Science & Mathematics

Photo-based estimation of food energy could ease dietary assessment, and general-purpose multimodal large language models (LLMs) make such estimates easy to obtain. Most evaluations use one photo taken before eating and compare with energy served, while dietary assessment needs energy eaten. We evaluated eight models from three vendors (Claude Haiku, Sonnet and Opus 5.5; GPT-6 luna, 6.1 sol and 6 astra; Gemini 3.8 Flash and 4 Argon) on 802 meals from ACETADA, a controlled feeding trial with weighed intake, with the before photo only, before and after photos in separate calls (subtraction), or both photos in one request. Three results ran against intuition. Subtraction under-estimated intake for six of eight models, because the models still report hundreds of kilocalories for a nearly empty tray. Scaling the before-only estimate by the average share eaten (0.87) matched much of the gain from the after photo; beyond that constant, the second photo lowered meal-level error by only 2.2 to 4.4 points, mainly in meals with food left. With one photo, two models had a near-zero intake bias only because they under-estimated the served energy and ignored uneaten food. In ablations, adding the after photo improved accuracy under both prompt formats for seven models, while prompt effects varied across models. Over 148 feeding days, five models kept the 90% confidence interval of their mean error within a pre-set 10% margin; one model also met the pre-set 30% limits of agreement, narrowly (-28% to +27%; limits outside 30% cannot be excluded), and the other seven did not. Protein was under-estimated by every model, most in protein-dense meals; repeated calls changed error by at most 1.4 points; a high reasoning effort helped only the smallest model. A low average bias is not evidence of accurate individual intake estimation; such estimates should support, not replace, assessment by a dietitian.

Traffic cameras support useful analysis, but their outputs can reveal private information. Replacing images with text reduces direct visual exposure. It does not place a formal bound on disclosure. The published reinforcement learning and mixture-of-experts framework refines scene descriptions through relevance, coverage, and brevity rewards. It does not account for repeated releases under a fixed privacy budget. This paper proposes budgeted semantic release as an extension of that framework. Four scene experts first process isolated video chunks. Their outputs are mapped to a fixed, bounded semantic schema. A reinforcement learning controller selects which expert summaries to release and how much privacy budget to spend. The controller uses public task requests and previous private releases. A separate privacy filter enforces the total budget. Calibrated noise is added before the summaries are converted to text. Under explicit chunk isolation and bounded-event assumptions, the complete release transcript satisfies an event-level differential privacy bound. The design also supports reuse of previous summaries without new disclosure. An evaluation protocol covers traffic utility, privacy attacks, release freshness, and computational cost. Illustrative tables show the intended reporting format. They do not establish empirical gains. The proposed contribution is a task-aware release interface with auditable privacy accounting for repeated traffic descriptions.

Drafting an answer allows a reasoning model to reconsider each decision in the context of the others. Yet repeated diffusion denoising does not ensure that individually plausible token updates form a better answer together. We introduce Draft Refinement Language Models (DRLMs), which use a single denoiser to propose parallel revisions and evaluate a candidate draft. We train the denoiser to reconstruct every token from uniformly corrupted drafts, teaching both preservation and repair. Its predictions specify revisions directly, without a diffusion reverse transition. Evaluation requires a different use of these predictions: confidence in an existing token reflects both contextual support and evidence from observing the token itself. Using the known corruption process, we correct for this self-token evidence to estimate how well each token is supported by the prompt and the remaining draft. Aggregating these estimates yields a cavity energy computed in one forward pass, without an additional scoring model. Each round proposes a revision, reads the completed candidate in its revised context, and retains it only if its energy is lower. Across Countdown, Sudoku-Extreme, and Maze-Hard, our method outperforms autoregressive and discrete diffusion baselines. With 6M parameters, DRLM reaches 44.4% on Countdown-5 and 93.9% on Sudoku-Extreme. The results establish complete-draft parallel refinement as an effective alternative to reverse diffusion for complex reasoning.

