking of privacy in AI systems. Traditional privacy frameworks, designed to protect isolated data points, fall short for sequential decision-making systems where sensitive information emerges from temporal patterns, behavioral strategies, and collaborative dynamics. Modern RL paradigms, such as federated RL (FedRL) and RL with human feedback (RLHF) in large language models (LLMs), exacerbate these challenges by introducing complex, interactive, and context-dependent learning environments that traditional methods do not address. In this position paper, we argue for a new privacy paradigm built on four core principles: multi-scale protection, behavioral pattern protection, collaborative privacy preservation, and context-aware adaptation. These principles expose inherent tensions between privacy, utility, and interpretability that must be navigated as RL systems become more pervasive in high-stakes domains like healthcare, autonomous vehicles, and decision support systems powered by LLMs. To tackle these challenges, we call for the development of new theoretical frameworks, practical mechanisms, and rigorous evaluation methodologies that collectively enable effective privacy protection in sequential decision-making systems.
arXiv:2504.11575v1 Announce Type: cross
Abstract: Detecting Distributed Denial of Service (DDoS) attacks in Multi-Environment (M-En) networks presents significant challenges due to diverse malicious traffic patterns and the evolving nature of cyber threats. Existing AI-based detection systems struggle to adapt to new attack strategies and lack real-time attack detection capabilities with high accuracy and efficiency. This study proposes an online, continuous learning methodology for DDoS detection in M-En networks, enabling continuous model updates and real-time adaptation to emerging threats, including zero-day attacks. First, we develop a unique M-En network dataset by setting up a realistic, real-time simulation using the NS-3 tool, incorporating both victim and bot devices. DDoS attacks with varying packet sizes are simulated using the DDoSim application across IoT and traditional IP-based environments under M-En network criteria. Our approach employs a multi-level framework (MULTI-LF) featuring two machine learning models: a lightweight Model 1 (M1) trained on a selective, critical packet dataset for fast and efficient initial detection, and a more complex, highly accurate Model 2 (M2) trained on extensive data. When M1 exhibits low confidence in its predictions, the decision is escalated to M2 for verification and potential fine-tuning of M1 using insights from M2. If both models demonstrate low confidence, the system flags the incident for human intervention, facilitating model updates with human-verified categories to enhance adaptability to unseen attack patterns. We validate the MULTI-LF through real-world simulations, demonstrating superior classification accuracy of 0.999 and low prediction latency of 0.866 seconds compared to established baselines. Furthermore, we evaluate performance in terms of memory usage (3.632 MB) and CPU utilization (10.05%) in real-time scenarios.
arXiv:2503.21620v3 Announce Type: replace
Abstract: The recent DeepSeek-R1 has showcased the emergence of reasoning capabilities in LLMs through reinforcement learning (RL) with rule-based rewards. Despite its success in language models, its application in multi-modal domains, particularly in graphic user interface (GUI) agent tasks, remains under-explored. To address this issue, we propose UI-R1, the first framework to explore how rule-based RL can enhance the reasoning capabilities of multimodal large language models (MLLMs) for GUI action prediction tasks. Specifically, UI-R1 introduces a novel rule-based action reward, enabling model optimization via policy-based algorithms such as Group Relative Policy Optimization (GRPO). For efficient training, we curate a small yet high-quality dataset of 136 challenging tasks, encompassing five common action types on mobile devices. Experimental results demonstrate that our proposed UI-R1-3B achieves significant improvements over the base model (i.e. Qwen2.5-VL-3B) on both in-domain (ID) and out-of-domain (OOD) tasks, with average accuracy gains of 22.1% on ScreenSpot, 6.0% on ScreenSpot-Pro, and 12.7% on ANDROIDCONTROL. Furthermore, UI-R1-3B delivers competitive performance compared to larger models (e.g., OS-Atlas-7B) trained via supervised fine-tuning (SFT) on 76K samples. These results underscore the potential of rule-based reinforcement learning to advance GUI understanding and control, paving the way for future research in this domain. Code website: https://github.com/lll6gg/UI-R1.
