arXiv:2606.05037v1 Announce Type: new
Abstract: When an AI agent calls an API and hits a validation error, it needs more than what went wrong -- it needs what to do next. A self-reflective API returns, on validation failure, a machine-readable recovery\_feedback.suggestions[] payload sufficient for the agent to repair the request and retry without external reasoning. On a leak-audited pilot ($N{=}30$ per cell, 3 LLMs, 10 adversarial tasks), structured suggestions lift task-completion rate by $+36.7$--$40.0$pp over plain-English diagnoses on Anthropic models (Fisher's exact $p \le 0.0022$), at $1.8$--$2.2\times$ better per-success token efficiency. The lift is not significant on gpt-4o-mini ($p{=}0.435$); a second-domain replication on a billing API confirms the pattern. The comparison only holds after auditing two undocumented classes of answer leakage in LLM benchmarks. We shipaudit\_prompt\_leakage.py as reusable CI infrastructure. Code and data: https://github.com/arquicanedo/self-reflective-apis.
Science Journals
arXiv:2606.04239v1 Announce Type: new
Abstract: This paper presents a state observer design for continuous linear time-invariant (LTI) systems subject to unknown bounded disturbances, that enforces a prescribed bound on the observer residual. The proposed observer augments a continuous-time Luenberger observer with state resets, triggered when the norm of the residual equals a pre-specified bound. The reset map guarantees contraction of the residual at jump instants while preserving the uniform boundedness properties of a standard Luenberger observer. The paper also establishes forward invariance of the residual envelope and non-expansiveness of the estimation error in a Lyapunov metric. Simulation results confirm the analysis. Under bounded disturbances, the residual stays within the prescribed bound. A standard Luenberger observer with the same gains violates this bound.
arXiv:2602.01146v2 Announce Type: replace
Abstract: Conversational assistants are increasingly integrating long-term memory with large language models (LLMs). This persistence of memories, e.g., the user is vegetarian, can enhance personalization in future conversations. However, the same persistence can also introduce safety risks that have been largely overlooked. Hence, we introduce PersistBench to measure the extent of these safety risks. We identify two long-term memory-specific risks: cross-domain leakage, where LLMs inappropriately inject context from the long-term memories; and memory-induced sycophancy, where stored long-term memories insidiously reinforce user biases. We evaluate 18 frontier and open-source LLMs on our benchmark. Our results reveal a surprisingly high failure rate across these LLMs - a median failure rate of 53% on cross-domain samples and 97% on sycophancy samples. To address this, our benchmark encourages the development of more robust and safer long-term memory usage in frontier conversational systems.
arXiv:2602.01083v2 Announce Type: replace
Abstract: Weight-space learning studies neural architectures that operate directly on the parameters of other neural networks. Motivated by the growing availability of pretrained models, recent work has demonstrated the effectiveness of weight-space networks across a wide range of tasks. SOTA weight-space networks rely on permutation-equivariant designs to improve generalization. However, this may negatively affect expressive power, warranting theoretical investigation. Importantly, unlike other structured domains, weight-space learning targets maps operating on both weight and function spaces, making expressivity analysis particularly subtle. While a few prior works provide partial expressivity results, a comprehensive characterization is still missing. In this work, we address this gap by developing a systematic theory for expressivity of weight-space networks. We first prove that all prominent permutation-equivariant networks are equivalent in expressive power. We then establish universality in both weight- and function-space settings under mild, natural assumptions on the input weights, and characterize the edge-case regimes where universality no longer holds. Guided by our theoretical results, we show that slight modifications to existing weight-space models yield a 34% improvement over prior SOTA, demonstrating the practical relevance of our framework.
