Science & Mathematics

We study deterministic causal quantization with pathwise error at most $\varepsilon$, using $ω$ shared representatives updated from transmitted addresses alone. For a midpoint realization with first-in, first-out replacement, we compare address entropy rates with process $\varepsilon$-entropy, the least entropy rate of any block quantizer meeting the same constraint. From $\varepsilon$-separated initialization, the gap is at most two bits per sample for every stationary source with finite scalar $\varepsilon$-entropy, without density, independence or mixing assumptions. A causal address-tree argument extends this comparison to any synchronized quantizer with cell overlap $m$, giving $\log_2 m$ bits. Lattice initialization with consistent boundary ownership realizes uniform grids with sharp one-bit scalar and process comparisons. Under an entropy condition and finite Fisher information rate $J$, the gap is at most $2J\varepsilon^2/(π^2\ln 2)$; general initializations add a term of order $ω\varepsilon\sqrt{J}$. The scalar grid estimate gives a nonasymptotic form of Rényi's theorem. Since a grid is causal, the price of causality is at most $\min\{1,2J\varepsilon^2/(π^2\ln 2)\}$ bits per sample, versus the high-resolution squared-error loss $\tfrac12\log_2(πe/6)$. These bounds certify $\varepsilon$-entropy brackets for autoregressive Laplace and Gaussian sources. For memoryless sources, the optimal causal rate equals the scalar $\varepsilon$-entropy. For a stationary process on finitely many $2\varepsilon$-separated centers, adding independent uniform noise of width $2\varepsilon$ gives exact Shannon lower-bound equality at every block length, attained by storing those centers. This includes histograms with finitely many bins. Storing two non-lattice centers beats every uniform grid of the same cell width by half a bit per sample.

Learning rate schedules are typically fixed before training begins, yet when fine-tuning large language models with low-rank adapters (LoRA), the appropriate scale varies substantially with the model, task, and adapter configuration. Practitioners re-tune a schedule for every new problem, yet a tuned schedule still cannot react to the training dynamics it encounters. We instead amortize learning rate selection across LoRA fine-tuning problems. We leave the base optimizer, Muon, untouched, so that its normalized update fixes the direction, and meta-learn only the global step size. Our method MuPACE is a small recurrent policy, meta-trained across a distribution of fine-tuning runs, that conditions on the observed trajectory and emits bounded adjustments to an interpretable analytic schedule. On 64 fine-tuning problems, MuPACE outperforms standard fixed schedules at a strong shared default on 64-78% of problems and is on par with the same schedules with per-task tuned peaks. Meta-trained only on models with up to 16 layers, MuPACE transfers one-shot to unseen tasks on models as deep as 64 layers, such as Qwen3-32B and Qwen3.8-27B, where a single run remains competitive with swept schedules.

As a prototypical latent world model, Joint-Embedding Predictive Architectures (JEPAs) provide strong semantic representations. However, dense patch representations expose a redundant interface to downstream predictors, which must implicitly infer spatial organization and temporal correspondence, presenting three challenges. First, object information is distributed across patches, forcing each downstream predictor to redundantly recover object structure that could otherwise be reused across tasks before modeling interactions. Second, the lack of explicit entity correspondence across time makes historical transitions expensive, restricting the predictor's flexibility in utilizing context. Third, without explicit transport structure, existing JEPA models typically use pointwise l1/l2 distances, which are insensitive to spatial displacement when predicted and goal supports are largely disjoint. We propose the Slot-based Joint-Embedding Predictive Architecture (Slot-JEPA), which decomposes dense representations into stable semantic slots and time-varying spatial masks. Pretrained once and frozen downstream, this decomposition exposes a reusable object-structured interface. Building on it, our Transition Memory Predictor (TMP) stores visual transitions as compact masks and retrieves useful context using statistics of corresponding slots. Its two branches separate slot-specific dynamics from shared cross-slot dynamics. We further use Unbalanced Optimal Transport (UOT) to supervise mask prediction and evaluate candidate plans by transporting mass between aligned predicted and goal masks, providing a signal better aligned with spatial progress. Across seven visual-understanding benchmarks, the slot-reconstructed features remain transferable. Slot-JEPA also improves goal-conditioned planning across six planar environments and four robotic arm planning tasks on the REALM-JEPA benchmark.