arXiv:2504.12011v1 Announce Type: cross
Abstract: Self-supervised learning (SSL) in graphs has garnered significant attention, particularly in employing Graph Neural Networks (GNNs) with pretext tasks initially designed for other domains, such as contrastive learning and feature reconstruction. However, it remains uncertain whether these methods effectively reflect essential graph properties, precisely representation similarity with its neighbors. We observe that existing methods position opposite ends of a spectrum driven by the graph embedding smoothness, with each end corresponding to outperformance on specific downstream tasks. Decomposing the SSL objective into three terms via an information-theoretic framework with a neighbor representation variable reveals that this polarization stems from an imbalance among the terms, which existing methods may not effectively maintain. Further insights suggest that balancing between the extremes can lead to improved performance across a wider range of downstream tasks. A framework, BSG (Balancing Smoothness in Graph SSL), introduces novel loss functions designed to supplement the representation quality in graph-based SSL by balancing the derived three terms: neighbor loss, minimal loss, and divergence loss. We present a theoretical analysis of the effects of these loss functions, highlighting their significance from both the SSL and graph smoothness perspectives. Extensive experiments on multiple real-world datasets across node classification and link prediction consistently demonstrate that BSG achieves state-of-the-art performance, outperforming existing methods. Our implementation code is available at https://github.com/steve30572/BSG.
arXiv:2504.08738v2 Announce Type: replace-cross
Abstract: The rapid growth of e-commerce has led to an overwhelming volume of customer feedback, from product reviews to service interactions. Extracting meaningful insights from this data is crucial for businesses aiming to improve customer satisfaction and optimize decision-making. This paper presents an AI-driven sentiment analysis system designed specifically for e-commerce applications, balancing accuracy with interpretability. Our approach integrates traditional machine learning techniques with modern deep learning models, allowing for a more nuanced understanding of customer sentiment while ensuring transparency in decision-making. Experimental results show that our system outperforms standard sentiment analysis methods, achieving an accuracy of 89.7% on diverse, large-scale datasets. Beyond technical performance, real-world implementation across multiple e-commerce platforms demonstrates tangible improvements in customer engagement and operational efficiency. This study highlights both the potential and the challenges of applying AI to sentiment analysis in a commercial setting, offering insights into practical deployment strategies and areas for future refinement.
arXiv:2504.11514v1 Announce Type: new
Abstract: Neural Networks (NNs) trained through supervised learning struggle with managing edge-case scenarios common in real-world driving due to the intractability of exhaustive datasets covering all edge-cases, making knowledge-driven approaches, akin to how humans intuitively detect unexpected driving behavior, a suitable complement to data-driven methods. This work proposes a hybrid architecture combining low-level Model Predictive Controller (MPC) with locally deployed Large Language Models (LLMs) to enhance decision-making and Human Machine Interaction (HMI). The DecisionxLLM module evaluates robotic state information against natural language instructions to ensure adherence to desired driving behavior. The MPCxLLM module then adjusts MPC parameters based on LLM-generated insights, achieving control adaptability while preserving the safety and constraint guarantees of traditional MPC systems. Further, to enable efficient on-board deployment and to eliminate dependency on cloud connectivity, we shift processing to the on-board computing platform: We propose an approach that exploits Retrieval Augmented Generation (RAG), Low Rank Adaptation (LoRA) fine-tuning, and quantization. Experimental results demonstrate that these enhancements yield significant improvements in reasoning accuracy by up to 10.45%, control adaptability by as much as 52.2%, and up to 10.5x increase in computational efficiency (tokens/s), validating the proposed framework's practicality for real-time deployment even on down-scaled robotic platforms. This work bridges high-level decision-making with low-level control adaptability, offering a synergistic framework for knowledge-driven and adaptive Autonomous Driving Systems (ADS).
arXiv:2503.21138v2 Announce Type: replace
Abstract: In order to reduce the cost of experimental evaluation for models, we introduce a computational theory of evaluation for prediction and decision models: build evaluation model to accelerate the evaluation procedures. We prove upper bounds of generalized error and generalized causal effect error of given evaluation models. We also prove efficiency, and consistency to estimated causal effect from deployed subject to evaluation metric by prediction. To learn evaluation models, we propose a meta-learner to handle heterogeneous evaluation subjects space problem. Comparing with existed evaluation approaches, our (conditional) evaluation model reduced 24.1\%-99.0\% evaluation errors across 12 scenes, including individual medicine, scientific simulation, social experiment, business activity, and quantum trade. The evaluation time is reduced 3-7 order of magnitude comparing with experiments or simulations.