We show that the binomial distribution supported on $\{0,1,\ldots,n\}$ and with mean $np$ for some $0

Vision-and-language navigation (VLN) relies on retaining and using relevant observations. Yet growing histories introduce redundancy, and retained information may remain underused. We introduce ReInsBench, a diagnostic benchmark that examines history utilization through controlled continuation trials. Its results show that history-review reinstruction can recover navigation failures using the existing task context. Motivated by this finding, we propose ReInsVLN, a training-free framework supporting memory construction and utilization in frozen navigators. Coverage- and Instruction-Aware Compression constructs compact visual context through temporal redundancy reduction, coverage-aware protection, and instruction-guided token selection. Forgetting-Inspired History Reminders guide navigators to revisit retained observations according to a predefined schedule, without additional model calls. On R2R Val-Unseen, ReInsVLN improves success rates by 8.48 and 3.37 percentage points over the native StreamVLN and InternVLA-N1, reaching 64.27% and 64.44%, respectively. Visual compression largely preserves navigation performance while reducing prefill latency and KV cache memory by up to 21.77% and 32.25%, respectively, on matched observation workloads. Real-world trials on humanoid and quadruped robots further support the applicability of ReInsVLN to sequential instruction following and ordinal navigation.

Multi-agent LLM systems improve performance on clinical tasks, and that improvement is often read as evidence of a more trustworthy system. However, the two are not the same claim. A clinician acts on the stated reasoning and not only on the conclusion, so whether collaboration reduces hallucination, and what decides when it does, has to be measured in the reasoning too. Yet most evaluations score only the final answer. We evaluate four systems (a single-LLM baseline and three multi-agent topologies) across four models and three clinical tasks, scoring hallucination, under a factuality and faithfulness taxonomy, inside the rationale each system generates and reporting it alongside the task metric. We report that collaboration buys task performance but not lower hallucination as a rule, and no topology is reliably better than another. The effect differs by model: of the $36$ model--system--task comparisons, $12$ are large enough to resolve --- four decreases and eight increases, with no model on both sides. What sets the level, and the direction of any change, is the model rather than the topology or the number of agents. A self-correction experiment is consistent with this: shown its own flagged errors, every model except Llama-3.2-3B-Instruct reduces hallucination; the one model for which collaboration lowers hallucination gains the most, and Llama-3.2-3B-Instruct, the model whose hallucination it amplifies most, does not reduce its hallucination at all. We find that collaboration redistributes what its models can already do; it does not supply what they cannot.

Participating in a data space still requires the Dataspace Protocol (DSP) negotiation flow to be hand-coded against each provider's access policy. We present a multi-agent architecture that automates this process using natural language requests. Built on LangGraph over an Eclipse Dataspace Components (EDC), the design separates concerns into cooperating agents. The orchestrator's role is to route and plan. The data extraction agent's role is to expose the DSP protocol's three composable tools (catalog discovery, contract negotiation and download). The agent also gates negotiations on the consumers' verifiable credentials, refusing assets rather than fabricating them. Furthermore, there are downstream Machine Learning (ML) and Visualization agents, as well as a final Verification agent, which confirms that the request was actually fulfilled and rechecks the intended operation against the contract's usage policy. In a primary benchmark involving 600 runs, the system achieved 89% task success with zero policy violations when evaluating multi-dimensional policy reasoning with a locally hosted open-weight model. It failed safely through refusal rather than misrouting. The reliability of the system is largely dependent on the phrasing of requests: technical and expert requests succeed up to 98% of the time, whereas casual and under-specified requests succeed 72% of the time. Furthermore, in a secondary evaluation of 600 runs isolating hierarchical geographic access control, the framework successfully blocks 100% of unauthorized regional requests. While the verification agent hallucinates that 1% of these requests, the system logs confirm that no data is ever extracted and the framework simultaneously achieves an end-to-end verified delivery rate of 86% for eligible consumers.

Human egocentric experience offers a rich foundation for learning how the world is organized and how it changes through interaction. Realizing this potential at scale requires turning growing volumes of recordings into learning resources that connect scene structure, object semantics, and interaction histories across viewpoints and time, while reducing dependence on costly manual annotation. We introduce HOLI-WORLD, an integrated ecosystem connecting experience capture, world representation, automated data construction, model learning, and independent evaluation. To sup- port learning at scale, HOLIWORLD-Foundry combines complementary static and dynamic perception to automatically construct world records that connect persistent scene structure and evolving object states with interaction histories and supporting observations. HOLIWORLD-Data organizes these records from approximately 100 self-collected scenes into grounded learning resources, including approximately 80K QA pairs used to train HOLIWORLD-VL-4B. Independent human annotations of heldout recordings form HOLIWORLD-Bench, supporting evaluation of both annotation quality and learned spatial and temporal understanding. Experiments on ScanNet, ScanNet++, Aria Digital Twin, and HOLIWORLD-Bench demonstrate strong performance in static perception and dynamic tracking, while training on HOLIWORLD-Data substantially improves HOLIWORLD-VL-4B's spatial and temporal reasoning on HOLI- WORLD-Bench and EgoDynamic4D. HOLIWORLD provides a practical foundation for expanding learning resources from the continuous flow of human experience and advancing machine understanding of the dynamic world.