The electricity system in Great Britain (GB) allows participants to deliberately produce or consume a different volume to their traded position and has incentivised this practice with a single imbalance price since 2015. In particular, non-physical traders (NPTs) often take an open position to imbalance without any intention to produce or consume power. However, new government proposals could restrict this practice due to concerns over supply-demand balancing feasibility and costs. This article presents the first empirical analysis of speculative imbalance positioning (SIP) by NPTs in the GB electricity system through its effects on imbalance volumes, balancing costs, the imbalance price, and associated settlement costs over the period 2021-2024. Our research uses party-level imbalance data in each settlement period to run counterfactual simulations of the electricity balancing mechanism without the aggregate NPT imbalance. Our results indicate that SIP by NPTs reduced the total net balancing volume required by 3.1-3.9 TWh (16-20%), reduced system imbalance standard deviation by 48-65 MWh (13-17%), reduced balancing costs by £347-396m (5.9-6.7%), and reduced supplier imbalance costs by £193-325m (65-75%) over the analysis period. These results suggest that consumer bills would have been higher without SIP by NPTs, since the extra costs would likely have been passed on to consumers through levies, although the potential for effects in the wholesale (primarily late-stage intraday) markets complicates this picture.

Wearable devices increasingly support personalized health and well-being applications, but post-deployment data can differ substantially from those used for training, while purely local adaptation may be limited by the data available on a single device during real-world operation. Conventional federated and decentralized learning rely on recurrent communication and stable connectivity, whereas existing opportunistic approaches often exploit available encounters without adequately considering their utility, compatibility, or communication cost. We propose PULSE, an event-driven framework for personalized opportunistic learning that treats each encounter as an online collaboration decision, selecting either no interaction or a promising nearby peer. Peer gradients are incorporated only when compatible with the local learning direction, while past interactions guide future peer selection. Across human activity recognition, fall detection, and stress recognition under synthetic and real contact traces, PULSE preserves or improves personalized classification performance, while reducing peer-assisted exchanges by 80-94% compared with opportunistic model averaging. On-device experiments further demonstrate its feasibility on wearable hardware: peer-assisted operations require limited energy, while the complete implementation occupies 149.8 KiB of RAM and 561.8 KiB of flash. The obtained results show that selective opportunistic collaboration can provide effective personalization while limiting communication and resource demands on constrained wearable devices.

Electrocardiography (ECG) is a cornerstone of cardiac diagnosis, yet most machine learning methods for ECG analysis lack interpretability, limiting their clinical adoption. We present IECG, a weakly-supervised ECG foundation model that jointly performs diagnostic classification and abnormality localization without requiring manual abnormal region annotations, overcoming a long-standing bottleneck in interpretable ECG learning. Central to IECG is a weakly-supervised attention strategy that explicitly incorporates clinical priors by deriving abnormality heatmap pseudo-labels from established ECG clinical knowledge. These pseudo-labels guide the model's CLS-token attention through a KL-divergence loss, encouraging it to focus on diagnostically relevant waveform regions. During pretraining, IECG is jointly optimized with clinically guided attention supervision, ECG-text contrastive alignment, and report captioning objectives, enabling it to learn rich multimodal representations while producing localized waveform evidence. By explicitly aggregating features from clinically guided attention regions, IECG grounds its diagnostic predictions in interpretable waveform evidence without sacrificing classification performance. Extensive experiments demonstrate that IECG achieves state-of-the-art abnormal region localization while maintaining strong classification performance, advancing trustworthy foundation models for ECG analysis.

The rapid growth of scientific literature makes survey writing increasingly challenging, motivating automatic survey generation. Yet, existing systems must search an open space of papers and an open space of concepts with no trace of how the community itself has navigated them, leading to incomplete scope, weak taxonomies, and unreliable citations. Our core insight is that author-written Related Work sections contain the human judgments needed to address both limitations: authors curate the citations relevant to their contributions and explain how the cited works relate, yielding paths for scope reconstruction and relational evidence for organization. We propose InsightRAG. Its insight-guided retrieval follows author-supported citation relations to reconstruct a verified research scope and structure the resulting relational evidence. It aggregates these judgments under a minimum-conflict objective into a taxonomy that reflects recurring author-written perspectives in the retrieved literature, guides generation, and bounds each section's evidence. Results on ten topics show gains in scope reconstruction, taxonomy quality, citation support, and reference reliability.