arXiv:2503.19653v3 Announce Type: replace-cross
Abstract: This paper identifies OpenSDI, a challenge for spotting diffusion-generated images in open-world settings. In response to this challenge, we define a new benchmark, the OpenSDI dataset (OpenSDID), which stands out from existing datasets due to its diverse use of large vision-language models that simulate open-world diffusion-based manipulations. Another outstanding feature of OpenSDID is its inclusion of both detection and localization tasks for images manipulated globally and locally by diffusion models. To address the OpenSDI challenge, we propose a Synergizing Pretrained Models (SPM) scheme to build up a mixture of foundation models. This approach exploits a collaboration mechanism with multiple pretrained foundation models to enhance generalization in the OpenSDI context, moving beyond traditional training by synergizing multiple pretrained models through prompting and attending strategies. Building on this scheme, we introduce MaskCLIP, an SPM-based model that aligns Contrastive Language-Image Pre-Training (CLIP) with Masked Autoencoder (MAE). Extensive evaluations on OpenSDID show that MaskCLIP significantly outperforms current state-of-the-art methods for the OpenSDI challenge, achieving remarkable relative improvements of 14.23% in IoU (14.11% in F1) and 2.05% in accuracy (2.38% in F1) compared to the second-best model in localization and detection tasks, respectively. Our dataset and code are available at https://github.com/iamwangyabin/OpenSDI.
arXiv:2504.12172v1 Announce Type: cross
Abstract: Arabic poetry is an essential and integral part of Arabic language and culture. It has been used by the Arabs to spot lights on their major events such as depicting brutal battles and conflicts. They also used it, as in many other languages, for various purposes such as romance, pride, lamentation, etc. Arabic poetry has received major attention from linguistics over the decades. One of the main characteristics of Arabic poetry is its special rhythmic structure as opposed to prose. This structure is referred to as a meter. Meters, along with other poetic characteristics, are intensively studied in an Arabic linguistic field called "\textit{Aroud}". Identifying these meters for a verse is a lengthy and complicated process. It also requires technical knowledge in \textit{Aruod}. For recited poetry, it adds an extra layer of processing. Developing systems for automatic identification of poem meters for recited poems need large amounts of labelled data. In this study, we propose a state-of-the-art framework to identify the poem meters of recited Arabic poetry, where we integrate two separate high-resource systems to perform the low-resource task. To ensure generalization of our proposed architecture, we publish a benchmark for this task for future research.
arXiv:2405.13640v2 Announce Type: replace-cross
Abstract: Reinforcement learning (RL) is an effective method of finding reasoning pathways in incomplete knowledge graphs (KGs). To overcome the challenges of a large action space, a self-supervised pre-training method is proposed to warm up the policy network before the RL training stage. To alleviate the distributional mismatch issue in general self-supervised RL (SSRL), in our supervised learning (SL) stage, the agent selects actions based on the policy network and learns from generated labels; this self-generation of labels is the intuition behind the name self-supervised. With this training framework, the information density of our SL objective is increased and the agent is prevented from getting stuck with the early rewarded paths. Our self-supervised RL (SSRL) method improves the performance of RL by pairing it with the wide coverage achieved by SL during pretraining, since the breadth of the SL objective makes it infeasible to train an agent with that alone. We show that our SSRL model meets or exceeds current state-of-the-art results on all Hits@k and mean reciprocal rank (MRR) metrics on four large benchmark KG datasets. This SSRL method can be used as a plug-in for any RL architecture for a KGR task. We adopt two RL architectures, i.e., MINERVA and MultiHopKG as our baseline RL models and experimentally show that our SSRL model consistently outperforms both baselines on all of these four KG reasoning tasks. Full code for the paper available at https://github.com/owenonline/Knowledge-Graph-Reasoning-with-Self-supervised-Reinforcement-Learning.