arXiv:2602.01027v2 Announce Type: replace
Abstract: Mixed-precision quantization is a promising approach for compressing large language models under tight memory budgets. However, existing mixed-precision methods typically suffer from one of two limitations: they either rely on expensive discrete optimization to determine precision allocation, or introduce hardware inefficiencies due to irregular memory layouts. We propose SFMP, a search-free and hardware-friendly mixed-precision quantization framework for large language models. The framework is built upon four novel ideas: Fractional bit-width, which extends integer bit-width for weight matrix to fractional value and transforms discrete precision allocation as a continuous problem; 2)Block-wise mixed-precision, enabling fine-grained precision within weight matrices while remaining hardware-friendly; 3)Row-column weight reordering, which aggregates salient weights via row and column reordering, incurring only a small activation reordering overhead during inference; 4)Unified GEMM kernel, which supports mixed-precision GEMM at arbitrary average bit-width. Extensive experiments demonstrate that SFMP outperforms state-of-the-art layer-wise mixed-precision methods under the same memory constraints, while significantly reducing quantization cost and improving inference efficiency. Code is available at https://github.com/Nkniexin/SFMP
arXiv:2606.04305v1 Announce Type: new
Abstract: We study online learning with an additional offline dataset in the stochastic linear bandit setting. Although this problem arises frequently in practice, the offline-to-online tradeoff remains poorly understood in structured environments. We propose a linear bandit algorithm that balances this tradeoff: it relies on offline data during early rounds, and increasingly favors exploration as the horizon grows. We establish regret bounds showing that our method is simultaneously competitive with both purely online and purely offline solutions. In particular, it achieves sublinear regret relative to the optimal action in the number of online interactions, while its regret relative to an offline reference decreases as the number of offline samples grows. Empirical results further demonstrate its effectiveness across various problem parameters.
arXiv:2606.04228v1 Announce Type: new
Abstract: AI scientist systems increasingly choose biological foundation models before they choose experiments. In protein pipelines, this creates a concrete engineering and scientific question: when is the cost of structural inference worth paying over a cheaper sequence-only model? We introduce the information bonus (IB), a task-level metric that measures the linearly accessible advantage of frozen single-sequence AlphaFold2 Evoformer representations over frozen ESM-2 embeddings under protein-level cross-validation. Across binding affinity regression (PDBbind, n=5,680), conformational flexibility (ATLAS molecular dynamics, 268 proteins), and allosteric-site classification (AlloSigDB, n=9,925 residues), IB is sharply mechanism-dependent. ESM-2 dominates binding affinity (IB=-0.141; Pearson r=0.449 vs. 0.307) and binary flexibility (IB=-0.060; AUROC 0.824 vs. 0.764; p=0.0017). AF2 single representations give the only above-chance allostery predictions (IB=+0.064; AUROC 0.548 vs. 0.485), revealing long-range geometric signal not recovered from sequence alone. We also identify a residue-level leakage artifact: naive residue splits inflate RMSF performance by 27-39% depending on the representation, enough to reverse representation rankings. These results turn representation selection into a measurable decision for AI-for-science systems.
arXiv:2505.03057v2 Announce Type: replace
Abstract: This paper addresses the $\mathcal{H}_2$-optimal approximation of linear dynamical systems with quadratic-output functions, also known as linear quadratic-output systems. Our major contributions are threefold. First, we derive interpolatory first-order optimality conditions for the linear quadratic-output $\mathcal{H}_2$ minimization problem. These conditions correspond to the mixed-multipoint tangential interpolation of the full-order linear- and quadratic-output transfer functions, and generalize the Meier-Luenberger optimality framework for the $\mathcal{H}_2$-optimal model reduction of linear time-invariant systems. Second, given the optimal interpolation data, we show how to enforce the interpolatory optimality conditions explicitly by Petrov-Galerkin projection of the full-order model. Third, to find the optimal interpolation data, we build on this projection framework and propose a generalization of the iterative rational Krylov algorithm for the $\mathcal{H}_2$-optimal model reduction of linear quadratic-output systems, called LQO-IRKA. Upon convergence, LQO-IRKA produces reduced linear quadratic-output systems that satisfy the interpolatory optimality conditions. The method only requires solving shifted linear systems and matrix-vector products, thus making it suitable for large-scale problems. Numerical examples are included to illustrate the effectiveness of the proposed method.
