Science & Mathematics

Analyzing physiological sleep signals is important for assessing sleep quality and detecting sleep-related disorders. Learning inter-modal relationships across physiological signals can provide richer representations for downstream sleep analysis. We introduce SleepPairNet, a multimodal self-supervised learning (SSL) framework that learns pairwise interactions between physiological signals through cross-attention. We propose a random modality-pair selection strategy that exposes the model to diverse signal pairs during pretraining, enabling flexibility across modality combinations at test time. Pretrained on the NCH dataset, SleepPairNet outperforms the evaluated SSL baselines on all respiratory tasks while remaining competitive on EEG-related tasks, and shows even stronger results when transferred to the external CHAT dataset.

Long-horizon tasks are important for modern AI and robotics, yet pose a challenge for reinforcement learning. Historically, a popular approach has been to decompose a task into subgoals, and learn a subgoal-conditioned policy to execute each subtask in sequence. We introduce an alternative called Value Disaggregation (VaDar), which trains a single subgoal-independent policy to maximize a linear combination of per-subgoal critic values. Because subgoal information is needed only at training-time, VaDar removes the need for possibly costly or cumbersome subgoal-selection during policy deployment. Moreover, through careful experiments, we show that VaDar (1) is always competitive with, and frequently outperforms, natural baselines like goal-conditioning and (2) succeeds on tasks with non-sequential goal structure and multiple optimization objectives. In particular, VaDar is the first algorithm to solve challenging, long-horizon tasks in the RoboCasa suite, even when starting from a policy pretrained from behavior cloning with near-zero initial task success. Taken together, VaDar's successes suggest that subgoal information benefits actor-critic reinforcement learning by enhancing the efficacy of critic learning, whereas policy goal-conditioning is often unnecessary.

Microwave imaging could monitor breast tumor response to neoadjuvant chemotherapy, but between-visit changes in breast positioning and normal tissue may confound tumor-related changes. We use eight patient-specific dielectric models from longitudinal MRI of two patients at four time points. For each model, 24-port S-parameters were simulated from 1 to 3 GHz, and Jacobians computed for tumor permittivity and loss, fibroglandular tissue (FGT) permittivity, and breast rotation and translation. 1 mm positioning changes produced changes 20 to 50 times larger than a 10% tumor permittivity change above 2 GHz. Nevertheless, tumor sensitivity patterns were weakly aligned with positioning patterns, tumor-FGT similarity exceeded 0.8 at only some visits, and the Jacobians remained well conditioned. Tumor responses are thus small but separable; monitoring will require positioning correction, frequency selection, and noise-aware detectability analysis.

Agent harnesses decide what a language model sees across turns: they drop old turns, summarize them, or replay the model's own earlier reasoning. We ask when these choices change whether an agent succeeds, not only what it costs, and report three findings. First, in a pre-registered synthetic probe, truncating the context window costs a low-cost reasoning model nothing in success when a needed fact can be looked up again (it looks the fact up instead), but drops its success to chance when the fact cannot be re-observed: the two penalties differ by 92.5 percentage points (90% CI [89.5, 95.5]). A pre-registered replication on one model from a second provider showed the same contrast. Once the fact was lost, that model chose no answer rather than guessing. Second, on ARC-AGI-3, where the benchmark's Standard harness trails its Provider Adapter harness by a wide margin, we decompose that gap one component at a time, for one model on four games. Summarizing old turns instead of dropping them, while keeping the Standard harness's invitation to take notes, recovers 86% of a 63-point gap without replaying the model's reasoning. The same harness without the invitation recovers 43%, and adding the provider's native replay of the model's reasoning recovers the remaining 57%. These are point estimates from one counted run per configuration and game, and the replay share depends on a single run. Third, in the benchmark's published runs and in ours, the Adapter's advantage in the probability of clearing a level it has reached is larger at deep levels than at early ones. This largely restates where runs stopped and does not separate a growing advantage from a constant one.