arXiv:2411.17404v3 Announce Type: replace
Abstract: LLMs exhibit advanced reasoning capabilities, offering the potential to transform natural language questions into mathematical models. However, existing open-source datasets in operations research domain lack detailed annotations of the modeling process, such as variable definitions, focusing solely on objective values, which hinders reinforcement learning applications. To address this, we release the StructuredOR dataset, annotated with comprehensive labels that capture the complete mathematical modeling process. We further propose BPP-Search, an algorithm that integrates reinforcement learning into a tree-of-thought structure using Beam search, a Process reward model, and a pairwise Preference algorithm. This approach enables efficient exploration of tree structures, avoiding exhaustive search while improving accuracy. Extensive experiments on StructuredOR, NL4OPT, and MAMO-ComplexLP datasets show that BPP-Search significantly outperforms state-of-the-art methods. In tree-based reasoning, BPP-Search excels in accuracy and efficiency, enabling faster retrieval of correct solutions. The StructuredOR dataset is available at https://github.com/tengwang0318/StructuredOR.
arXiv:2412.00127v2 Announce Type: replace-cross
Abstract: We introduce Orthus, an autoregressive (AR) transformer that excels in generating images given textual prompts, answering questions based on visual inputs, and even crafting lengthy image-text interleaved contents. Unlike prior arts on unified multimodal modeling, Orthus simultaneously copes with discrete text tokens and continuous image features under the AR modeling principle. The continuous treatment of visual signals minimizes the information loss for both image understanding and generation while the fully AR formulation renders the characterization of the correlation between modalities straightforward. The key mechanism enabling Orthus to leverage these advantages lies in its modality-specific heads -- one regular language modeling (LM) head predicts discrete text tokens and one diffusion head generates continuous image features conditioning on the output of the backbone. We devise an efficient strategy for building Orthus -- by substituting the Vector Quantization (VQ) operation in the existing unified AR model with a soft alternative, introducing a diffusion head, and tuning the added modules to reconstruct images, we can create an Orthus-base model effortlessly (e.g., within mere 72 A100 GPU hours). Orthus-base can further embrace post-training to better model interleaved images and texts. Empirically, Orthus surpasses competing baselines including Show-o and Chameleon across standard benchmarks, achieving a GenEval score of 0.58 and an MME-P score of 1265.8 using 7B parameters. Orthus also shows exceptional mixed-modality generation capabilities, reflecting the potential for handling intricate practical generation tasks.
arXiv:2504.11750v1 Announce Type: cross
Abstract: Large language model (LLM)-based inference workloads increasingly dominate data center costs and resource utilization. Therefore, understanding the inference workload characteristics on evolving CPU-GPU coupled architectures is crucial for optimization. This paper presents an in-depth analysis of LLM inference behavior on loosely-coupled (PCIe A100/H100) and closely-coupled (GH200) systems. We analyze performance dynamics using fine-grained operator-to-kernel trace analysis, facilitated by our novel profiler SKIP and metrics like Total Kernel Launch and Queuing Time (TKLQT). Results show that closely-coupled (CC) GH200 significantly outperforms loosely-coupled (LC) systems at large batch sizes, achieving 1.9x-2.7x faster prefill latency for Llama 3.2-1B. However, our analysis also reveals that GH200 remains CPU-bound up to 4x larger batch sizes than LC systems. In this extended CPU-bound region, we identify the performance characteristics of the Grace CPU as a key factor contributing to higher inference latency at low batch sizes on GH200. We demonstrate that TKLQT accurately identifies this CPU/GPU-bound transition point. Based on this analysis, we further show that kernel fusion offers significant potential to mitigate GH200's low-batch latency bottleneck by reducing kernel launch overhead. This detailed kernel-level characterization provides critical insights for optimizing diverse CPU-GPU coupling strategies. This work is an initial effort, and we plan to explore other major AI/DL workloads that demand different degrees of CPU-GPU heterogeneous architectures.