arXiv:2603.18577v2 Announce Type: replace
Abstract: Text-guided image editors can now manipulate authentic medical scans with high fidelity, enabling lesion implantation/removal that threatens clinical trust and safety. Existing defenses are inadequate for healthcare. Medical detectors are largely black-box, while MLLM-based explainers are typically post-hoc, lack medical expertise, and may hallucinate evidence on ambiguous cases. We present MedForge, a data-and-method solution for pre-hoc, evidence-grounded medical forgery detection. We introduce MedForge-90K, a large-scale benchmark of realistic lesion edits across 19 pathologies with expert-guided reasoning supervision via doctor inspection guidelines and gold edit locations. Building on it, MedForge-Reasoner performs localize-then-analyze reasoning, predicting suspicious regions before producing a verdict, and is further aligned with Forgery-aware GSPO to strengthen grounding and reduce hallucinations. Experiments demonstrate state-of-the-art detection accuracy and trustworthy, expert-aligned explanations.
arXiv:2606.04393v1 Announce Type: new
Abstract: To overcome the difficulty of single nanostructures in approaching the theoretical limit of chiroptical performance, we design a single plasmonic twisted dimer cavity whose magnetic gap plasmon mode enables magnetic polarization near-field engineering for high chirality. The structure exhibits strong extinction under circularly polarized excitation with one handedness, while its response to the orthogonally circularly polarized light is almost perfectly suppressed, yielding a chiral g-factor as high as 1.94. Meanwhile, the structure demonstrates strong chiral-selective spin-orbit angular momentum conversion: the conversion efficiency is ~95% under circularly polarized excitation with one handedness and only ~1% under the other. By tuning geometric parameters, the g-factor can be continuously adjusted from 0 to 1.94. Without relying on periodic coupling or collective effects, this work achieves near-perfect chirality and highly efficient angular momentum manipulation solely through intrinsic near-field matching, providing a new design strategy and theoretical basis for highly selective, ultra-compact integrated chiral photonic devices.
arXiv:2606.04612v1 Announce Type: new
Abstract: Large Language Models (LLMs) are vulnerable both to hallucination and adversarial manipulation. Although these problems are closely related, existing defences typically address them separately. We investigate a hybrid defence framework that combines entropy-based models, designed to reduce hallucinations, with uncertainty-based models and geometric-based models, designed to reduce vulnerability. Under in-domain tests on Natural Language Understanding datasets (FEVER, HotpotQA, CSQA, SIQA) we find our hybrid model improves both clean-task performance (up to 43.34\% increase in accuracy) and adversarial robustness (up to 64.92\% improvement in accuracy and 62.27\% reduction in attack success rate). For out-of-distribution datasets (AeroEngQA, CPIQA) we see similar adversarial robustness from our hybrid model (up to 57.14\% improvement in accuracy). For prompt injection (SafeGuard) and jailbreak detection (AdvBench, DAN) datasets our hybrid model is also very strong (up to 51\% reduction in attack success rate compared to state of the art baseline models). Overall, our results show that combining entropy, uncertainty and geometric features provides a more effective defence strategy than using any single feature alone for both in-domain and out-of-distribution tasks.
arXiv:2606.04231v1 Announce Type: new
Abstract: Recent advances in multimodal retrieval-augmented generation (MM-RAG) have shifted toward minimal parsing, relying on page-level images for producing retriever embeddings and for answer generation. While efficient, this trend often neglects explicit handling of the rich, structured information in complex enterprise documents, instead depending on pre-trained embeddings or vision-language models to implicitly capture such structure. In this work, we take a more direct approach: MM-BizRAG proactively extracts and represents document structure via a document structure-aware split that dynamically routes documents through orientation-specific ingestion pipelines, applying explicit layout-aware parsing for vertically structured documents (e.g., reports) and holistic page-level representations for horizontally structured documents (e.g., slide decks). A unified LLM-driven artifact transformation pipeline with placeholder-based positional alignment preserves natural reading order, while inference-time multimodal assembly decouples retrieval representations from generation context, enabling richer, more grounded answers without any finetuning requirement. Through experiments on a large, heterogeneous enterprise dataset and two public benchmarks (SlideVQA and FinRAGBench-V), MM-BizRAG consistently outperforms state-of-the-art vision-centric baselines by up to 32% points, with especially strong gains on report-style layouts. Furthermore, we introduce FastRAGEval, a single-call LLM Judge metric for fine-grained generative recall that halves RAGChecker's cost while achieving stronger human alignment.