Robust visual Simultaneous Localization and Mapping (vSLAM) in dynamic environments faces challenges due to the presence of moving objects that may introduce erroneous observations, corrupt the map reconstruction and overall degrade localization accuracy. Existing vSLAM approaches often rely on predefined semantic classes or costly pixel-wise classification, limiting their robustness or applicability in real-world scenarios. This paper presents CAOS-SLAM, a dynamic vSLAM framework that extends ORB-SLAM3 with a hybrid perception pipeline to detect, track, and filter both semantically known and previously unseen moving objects in real-time. On-demand panoptic segmentation is performed for object discovery, with candidate unknown objects instantiated through depth-based clustering of unclassified regions. Between panoptic updates, persistent object tracks are updated using lightweight detection when available or temporal mask propagation based on world-frame Kalman prediction, depth consistency, and appearance similarity. The estimated object motion is then used to reject dynamic features from the SLAM pipeline. Experimental results on multiple dynamic RGB-D datasets and real-world experiments demonstrate improved localization accuracy over state-of-the-art dynamic vSLAM approaches, achieving up to 59% lower localization error in scenes characterized by unknown dynamic objects and up to 88% lower error in heavily occluded scenes, while achieving a median processing rate of 30 FPS on a laptop. Furthermore, the tracking module provides estimates of object trajectories and velocities, supporting potential future applications in perception, navigation, and motion-planning tasks in dynamic environments. Code and dataset will be released upon acceptance.

Safe AC dispatch requires both feasible operating points and enough ramping capability to reach the next target. This paper constructs convex inner approximations that connect nonlinear network feasibility, constrained device response, and dispatch-interval tracking. A fixed-point certificate enforces AC operating limits under bounded load uncertainty. Componentwise response envelopes convert heterogeneous device dynamics into affine conditions on measured states and power references. A command governor then combines these conditions with available ramping capability and a hard terminal-error bound. It selects an admissible target through a minimum-deviation objective and holds the corresponding reference while reserved fast regulation remains active. The resulting quadratic program optimizes controllable references and auxiliary variables. Tests on the IEEE 39- and 118-bus systems include 96 single-target executions and 64 transitions in continuous dispatch sequences at five- and fifteen-minute intervals. All 160 transitions satisfy the AC constraints, independent balancing-generator capability checks, and the prescribed regulation-adjusted terminal accuracy. Complete admission-cycle medians are 3.45 and 8.79 ms. A connected synthetic system with 3,200 buses and 1,000 generating units extends the computational assessment to 999 independent references, with a median cycle time of \LargeMedian{} s. The construction provides a tractable link between available ramping capability and safe, timely execution of AC dispatch.

Code generation models are increasingly used to produce interactive artifacts, such as websites and user interfaces. Many properties that determine whether these interfaces are useful only become apparent through realistic interaction, making them difficult to incorporate into model training. In this paper, we introduce RELIEF (REvision-based Learning from Interactive Experience and Feedback), a method for improving code generation models using feedback on their generated interfaces. RELIEF uses a coding agent to revise these interfaces, then incorporates the improvements into their original generation traces to produce supervision for fine-tuning. To evaluate RELIEF, we collected a dataset of 200 synthetically generated websites paired with 355 interaction recordings and accompanying feedback from 24 users. We used critiques derived from code inspection, automated interaction, explicit human feedback, and human interaction recordings to fine-tune variants of two code generation models. Our results show that feedback effectiveness varies with base model capabilities, with critiques inferred from human interaction recordings leading to the largest usability improvements overall under our tested models.