arXiv:2410.09080v2 Announce Type: replace
Abstract: Growing evidence suggests that social determinants of health (SDoH), a set of nonmedical factors, affect individuals' risks of developing Alzheimer's disease (AD) and related dementias. Nevertheless, the etiological mechanisms underlying such relationships remain largely unclear, mainly due to difficulties in collecting relevant information. This study presents a novel, automated framework that leverages recent advancements of large language model (LLM) and natural language processing techniques to mine SDoH knowledge from extensive literature and integrate it with AD-related biological entities extracted from the general-purpose knowledge graph PrimeKG. Utilizing graph neural networks, we performed link prediction tasks to evaluate the resultant SDoH-augmented knowledge graph. Our framework shows promise for enhancing knowledge discovery in AD and can be generalized to other SDoH-related research areas, offering a new tool for exploring the impact of social determinants on health outcomes. Our code is available at: https://github.com/hwq0726/SDoHenPKG
arXiv:2504.02623v3 Announce Type: replace
Abstract: Large language models (LLMs) demonstrate strong potential as agents for tool invocation due to their advanced comprehension and planning capabilities. Users increasingly rely on LLM-based agents to solve complex missions through iterative interactions. However, existing benchmarks predominantly access agents in single-mission scenarios, failing to capture real-world complexity. To bridge this gap, we propose the Multi-Mission Tool Bench. In the benchmark, each test case comprises multiple interrelated missions. This design requires agents to dynamically adapt to evolving demands. Moreover, the proposed benchmark explores all possible mission-switching patterns within a fixed mission number. Specifically, we propose a multi-agent data generation framework to construct the benchmark. We also propose a novel method to evaluate the accuracy and efficiency of agent decisions with dynamic decision trees. Experiments on diverse open-source and closed-source LLMs reveal critical factors influencing agent robustness and provide actionable insights to the tool invocation society.
arXiv:2504.11558v1 Announce Type: cross
Abstract: We introduce the Error Broadcast and Decorrelation (EBD) algorithm, a novel learning framework that addresses the credit assignment problem in neural networks by directly broadcasting output error to individual layers. Leveraging the stochastic orthogonality property of the optimal minimum mean square error (MMSE) estimator, EBD defines layerwise loss functions to penalize correlations between layer activations and output errors, offering a principled approach to error broadcasting without the need for weight transport. The optimization framework naturally leads to the experimentally observed three-factor learning rule and integrates with biologically plausible frameworks to enhance performance and plausibility. Numerical experiments demonstrate that EBD achieves performance comparable to or better than known error-broadcast methods on benchmark datasets. While the scalability of EBD to very large or complex datasets remains to be further explored, our findings suggest it provides a biologically plausible, efficient, and adaptable alternative for neural network training. This approach could inform future advancements in artificial and natural learning paradigms.
arXiv:2501.11107v2 Announce Type: replace-cross
Abstract: Chaos Engineering (CE) is an engineering technique aimed at improving the resiliency of distributed systems. It involves artificially injecting specific failures into a distributed system and observing its behavior in response. Based on the observation, the system can be proactively improved to handle those failures. Recent CE tools implement the automated execution of predefined CE experiments. However, defining these experiments and improving the system based on the experimental results still remain manual. To reduce the costs of the manual operations, we propose ChaosEater, a system for automating the entire CE operations with Large Language Models (LLMs). It predefines the agentic workflow according to a systematic CE cycle and assigns subdivided operations within the workflow to LLMs. ChaosEater targets CE for Kubernetes systems, which are managed through code (i.e., Infrastructure as Code). Therefore, the LLMs in ChaosEater perform software engineering tasks to complete CE cycles, including requirement definition, code generation, debugging, and testing. We evaluate ChaosEater through case studies on both small and large Kubernetes systems. The results demonstrate that it stably completes reasonable single CE cycles with significantly low time and monetary costs. The CE cycles are also qualitatively validated by human engineers and LLMs.