arXiv:2606.05121v1 Announce Type: new
Abstract: Audio is an inherently interactive modality, yet today's Large Audio Language Models (LALMs) are offline, and streaming audio models each handle only a single task such as streaming ASR or voice chatting. It is time to unify them into one online LALM: a model that, through an always-on perceive-decide-respond loop, listens to sound, environment, and instructions in real time and reacts on the fly. We formalize this regime as the Audio Interaction Model, and realize it with Audio-Interaction, a unified streaming model that retains offline task execution while adding online general audio instruction following, from dialogue to full voice chatting, deciding when to respond from the semantics of the stream. To enable this, we propose SoundFlow, a framework that instantiates the perceive-decide-respond loop end to end, from data to training to deployment, through streaming-native data construction, comprehension-aware training, and asynchronous low-latency inference for stable real-time interaction. We further construct StreamAudio-2M, a 2.6M-item streaming corpus spanning 7 fundamental abilities and 28 sub-tasks, and Proactive-Sound-Bench for evaluating proactive audio intervention. Across 8 benchmarks, Audio-Interaction preserves competitive performance on mainstream audio tasks while unlocking capabilities inaccessible to offline LALMs, including real-time ASR, streaming audio instruction following, and proactive help.
arXiv:2505.01351v2 Announce Type: replace
Abstract: Recognizing player state is only one component of affective game adaptation; inferred experience must also be translated into adaptive interventions that modify gameplay or game content. Although player experience modeling and content adaptation are established research areas, fewer studies examine how sensing, modeling, and adaptation are integrated into complete, empirically evaluated gameplay systems. This PRISMA-guided systematic review analyzes 23 empirical studies published from January 1, 2015, to December 31, 2025, that implement a complete experience-driven loop defined here as the combination of player data acquisition, player experience modeling, and adaptive game content. Complete-loop systems were relatively uncommon in the retrieved corpus, and the selected systems were predominantly oriented toward dynamic difficulty adjustment, engagement, rehabilitation, or performance-related goals. Game telemetry was the dominant input modality, while non-invasive sources with affective relevance, such as facial expression analysis and peripheral interaction data, were less common. Knowledge-based methods, including rule-based systems and heuristics, dominated both modeling and adaptation because of their interpretability and low deployment requirements, whereas machine learning approaches were less frequent and remained constrained by data availability, transparency, and runtime integration challenges. Most importantly, affective information was often used to support challenge calibration or related adaptation objectives, while stress, anxiety, horror, and related affective states were rarely addressed as explicit adaptation targets. These findings identify a gap within this review scope: affective information may enter an adaptive loop without making affective state the objective of adaptation.
arXiv:2505.01122v3 Announce Type: replace
Abstract: To facilitate accountability and transparency, the Digital Services Act (DSA) sets up a process through which Very Large Online Platforms (VLOPs) need to grant vetted researchers access to their internal data (Article 40(4)). Operationalising such access is challenging for at least two reasons. First, data access is only available for research on systemic risks affecting European citizens, a concept with high levels of legal uncertainty. Second, data access suffers from an inherent standoff problem. Researchers need to request specific data but are not in a position to know all internal data processed by VLOPs, who, in turn, expect data specificity for potential access. In light of these limitations, data access under the DSA remains a mystery. To contribute to the discussion of how Article 40 can be interpreted and applied, we provide a concrete illustration of what data access can look like in a real-world systemic risk case study. We focus on the 2024 Romanian presidential election interference incident, the first event of its kind to trigger systemic risk investigations by the European Commission. During the elections, one candidate is said to have benefited from TikTok algorithmic amplification through a complex dis- and misinformation campaign. By analysing this incident, we can comprehend election-related systemic risk to explore practical research tasks and compare necessary data with available TikTok data. In particular, we make two contributions: (i) we combine insights from law, computer science and platform governance to shed light on the complexities of studying systemic risks in the context of election interference, focusing on two relevant factors: platform manipulation and hidden advertising; and (ii) we provide practical insights into various categories of available data for the study of TikTok, based on platform documentation, data donations and the Research API.