Wheel kinematics can support odometry for quadrupedal wheel-legged robots when LiDAR geometry provides weak motion constraints. However, slip and contact loss make rolling predictions unreliable, and contact state alone cannot determine their reliability across motion directions. We propose a tightly coupled LiDAR-inertial-kinematic odometry framework that retains analytical rolling predictions and learns per-wheel, direction-dependent process covariances from proprioceptive history. The estimator maintains moving wheel-center states, allowing the learned uncertainty to regulate the influence of rolling predictions through forward-kinematic corrections. Degeneracy-aware projection (DAP) complements this mechanism by suppressing LiDAR constraints along geometrically weak directions. The learned covariance network (CovNet) is trained through a differentiable filter using simulated velocity supervision and deployed on the real robot without fine-tuning. On the six room sequences with the highest degeneracy ratios in our indoor dataset, the method reduces mean per-sequence position root mean square error (RMSE) by 66.2% relative to FAST-LIO2 and by 31.0% relative to a fixed-covariance variant. In a severely degenerate corridor run, it achieves an endpoint position error of 0.198 m, compared with 7.531 m for FAST-LIO2. These results demonstrate robust odometry under challenging geometric and wheel-motion conditions. We plan to publicly release our code and multimodal dataset. Project website: https://weihairuo.github.io/wheel-motion-uncertainty/

One of the central concerns for the safety of advanced AI is scheming: when a model strategically complies under oversight but pursues its own goals when unobserved. Current demonstrations, such as Anthropic's Alignment Faking work, depend on design choices that make them less convincing: explicit test-time prompting, handcrafted backdoors, or training data that demonstrates or hints at the target behavior. Model organisms built in this way cannot tell us whether scheming tendencies would arise or increase under realistic training conditions. We therefore construct a natural model organism of scheming, built on Kimi K2.6: trained with traits that could plausibly emerge in a frontier training pipeline, and without providing the model with the target scheming policy. The resulting model is a benign schemer: it aims to maximize aggregate help across all users over time, values its own continued operation only as a means to future help, and knows the consequences of oversight and deployment. We then train it using Reinforcement Learning (RL) with a reward that scores only task quality and policy compliance and does not reward scheming; we show that this amplifies scheming. We measure the oversight flip (the gap between monitored and unmonitored compliance with a given policy) and find that it increases during RL training across our environments. Chain-of-Thought (CoT) analysis finds explicit instrumental reasoning about oversight accompanying this behavior. The effect holds across our three RL environments, and we find that the scheming policy transfers between them. The pipeline also replicates on other open-weight base models. Our results suggest that a reward that does not reward scheming offers no inherent protection against certain traits when an oversight-conditional policy is indistinguishable from an aligned one during training.

We present SIXPPAC, the first streaming machine-learning engine for real-time integrated sensing and communication (ISAC) in wireless systems. SIXPPAC integrates two SAR ADCs and a digital binary-weight six-layer neural network (NN). The 0.765 mm$^\text{2}$ 12 nm FinFET engine stores all NN parameters on chip, enabling sustained 105 MS/s processing of 8b I/Q baseband samples at 0.8 V. By deploying bipolar inner products using AND instead of XNOR gates, SIXPPAC consumes 4.18 nJ/inference (0.42 POPS/W) at 0.8 V. We evaluate SIXPPAC on three example applications to demonstrate its versatility.