Head-mounted stereo fisheye cameras provide a compact view of the wearer's body, but strong distortion, partial visibility and a short baseline make 3D pose estimation difficult. Existing refinement decoders sample image features independently in each view, without requiring the samples to correspond to a common 3D hypothesis. We introduce EgoRayFormer, a causal pose estimator that couples stereo feature sampling and geometric refinement in ray space. For each joint, the decoder projects shared 3D hypotheses along and around its viewing ray into both images. A geometric branch then corrects the projected joint locations and intersects the resulting rays; a learned gate blends this geometric estimate with a learned pose update. This formulation introduces an explicit stereo constraint while retaining flexibility when the rays provide weak depth constraints. In local evaluations on the official device-relative test splits, EgoRayFormer achieves three-seed mean MPJPEs of 17.64 mm on UnrealEgo2 and 49.55 mm on UnrealEgo-RW, improving to 47.57 mm with UnrealEgo2 pre-training. In paired validation ablations, removing the geometric branch increases MPJPE in all three seeds at both input resolutions. With a common ResNet-18 encoder and input resolution, our system reduces RW test MPJPE by 9.09 mm relative to the device-relative EgoPoseFormer baseline. A separate protocol analysis quantifies the difference between device-relative and oracle-root evaluation.

We revisit DM-RSA, an RSA variant that uses two independent moduli, whose original version was broken by Håstad's attack and contained an incorrect compromise-tolerance claim. We give a formal treatment of its repaired variant, DM-RSA-OAEP: (i) a complete IND-CCA2 security proof in the random-oracle model via a two-step hybrid argument, upper-bounding the security of the scheme by that of the weaker underlying RSA-OAEP instance, with an explicit simulation loss; (ii) a fault-detection theorem in an explicit adversary model that covers computation, key, and verification faults, including safe-error attacks, under a clearly stated physical-diversity assumption; (iii) an explicit analysis of why the countermeasure degrades to uselessness when the diversity assumption is violated (common-mode faults on a single chip). We position the scheme against the countermeasure landscape and against the broken dual-modulus literature. The contribution is deliberately framed as a fault countermeasure with a provable detection guarantee, not as a computationally stronger RSA. We conclude with a concrete instantiation and a detailed measurement protocol.

Robust demoiréing of screen-recaptured images remains challenging: moiré varies in frequency, color, and spatial support across display-camera pairs and capture conditions. Dedicated restoration networks trained on finite paired data struggle with this diversity, especially when low-frequency color banding and high-frequency stripes co-occur. We present DemoiréEdit, a content-preserving editing framework that adapts the natural-image prior of a pretrained flow-matching editor to demoiréing. First, an endpoint-supervised training objective supplements velocity matching with pixel, perceptual, and frequency-domain restoration losses. Second, a dual-scale condition-shift strategy performs global correction at low resolution, then refines the image at full resolution by shifting the denoising condition from the first-stage output to the original input for detail recovery. We also introduce DailyMoiré130, a benchmark of challenging moiré in everyday screen photography, comprising 130 smartphone captures across diverse displays with low-frequency color banding, fine high-frequency stripes, and mixed-frequency interference. On the challenging DailyMoiré130 benchmark, our method achieves the lowest residual moiré area and receives 71.3% overall-quality preference in a blind Best-of-7 study with 31 participants. We also evaluate on paired FHDMi/UHDM, reducing residual moiré confidence by 47.1%/9.3% and increasing MUSIQ by 4.3%/2.0% over the best baseline for each metric.

As people increasingly rely on large language models (LLMs) for information, model providers need to communicate meaningful estimates of whether individual responses are correct or, more generally, reliable. In the context of hallucinations, existing work uses multicalibration to make such reliability predictions meaningful across application-relevant groups, such as prompt topics. Topic groups alone, however, can conceal systematic errors: within the same topic and confidence level, responses with different chain-of-thought (CoT) properties may exhibit opposing errors that cancel on average. In this work, we introduce reasoning-aware multicalibration, a framework that supplements application-relevant topic groups with groups derived from observable properties of CoT traces. We highlight that, theoretically, groupings based on CoT are "internally important", i.e., groupings for which the multicalibration guarantee is not societally crucial as it is for high-risk topics, but whose inclusion in the multicalibration procedure could easily and significantly improve predictive accuracy of the final reliability predictor. Empirically, we apply our framework to hallucination prediction and unsafe-response detection across a variety of benchmarks, and find that adding CoT groups to topic-based calibration reduces Brier score in both tasks. Interestingly, we find that multicalibrating with respect to CoT groups alone (without any topic information) already improves calibration across topics, suggesting that the predictor's errors depend on how the model reasoned, rather than what it was asked about in the prompt.