arXiv:2410.15954v3 Announce Type: replace-cross
Abstract: Time series classification underpins critical applications such as healthcare diagnostics and gesture-driven interactive systems in multimedia scenarios. However, time series class-incremental learning (TSCIL) faces two major challenges: catastrophic forgetting and intra-class variations. Catastrophic forgetting occurs because gradient-based parameter update strategies inevitably erase past knowledge. And unlike images, time series data exhibits subject-specific patterns, also known as intra-class variations, which refer to differences in patterns observed within the same class. While exemplar-based methods fail to cover diverse variation with limited samples, existing exemplar-free methods lack explicit mechanisms to handle intra-class variations. To address these two challenges, we propose TS-ACL, which leverages a gradient-free closed-form solution to avoid the catastrophic forgetting problem inherent in gradient-based optimization methods while simultaneously learning global distributions to resolve intra-class variations. Additionally, it provides privacy protection and efficiency. Extensive experiments on five benchmark datasets covering various sensor modalities and tasks demonstrate that TS-ACL achieves performance close to joint training on four datasets, outperforming existing methods and establishing a new state-of-the-art (SOTA) for TSCIL.
arXiv:2504.12012v1 Announce Type: new
Abstract: Hallucinations in Large Language Models (LLMs) are widely regarded as errors - outputs that deviate from factual accuracy. However, in creative or exploratory contexts, these "mistakes" may represent unexpected avenues for innovation. We introduce Purposefully Induced Psychosis (PIP), a novel approach that amplifies LLM hallucinations for imaginative tasks such as speculative fiction, interactive storytelling, and mixed-reality simulations. Drawing on Herman Melville's Moby-Dick, where Pip's "madness" reveals profound insight, we reframe hallucinations as a source of computational imagination rather than a flaw. Our method fine-tunes LLMs to encourage speculative, metaphorical, and surreal outputs - hallucinations that are useful when factual accuracy is not the chief objective. Inspired by the consensual illusions of theater and stage magic, PIP situates these creative missteps in contexts where users willingly suspend disbelief, thereby transforming "errors" into catalysts for new ways of thinking. We discuss potential applications, design principles for ensuring user consent, preliminary observations, and implications for broader AI ethics and human-AI collaboration.
arXiv:2311.05418v2 Announce Type: replace-cross
Abstract: The scientific community is increasingly recognizing the importance of generalization in medical AI for translating research into practical clinical applications. A three-level scale is introduced to characterize out-of-distribution generalization performance of medical AI models. This scale addresses the diversity of real-world medical scenarios as well as whether target domain data and labels are available for model recalibration. It serves as a tool to help researchers characterize their development settings and determine the best approach to tackling the challenge of out-of-distribution generalization.
arXiv:2504.11765v1 Announce Type: new
Abstract: Recent large language models (LLMs) face increasing inference latency as input context length and model size continue to grow. In particular, the retrieval-augmented generation (RAG) technique, which enhances LLM responses by incorporating external knowledge, exacerbates this issue by significantly increasing the number of input tokens. This expansion in token length leads to a substantial rise in computational overhead, particularly during the prefill stage, resulting in prolonged time-to-first-token (TTFT). To address this issue, this paper proposes a method to reduce TTFT by leveraging a disk-based key-value (KV) cache to lessen the computational burden during the prefill stage. We also introduce a disk-based shared KV cache management system, called Shared RAG-DCache, for multi-instance LLM RAG service environments. This system, together with an optimal system configuration, improves both throughput and latency under given resource constraints. Shared RAG-DCache exploits the locality of documents related to user queries in RAG, as well as the queueing delay in LLM inference services. It proactively generates and stores disk KV caches for query-related documents and shares them across multiple LLM instances to enhance inference performance. In experiments on a single host equipped with 2 GPUs and 1 CPU, Shared RAG-DCache achieved a 15~71% increase in throughput and up to a 12~65% reduction in latency, depending on the resource configuration.
arXiv:2502.06152v3 Announce Type: replace
Abstract: Multiple agents -- including humans and AI models -- are often paired on decision tasks with the expectation of achieving complementary performance, where the combined performance of both agents outperforms either one alone. However, knowing how to improve the performance of a human-AI team is often difficult without knowing more about what particular information and strategies each agent employs. We provide a decision-theoretic framework for characterizing the value of information -- and consequently, opportunities for agents to better exploit available information -- in AI-assisted decision workflows. We demonstrate the use of the framework for model selection, empirical evaluation of human-AI performance, and explanation design. We propose a novel information-based explanation technique that adapts SHAP, a saliency-based explanation, to explain information value in decision making.