arXiv:2606.04930v1 Announce Type: new
Abstract: Real-time data analysis requires the ability to accurately and adaptively address nonlinear dynamics in a nonstationary data stream while preserving computational efficiency. However, nonlinear dynamics are so complex that capturing dynamically changing nonlinear patterns and utilizing them for downstream tasks under strict time constraints is nontrivial. To bridge the gap between nonlinear complexity and computational tractability, this study applies Koopman operator theory, which states that nonlinear dynamics can be represented as linear transitions in an infinite-dimensional space. Building upon finite-dimensional approximations of this operator, we present AdaKoop, an efficient streaming algorithm for modeling nonlinear dynamics over nonstationary data streams. Our approach utilizes a probabilistic framework grounded in Koopman operator theory, treating both raw observations and reproducing kernel Hilbert space (RKHS) features as emissions from latent vectors. This dual-view formulation allows nonlinear dynamics to be expressed as a tractable linear system. Therefore, AdaKoop enables the efficient and stable modeling of nonlinear dynamics in a streaming fashion, avoiding the prohibitive computational costs of iterative nonlinear optimization. Furthermore, to address nonstationarity in data streams, AdaKoop adaptively detects the switching of patterns via statistical hypothesis testing for abrupt pattern shifts and incrementally updates model parameters to handle continuous changes. Extensive experiments on a total of 71 practical benchmark datasets across various domains demonstrate that AdaKoop outperforms state-of-the-art methods in terms of real-time forecasting accuracy and computational efficiency.
arXiv:2606.04829v1 Announce Type: new
Abstract: Building a general-purpose whole-body controller is essential for enabling diverse motion capabilities in humanoid robots across a wide range of downstream tasks, including locomotion and loco-manipulation. Different tasks rely on distinct motion reference modalities: locomotion primarily depends on coordinated robot joint trajectories, whereas manipulation requires precise end-effector trajectory tracking. Existing methods often overlook the representational mismatch between dense robot joint angles and sparse end-effector poses. To address this, we propose Multi-Modal Mimic (M3imic), a versatile multi-modal whole-body control framework that unifies heterogeneous motion reference modalities, including robot joint angles, human pose trajectories, and end-effector poses, using modality-specific encoders to map them into a shared latent space. Leveraging large-scale reinforcement learning in the simulator, we train a single policy that achieves sim-to-real transfer across multiple motion reference modalities without modality-specific retraining. Extensive simulation and real-world experiments on the Unitree G1 robot are conducted to evaluate the proposed framework. In simulation, the policy achieves a peak success rate of 98.42\% on an unseen test dataset, demonstrating its exceptional generalization capability. The code is available at https://github.com/Renforce-Dynamics/MultiModalWBC
arXiv:2606.04831v1 Announce Type: new
Abstract: Intestinal villi undergo pendular-wave motility -- an active, propagating tissue motion driven by underlying longitudinal muscles. This motility drives irreversible, counter-wave fluid pumping, akin to the antiplectic metachrony of ciliary carpets, and generates a viscous mixing boundary layer above the villi tips, whose height is controlled by flow inertia. Using a simplified 2D model of the rat duodenum, we quantify the system's viscous energy dissipation and axial pumping efficiency. In contrast to the classical Stokes' second problem, we show that the fluid volume dominating energy dissipation is dictated by the intervillous geometry, remaining insensitive to the dynamically varying viscous mixing boundary layer height. The computed pumping efficiency is orders of magnitude lower than that of canonical peristalsis for equivalent flux pumping. We thus infer that bulk fluid pumping is not the primary biophysical function of propagating pendular-wave motility; instead, we postulate that its main role is to shear the mucus barrier layer over the villi-lined mucosa. Comparing the strain rate in the barrier region with canonical peristaltic reference values for a villi-free wall strongly supports our hypothesis. Finally, for biomimetic microfluidic applications, geometric optimization reveals that pumping efficiency scales quadratically with the channel-to-villi height ratio in Stokes flow, whereas in the inertial regime, dynamic flux confinement renders this geometric optimization strategy redundant.