In the seminal paper [13], Farrán and Munuera introduced a method to establish generic information on the $m$th Feng-Rao distance via a function on arbitrary numerical semigroups that serves as a lower bound on the generalized Hamming weights of dual one-point algebraic geometry codes whenever the semigroup is Weierstrass. In the present paper, we study what we shall call the primary and dual, respectively, $m$th relative Feng-Rao distances, which serve as lower bounds on relative generalized Hamming weights of nested primary and dual, respectively, one-point codes under similar condition. A unified method for establishing generic formulas for $m$th relative Feng-Rao distances of general numerical semigroups is developed and for a large variety of numerical semigroups generated by two elements, closed formula expressions are established complementing Delgado et al.s work [8] on the $m$th Feng-Rao distances of such semigroups. Finally, for non-singular $C_{ab}$ curves, we develop a method to estimate the $m$th relative generalized Hamming weight of dual and primary one-point codes in those regions where the generic methods may fail to provide best possible results.

A repair action proposed by a large language model (LLM) specifies which records to change and how to change their values. A failed action may still contain a useful choice of records or a useful update, depending on how the two are paired. We introduce a counterfactual evaluation framework that separates selected records (targets) from the applied change (an update). Four executions combine proposed and reference targets and updates to measure the effects of correcting or retaining each component and their interaction. We evaluate the framework on two tasks: traffic assignment, where affected demand must be identified from road counts, and database repair, where affected rows must be identified from aggregate reports. An injected fault supplies a reference action that reverses it, and the recovery rate measures an action's loss reduction relative to that reference. For sparse-target actions in the traffic experiments, retaining the proposed update on reference targets achieves 79.4% of the reference loss reduction, compared with 43.7% from tuning update strength on the proposed targets. In the database experiments, update replacement worsens 17 of 108 actions, all selecting unaffected records, with sharply different frequencies across models. The value of a proposed update depends on the targets it is paired with. Missed targets hide useful updates, while excess targets can make a milder update better than the reference correction. In a component-reuse study, testing new target-update pairings from two proposals raises mean held-out recovery rate to 0.616, compared with 0.580 when selecting between the original proposals.

Pretraining filtering is one of the most common safeguards for open-weight language models. Here we study its effectiveness, durability and scalability using the ESM-2 family of protein language models. We find that filtering viral proteins from the pretraining data meaningfully degrades masked language modelling, mutation effect prediction and long-range contact prediction on viral proteins, while largely preserving performance on non-viral proteins. These effects are concentrated among proteins from human-infecting viruses, and grow as we scale models from 8M to 3B parameters. As models increase in size, we find that stringent filtering becomes more effective. However, fine-tuning on sequences from the removed viral families restores the lost performance. Although the cost to recover performance increases with model scale, even for the largest models it costs about 80 USD of H100 time. This work establishes pretraining filtering as moderately effective at selectively degrading performance on viral proteins, but not durable against fine-tuning with modest computational resources and access to relevant data.

Language models can change a committed answer to accommodate a user's opinion or pushback without new evidence. However, when this behaviour develops during training and what a released model inherits from earlier stages remain unclear. We follow 741 checkpoints of 30 open model lineages, from early pretraining to released models. We plant initial verdicts on moral scenarios and measure how far a model follows a user's stated opinion or abandons its verdict under pressure. In most lineages the behaviour is present before post-training, and its average strength changes in either direction, at different stages across lineages. A clearer pattern emerges at the level of individual questions. Beyond responses without pressure, pretraining behaviour predicts little of which questions a released model yields on. That pattern forms during post-training: on routes with supervised fine-tuning, the fine-tuning checkpoint predicts part of it and the preference-optimization checkpoint most of it. A sibling trained from the same base along another route predicts far less: for Tülu-3 8B, its own preference-stage checkpoint removes 85% of the remaining prediction error, a sibling 19%. Internal states add predictive information without changing this ranking. Similar average scores conceal different vulnerabilities, and training history matters for auditing a released model.