Real-time video generation, interactive world models, and on-device image synthesis are almost a reality. Diffusion Transformers (DiT) achieve excellent sample quality, shifting the bottleneck to the throughput of the models themselves. Natural images are inherently sparse, as large portions of an image are often background, contribute little to the scene content, and are easy to denoise. DiTs fail to exploit this sparsity, as they process image tokens statically and uniformly. We propose MoDDiT, a simple architecture modification that learns to route tokens through both MLP and self-attention operations. By routing tokens through both blocks independently, MoDDiT allows for dynamic compute allocation in MLPs that varies across tokens, images, and denoising timesteps. Using MoDDiT to train image generation models from scratch significantly reduces computational costs of training and inference, processing as few as 12.5% of the tokens. By increasing sparsity and model depth, MoDDiT reduces costs and improves generation performance over unrouted baselines. The learned allocation is interpretable, revealing how computation varies with visual content and noise level. We also use MoDDiT to train video generation models, where allocation correlates with motion, and to distill a dense bidirectional video model into a sparse autoregressive student.

AI-generated images linked to news events and online claims can mislead audiences, enable fraud, or cause public controversy. We introduce WildAIGC-Bench, built through recurring collection across news websites, fact-checking sources, and related social media content. The pipeline combines broad coverage, timely updates, and human-verified labels through scheduled discovery, automatic screening and preliminary annotation, and human review of image-specific evidence. The benchmark contains 407 verified AI-generated or edited images, each accompanied by a source webpage link and an evidence summary. Images retain the conditions encountered online, including compression, screenshots, and recapture. Evaluation of dedicated visual detectors, multimodal large language models, and a simple Perception Encoder baseline reveals substantial recall gaps relative to existing benchmarks. WildAIGC-Bench provides both a practical detection challenge and a basis for continued data expansion and source-based verification. The dataset is released at https://github.com/YuZijian/WildAIGC-Bench.

Existing continuous-time graph learning methods rely heavily on complex neural architectures to encode ever-expanding interaction histories. However, the future conditional distribution induced by complex histories can often be characterized by a low-dimensional predictive state manifold, implying that only a few predictive degrees of freedom are essential for future interactions. We therefore argue that continuous-time graph learning should identify low-dimensional states sufficient for future prediction rather than reconstruct the full history. To make this perspective measurable, we define and independently estimate the intrinsic predictive dimension, which quantifies the effective predictive degrees of freedom required by a given task. Specifically, we remove the history-independent marginal from the history-conditioned future distribution, decompose the resulting predictive residual operator, and estimate the minimum effective rank needed to preserve the predictive power of the full history. Experiments across datasets show that model performance saturation is largely determined by the intrinsic predictive dimension. Based on this finding, we propose PreDyG, a lightweight model whose capacity is matched to the intrinsic predictive dimension. PreDyG achieves superior predictive performance with only 0.07\% of the parameters of a representative complex baseline.

Variable voxel spacing changes more than image resolution: it alters the discrete sampling lattice, tensor geometry, and anatomical evidence available to a 3D segmentation network. While conventional pipelines restore images to a canonical grid, direct-grid deployment requires the model itself to tolerate these changes. We observe that candidate grids are not exchangeable views -- a coarse grid removes more structural information than a mildly resampled one -- yet standard consistency objectives assign them equal influence. We therefore propose LatticeRank, a training-only framework that aligns candidate posteriors in physical space and allocates consistency supervision by voxel-spacing order: among endpoint-matched candidates, the auxiliary loss concentrates on the coarsest requested grid, while inference retains a single network forward with no candidate branch or ensemble. On MSD Task04, LatticeRank improves coarse-grid Extra Dice over mean consistency by 0.03436 across five seeds while meeting the prespecified reference-grid noninferiority criterion. A controlled ladder shows that grid exposure provides the largest recovery, posterior consistency adds a second gain, and spacing-ranked allocation yields the strongest direct-grid result; requested spacing thus provides a simple, model-independent ordering signal for direct-grid consistency learning.