arXiv:2606.04613v1 Announce Type: new
Abstract: Vision-Language Models (VLMs) struggle when applied to medical image-text data, yet the tools available to diagnose this failure remain limited. Existing representation alignment metrics are symmetric, collapsing both modalities into a single score and hiding which modality drives cross-modal degradation. We introduce the Spectral Alignment Score (SAS), an asymmetric metric that projects both modalities onto the principal eigenbasis of an anchor modality and computes eigenvalue-weighted per-eigenmode correlations, resulting in directional scores whose difference quantifies modality information imbalance. We embed SAS within a benchmarking framework evaluating 15 VLMs across natural and medical image-text datasets alongside 6 alignment metrics and bidirectional retrieval. Our experiments show that medical images retain richer structural information than their paired clinical reports, a directional asymmetry invisible to all competing metrics, and that SAS achieves the strongest zero-label correlation with retrieval performance in the medical domain, positioning it as a practical diagnostic tool for clinical deployment. Code is available at this URL: https://github.com/iamalegambetti/medical-vlms-assessment.
arXiv:2606.04929v1 Announce Type: new
Abstract: LLM post-training proceeds through multiple stages, e.g., supervised fine-tuning (SFT) followed by reinforcement learning from human feedback (RLHF) or direct preference optimization (DPO), where each stage draws data from different, potentially untrusted sources. Existing literature assumes data poisoning attacks may occur at each training stage, but neglects the possibility of multiple attackers. To study the trustworthiness of the entire post-training pipeline, we propose the threat model of sequential data poisoning, where multiple adversaries separately poison the SFT and preference datasets. Under this threat model, we identify the single-attacker illusion: each adversary, evaluated in isolation, appears to pose a negligible threat. Yet when adversaries collaborate across stages, the true vulnerability is revealed. In the SFT $\to$ DPO pipeline, their contributions are additive: splitting a fixed poison budget across stages outperforms concentrating it in either stage alone. In the SFT $\to$ PPO pipeline, their contributions are complementary: neither SFT nor reward model poisoning succeeds individually, yet their combination does. These findings show that security analyses of individual post-training stages systematically underestimate compound vulnerabilities that emerge only from their interaction. Code is available at https://github.com/jcksanderson/sequential-poisoning.
arXiv:2606.04027v1 Announce Type: new
Abstract: Diffusion large language models (dLLMs) generate text by iteratively denoising partially masked sequences under bidirectional context, exposing a safety surface distinct from autoregressive LLMs. Because mask tokens are native inputs and tokens are committed by confidence rather than position, harmful content can be induced through infilling and outside the monitored prefix. Existing jailbreaks either miss this native infill capability or rely on low-diversity mask-bearing templates applied uniformly across goals, with little structural adaptation or accumulated attack experience. We propose MaskForge, a fully black-box adaptive attack that casts dLLM red-teaming as optimized search over a growing library of structural patterns. MaskForge abstracts successful attempts into reusable schemas, selects goal-compatible patterns with a UCB bandit, and invokes a scorer-guided fallback when the current library fails. Successful attempts are distilled back into the pattern library, enabling experience to accumulate across goals. Across five public dLLMs and three benchmarks, MaskForge achieves an average attack success rate of 79.3%, a 17.6% relative improvement over the strongest competing dLLM baseline. The matured pattern library further transfers to AdvBench without any updates, achieving a 88.2% attack success rate and a 67% relative improvement over the strongest competing baseline.