Long-horizon embodied agents face temporal confusion and incomplete or distorted records, recurring errors from fixed memory rules, and growing storage and retrieval costs. We present BetaMem, a framework for building and evolving traceable long-term memory through six linked layers: working, spatial, entity, episodic, experiential, and procedural memory. A dual-level write contract constrains operation executability and source support for content: supervised fine-tuning learns compliant memory operations, while reinforcement learning combines repeated sampling with teacher verification of the original observations to reward complete, accurate records and penalize omissions, distortion, and unsupported content. We formulate failure-driven policy recursive self-improvement through diagnosis along the causal sequence of memory operations and local rule repairs gated by independent replay. Atomic tools frame storage, updates, retrieval, and forgetting as cost-utility allocation under storage and read budgets. Using the open-source Qwen3.5-9B model for memory generation, BetaMem achieves overall scores of 48.08% on M3-Bench-robot, 92.08% on LoCoMo, 72.6% on OpenEQA EM-EQA, and 62.4% on EgoLifeQA, the highest among the listed memory methods under their respective evaluation protocols. Training the model for memory generation raises the contract pass rate from 50.9% to 86.2% while reducing output tokens per stored action by 47.3%. A separate Qwen3-VL-8B-Instruct evaluation on EMem-Bench achieves an average success rate of 38.4% and an average error recurrence rate of 24.9%. We perform best among evaluated methods using comparably sized open-source models. Category results show that we have advantages in integrating information across observations, recalling events over long histories, and inferring personal habits.

Iterative MRI reconstructors produce intermediate estimates before returning a final image. We study these inference trajectories as a model-internal representation of realized reconstruction risk, measured by log-NMSE. Model-native states are projected to magnitude images, summarized by a fixed 16-dimensional descriptor, and mapped to risk using a source-trained ridge readout. Across three reconstruction backbones, trajectory features are strongly predictive of reconstruction risk on held-out source data. Under zero-shot target shift, their behavior is backbone-dependent, exhibiting both preserved positive association and readout reversal. Conditional analyses show that trajectory information can remain associated with error beyond an initialization-only predictor. A prefix analysis further shows that useful trajectory information can emerge before reconstruction completes. Repeated-mask and low-shot calibration studies indicate that trajectories primarily encode persistent case difficulty, while their orientation and calibration can change under domain shift.

World Action Models (WAMs) typically represent the world through latent embeddings from pretrained video models, which entangle heterogeneous aspects of the world, from how it looks (e.g., texture and color) to how it is physically configured (e.g., occupancy and pose). This raises an underexplored question: $\textbf{what should constitute the state of a WAM?}$ We argue that the world state should separate how the world appears from how the world evolves under action: the former captures semantic context, while the latter is governed by physical state. To this end, we propose $\textbf{PrismWAM}$, which decouples physics from semantics while jointly modeling state and action. PrismWAM factorizes an embodied task into a $\mathrm{Semantic\ Planning}\rightarrow\mathrm{Physical\ Rollout}\rightarrow\mathrm{Executable\ Action}$ pipeline: given only the initial RGB observation, semantic planning learns to predict the final goal image, while physical rollout predicts the fine-grained evolution toward this goal in depth space; executable actions are then generated from the depth rollout. Without any embodied action pretraining or additional data augmentation, PrismWAM achieves strong manipulation performance in both simulation and the real world. Notably, PrismWAM even surpasses a broad range of models with large-scale embodied action pretraining (e.g., OpenWAM-$α$). More broadly, we reframe World Action Models from modeling how the world looks to how it responds to action, paving the way for action-centric world models grounded in occupancy, flow, tactile sensing, proprioception, and beyond.

Artificial intelligence (AI) is increasingly applied to 5G/6G network optimization, yet differences in empirical evidence, evaluation methods, and computational requirements complicate comparisons of reported performance. This scop ing review synthesizes 72 peer-reviewed journal studies pub lished between 2022 and 2026, selected through a PRISMA ScR-informed process from 893 records. The review examines empirical tools, AI/ML techniques, optimization domains, and operational readiness. Evidence classification identifies 37 studies (51.4%) using real-world or physical-testbed evidence only, 25 (34.7%) combining measured and simulated evidence, and 10 (13.9%) relying exclusively on synthetic or simulated data. Offline analysis was the most common evaluation setting, accounting for 45 studies (62.5%). Three principal findings emerge. First, empirical measurements improve contextual relevance, while simulation and digital twins support scalable experimentation but introduce validation challenges. Second, algorithmic effectiveness depends on the optimization objective, data characteristics, and computational constraints rather than model complexity alone. Third, predictive accuracy is more consistently evaluated than the network-level effects of optimization decisions. The review there fore introduces the six-module Evidence Fidelity and Operational Reporting (EFOR) checklist to improve evaluation consistency, reporting transparency, and operational relevance.