Persuasive writing exploits cognitive biases, recruiting a reader's mental shortcuts to shape belief, often without the reader noticing. We call the linguistic features exploiting such mental shortcuts cognitive bias triggers. As large language models (LLMs) increasingly mediate how people encounter news, summarizing it, ranking it, and answering questions over it, a natural question is whether they can recognize the features of a text that bias a human reader. We introduce BAIT (Bias Annotation In Text), a resource for detecting cognitive bias triggers in news. BAIT pairs a schema of nine triggers, each specified in an annotator codebook, with 50 English news articles annotated by trained annotators at the article and span levels and adjudicated into a gold standard. On this data we define two tasks, detecting which triggers a text contains and locating the spans that carry them. We then benchmark six frontier LLMs zero-shot and with three or ten in-context examples. The models approach human agreement on which triggers an article contains, but rarely recover its full set of triggers, and they reach less than half of human agreement in locating the passages that carry them. Moreover, they over-predict triggers that come with easy surface cues, and most of the passages they mark for these triggers are not triggers, yet carry the trigger's surface cue far more often than random text. These findings suggest that the evaluated LLMs respond to surface cues rather than to the mechanisms that make a passage manipulative. We release the schema, the codebook, standoff annotations, and the evaluation pipeline.

AI agents keep long-term memory, much of which comes from web pages and tool results. Agent harnesses, the software that runs the model, do not record who may act on this memory, so an instruction hidden in a fetched page can be treated as the user's own instruction. We propose ClawMMU, a protection layer that labels every piece of memory with its owner, scope, origins, and authority. ClawMMU checks these labels whenever content enters the model's context, is written to storage, returns from storage, or drives a tool call. Its central rule is that information may rise in authority, but authority never lifts itself. Because the checks depend on where content came from, not on what it says, memory attacks become classical protection faults that the harness stops even when the model is fooled.

This paper addresses decentralized estimation and control of heterogeneous multi-agent systems communicating over directed switching topologies. The objective is to satisfy cooperative tasks that require coordination among agents according to a time-invariant task graph. To address the potential mismatch between the communication and task graphs, each agent is equipped with a k-hop Prescribed Performance State Observer (k-hop PPSO) to estimate the states of agents up to k communication hops away using only information exchanged with its 1-hop neighbors. A novel prescribed performance funnel construction is developed to account for the switching communication topology, and a sufficient lower bound on the dwell-time is derived to guarantee the prescribed performance specifications under topology variations. Further analysis shows that, if the desired control objective is achieved with exact knowledge of the required nonlocal states and the resulting closed-loop system is uniformly set-ISS with respect to the state estimation error, then replacing these states with their estimates guarantees convergence to a neighborhood of the desired set. The size of this neighborhood can be made arbitrarily small by appropriately selecting the observer performance bounds. A simulation study demonstrates task satisfaction under directed switching topologies.

Sparse attention reduces the computational cost of long-sequence video Transformers, but three-dimensional token layouts often leave selected blocks partially occupied, wasting physical tile capacity. We present KPage, a block-sparse attention scheme that packs valid 16-row K/V pages across logical-block boundaries into large tiles directly during loading, without materializing compacted tensors in high-bandwidth memory. KPage also routes Q blocks between 64x256 and 128x128 tiles to reduce query-side padding. Exact masking and K/V-owned page packing preserve the original sparse graph in forward and backward execution. Across five representative workloads, KPage achieves 1.35x-2.06x bare forward-kernel speedup over NVIDIA Block-Sparse Attention (BSA), with a geometric mean of 1.77x. It is near parity for fully occupied blocks (1.03x) but achieves 2.29x geometric-mean speedup on 3D layouts with partial blocks. KPage reduces physical query-key pair capacity from 334.1G to 177.5G while preserving 160.6G useful interactions. Across eight layouts from open-source SSTA, STA, and VSA systems, KPage achieves 1.13x-4.25x forward and 1.02x-2.71x forward-plus-backward speedup over the strongest exact sparse baselines. With independent Q/K partitions, KPage also achieves 1.49x speedup over the strongest exact sparse baseline and 1.57x over BSA on the SVG2 Wan 14B/720p configuration. After compliance review, we will open-source KPage and submit a pull request to NVIDIA cuDNN Frontend.