arXiv:2404.15065v2 Announce Type: replace-cross
Abstract: Graph neural networks are becoming increasingly popular in the field of machine learning due to their unique ability to process data structured in graphs. They have also been applied in safety-critical environments where perturbations inherently occur. However, these perturbations require us to formally verify neural networks before their deployment in safety-critical environments as neural networks are prone to adversarial attacks. While there exists research on the formal verification of neural networks, there is no work verifying the robustness of generic graph convolutional network architectures with uncertainty in the node features and in the graph structure over multiple message-passing steps. This work addresses this research gap by explicitly preserving the non-convex dependencies of all elements in the underlying computations through reachability analysis with (matrix) polynomial zonotopes. We demonstrate our approach on three popular benchmark datasets.
arXiv:2504.11478v1 Announce Type: cross
Abstract: We propose a simple yet effective zero-shot framework for subject-driven image generation using a vanilla Flux model. By framing the task as grid-based image completion and simply replicating the subject image(s) in a mosaic layout, we activate strong identity-preserving capabilities without any additional data, training, or inference-time fine-tuning. This "free lunch" approach is further strengthened by a novel cascade attention design and meta prompting technique, boosting fidelity and versatility. Experimental results show that our method outperforms baselines across multiple key metrics in benchmarks and human preference studies, with trade-offs in certain aspects. Additionally, it supports diverse edits, including logo insertion, virtual try-on, and subject replacement or insertion. These results demonstrate that a pre-trained foundational text-to-image model can enable high-quality, resource-efficient subject-driven generation, opening new possibilities for lightweight customization in downstream applications.
arXiv:2504.11844v1 Announce Type: new
Abstract: To what extent do LLMs use their capabilities towards their given goal? We take this as a measure of their goal-directedness. We evaluate goal-directedness on tasks that require information gathering, cognitive effort, and plan execution, where we use subtasks to infer each model's relevant capabilities. Our evaluations of LLMs from Google DeepMind, OpenAI, and Anthropic show that goal-directedness is relatively consistent across tasks, differs from task performance, and is only moderately sensitive to motivational prompts. Notably, most models are not fully goal-directed. We hope our goal-directedness evaluations will enable better monitoring of LLM progress, and enable more deliberate design choices of agentic properties in LLMs.
arXiv:2504.11460v1 Announce Type: cross
Abstract: In this study, we present our methodology for two tasks: the Behavioural Ambivalence/Hesitancy (BAH) Recognition Challenge and the Emotional Mimicry Intensity (EMI) Estimation Challenge, both conducted as part of the 8th Workshop and Competition on Affective & Behavior Analysis in-the-wild. Building on previous work, we utilize a Wav2Vec 2.0 model pre-trained on a large podcast dataset to extract various audio features, capturing both linguistic and paralinguistic information. Our approach incorporates a valence-arousal-dominance (VAD) module derived from Wav2Vec 2.0, a BERT-like encoder, and a vision transformer (ViT) with predictions subsequently processed through a long short-term memory (LSTM) architecture for temporal modeling. In this iteration, we integrate the textual and visual modality into our analysis, recognizing that semantic content provides valuable contextual cues and underscoring that the meaning of speech often conveys more critical insights than its acoustic counterpart alone. Fusing in the vision modality helps in some cases to interpret the textual modality more precisely. This combined approach yields significant performance improvements over baseline methods.
arXiv:2504.11829v1 Announce Type: cross
Abstract: Generation capabilities and language coverage of multilingual large language models (mLLMs) are advancing rapidly. However, evaluation practices for generative abilities of mLLMs are still lacking comprehensiveness, scientific rigor, and consistent adoption across research labs, which undermines their potential to meaningfully guide mLLM development. We draw parallels with machine translation (MT) evaluation, a field that faced similar challenges and has, over decades, developed transparent reporting standards and reliable evaluations for multilingual generative models. Through targeted experiments across key stages of the generative evaluation pipeline, we demonstrate how best practices from MT evaluation can deepen the understanding of quality differences between models. Additionally, we identify essential components for robust meta-evaluation of mLLMs, ensuring the evaluation methods themselves are rigorously assessed. We distill these insights into a checklist of actionable recommendations for mLLM research and development.