arXiv:2606.04321v1 Announce Type: new
Abstract: Agentic AI deployments face a recurring design tension: heavy human oversight limits scale, while broad autonomy outruns accountability. Neither posture provides the governance infrastructure required for responsible delegation. We present the Digital Apprentice, a framework for scalable, safe AI agency in which autonomy is earned, not assumed. The Digital Apprentice is a developmental learner that internalizes the tacit methodology of a directing human, graduating through per-skill autonomy tiers only when empirical evidence justifies it. The result is an agent that becomes genuinely useful over time while remaining aligned to a specific human's standards. Three architectural components make this possible. (1) Methodology capture, distilling a directing professional's tacit approach into structured assets. (2) Authorization, with autonomy escalation gated by explicit human approval. (3) Continuous alignment, correcting drift at runtime and converting each correction into owned preference data. We instantiate this framework as an inference-time control plane. We mathematically model the quality framework and discuss policies and techniques designed to raise quality. We apply the framework to an open professional corpus, and we show how catching data drift and applying a different technique at runtime recovers degraded quality dimensions under traffic shift. The implication extends beyond any single application. We believe these three pillars, stitched together as a system, form a safer and more viable path to agentic systems that can scale without sacrificing trust.
arXiv:2606.04320v1 Announce Type: new
Abstract: Relational Foundation Models (RFMs) promise a single pre-trained predictor that, given any relational database, returns predictions in one forward pass via relational in-context learning (ICL). Yet a substantial gap separates open RFMs from their commercial counterparts, and the origin of this gap has not been systematically understood. We dissect a representative framework, the Relational Transformer (RT), from two perspectives. Model side: we show that RT performs relation-level ICL, and a kernel regression view shows it fails when sparse label-cell coverage yields an underdetermined regression. Data side: we ablate RT's pre-training source and find that existing synthetic-only pre-training and in-distribution pre-training drive the same architecture into different regimes, lazy vs. feature-learning. Probing this gap reveals that the missing ingredient is a support-identifiable relational latent in the label-generation process. These two diagnoses translate into (1) a dual-stage ICL architecture that combines the relational backbone with a batch-level ICL layer lifted from a pre-trained tabular foundation model to overcome relation-level label scarcity, and (2) a homophily-aware synthetic plus continual real-data pre-training mixture, augmented with a prototype-based regularization. These choices define OpenRFM, a simple yet effective RFM that improves average task performance by approximately 30% over the RT backbone and surpasses the commercial model KumoRFMv1 on a large set of evaluation tasks.
arXiv:2601.02369v2 Announce Type: replace
Abstract: The concentration of digital payment transactions in just two UPI apps like PhonePe and Google Pay has raised concerns of duopoly in India s digital financial ecosystem. To address this, the National Payments Corporation of India (NPCI) has mandated that no single UPI app should exceed 30 percent of total transaction volume. Enforcing this cap, however, poses a significant computational challenge: how to redistribute user transactions across apps without causing widespread user inconvenience while maintaining capacity limits? In this paper, we formalize this problem as the Minimum Edge Activation Flow (MEAF) problem on a bipartite network of users and apps, where activating an edge corresponds to a new app installation. The objective is to ensure a feasible flow respecting app capacities while minimizing additional activations. We further prove that Minimum Edge Activation Flow is NP-Complete. To address the computational challenge, we propose scalable heuristics, named Decoupled Two-Stage Allocation Strategy (DTAS), that exploit flow structure and capacity reuse. Experiments on large semi-synthetic transaction network data show that DTAS finds solutions close to the optimal ILP within seconds, offering a fast and practical way to enforce transaction caps fairly and efficiently.
arXiv:2606.04384v1 Announce Type: new
Abstract: Machine learning's reliance on sensitive data necessitates privacy-preserving techniques like Differentially Private Stochastic Gradient Descent (DPSGD). However, DPSGD suffers from substantial utility degradation and slow convergence due to gradient clipping and noise injection. Prior works have attempted to improve DPSGD from various perspectives; notably, the Differentially Private Selective Update and Release (DPSUR) algorithm has achieved remarkable model utility. However, the privacy accounting in DPSUR overlooks the variation in sampling probability introduced by the selective release mechanism, which compromises the rigor of its privacy guarantees. To address these limitations, we re-evaluate the privacy analysis of the selective release mechanism and propose a novel algorithm: Differentially Private Selective Release based on Clipped Gradients (DPSR-CG). Through a rigorous, newly derived privacy analysis and extensive experiments on multiple datasets (MNIST, CIFAR-10, IMDB, and FMNIST), we demonstrate that our DPSR-CG mechanism maintains strict privacy guarantees while achieving exceptional model performance.