In this paper, we consider mixed-motive games (MMGs) which model various real-world multi-agent scenarios where self-interested agents must both cooperate and compete. We identify an issue with existing gifting methods, where agents become overly dependent on incoming gifts and fail to learn reward-yielding behaviors, a phenomenon we call gift over-reliance. To address this, we propose a refusal-augmented gifting framework that balances cooperation and competition in MMGs. Our approach introduces an adaptive refusal mechanism that allows agents to selectively reject a portion of received gifts from their policy-update signals, thereby mitigating gift over-reliance and preventing the agents from being exploited by other agents. This mechanism is gifting-method-agnostic, serving as a plug-in for existing schemes while enabling decentralized learning. We demonstrate that across diverse MMG environments and gifting schemes, our mechanism consistently reduces gift over-reliance and prevents lock-in to sacrificing policies, improving $\alpha$-fairness as a result, without necessitating major changes to existing training pipelines.
Scientific discovery from the literature requires assessing connections among findings obtained under different study conditions. Source attribution makes individual findings traceable, but does not determine what they jointly support. We introduce GI-TraceBench, a benchmark of 10,282 questions in gastrointestinal oncology that unifies provenance question answering, knowledge discovery, and hypothesis evaluation through shared Paper, Claim, and Evidence Span records. We further propose FindLink, a two-stage framework for provenance-grounded scientific reasoning. Premise Grounding constructs source-bound premises with explicit study context and support status. Structured Reasoning organizes these premises into support plans that represent both supported relations and unresolved conditions, then generates evidence-conditioned answers using deterministic compiler feedback. The compiler verifies structural and source-binding consistency, but does not assess scientific entailment. On the 1,545-question fixed-evidence test set, FindLink improves the balanced family macro score from 67.95 to 87.25 over direct supervised fine-tuning of the same backbone. A retrained variant without Structured Reasoning achieves slightly higher premise-identifier F1, but substantially lower relation and evidence-sufficiency F1. These results indicate that scientific reasoning requires not only identifying grounded premises, but also assessing what connections those premises support.
Large language models have demonstrated strong capabilities in structured reasoning and code generation, yet identifying an assembly fault does not guarantee the ability to generate executable rework programs. In interlocking assemblies, a locally plausible repair program may violate blocking dependencies, omit required recovery operations, or terminate in an incorrect configuration. We identify this challenge as fault-conditioned executable rework generation, where models must jointly reason about structural failures and generate programs that restore constrained assembly states. To benchmark this capability, we introduce FORGE-Bench, containing 4,000 tasks across 26 interlocking assemblies with five assemblies held out for evaluating generalization to unseen structures. We further propose FORGE (Fault-Oriented Rework Code Generation for Executable Interlocking Assembly), a structured state-to-program framework that decomposes rework generation into Rework Structuring and Action Grounding. Rework Structuring captures fault-induced recovery dependencies, while Action Grounding maps these dependencies into state-conditioned executable actions. Experiments show that locally reasonable repair code is insufficient for reliable assembly recovery, and that connecting structural recovery reasoning with executable action realization is essential for verified target restoration.
A GUI command can be correct at the interface level yet still produce an incorrect assembly. An agent must identify the relevant components, respect their structural relationships, and determine which parts should move during recovery. We introduce MAGAR-Bench, a benchmark of 4,000 multimodal samples from 26 assembly families that separates Assembly Decision, Atomic Action, and Minimal Rework. For rework, the objective counts distinct moved parts within the validated recovery candidate set for each state, rather than the number of actions. We also introduce MAGAR, a selective correction framework built on a frozen policy obtained by supervised fine-tuning. Its key design matches correction granularity to output dependencies. Structured candidates revise assembly fields and rework scope, while complete action candidates keep native GUI primitives coupled with their parameters. A learned selector retains the original prediction unless a candidate passes a task-specific intervention threshold. On recorded states from held-out assembly families, MAGAR improves Overall, the task-balanced aggregate, from 37.50 to 41.39, with gains on all three tasks. Across 209 interventions, 70 correct an error and 22 introduce one. These results support task-specific correction while highlighting the remaining tradeoff between fixing errors and preserving reliable predictions.
System identification studies the canonical problem of reliably learning system-dynamics from state and input trajectories. One of its important applications consists of estimating a feedback gain matrix that can stabilize an unknown system. This paper focuses on the analysis of such adaptive stabilization schemes for multivariate linear time-invariant state-space models. The system instability practically renders trajectories short, i.e., the observed data is notably limited. So, it is crucial to tightly consider effects of \emph{all} involved quantities including length and number of trajectories, dimension, controllability Gramian, and signal-to-noise ratio. We put forward a simple, practical, and reliable relationship between the chance that adaptive stabilization (based on system identification) fails to stabilize the system, in terms of all the above-mentioned quantities. Our findings show that despite exponential effects of all other quantities on failure probability, such effects do not hold for signal-to-noise ratio. More specifically, this study shows that the strength of the excitation signal helps stabilization only \emph{logarithmically}; a very important phenomenon that we explain under the term \emph{signal saturation}.
Action-conditioned world models are increasingly trained on data from randomized experiments, and whether their predictions respond to actions is usually checked by replacing or removing the actions and measuring the loss in accuracy. Taking biological experiments as the guiding case, we show that such diagnostics report action effects where there are none: under autocorrelated designs, future actions stand in for history the model does not see. When the design law is known, a conditional randomization test written as an e-value is valid in finite samples whatever the model, provided the resampler conditions on the full state the design used. A second failure arises whenever many sites share one action sequence, as in world models: the pooled score turns small model errors along the action contrast into overwhelming evidence against a true effect. In a pixel world, its median log e-value fell from -17.3 to -6832 as frames grew from 64 to 4096 pixels, while the oracle's rose. Tempering the score changes the bet, not the validity. Our own preregistered analysis of optogenetic ERK experiments failed in exactly this way; a tempered statistic, registered before the data were obtained, then rejected the null on a never-opened field (E = 1.1e10). Every single tempering factor we tried, estimated from residuals or tuned on validation data, collapsed in some setting; mixtures over a fixed grid of factors never did. In registered tests on three world-model families trained by others, the pooled test rejected in 0 of 96 units of a diffusion model in which actions mattered, and reached the registered power bar in 10 of 26 Atari games under IRIS's own token likelihood and 1 of 6 control tasks under TD-MPC2; mixtures fixed in advance reached it in all 96 units, 24 of 26 games and 6 of 6 tasks. Under exact nulls the statistics made 4 rejections in 22,848 tests.
Future 6G wireless networks are expected to evolve toward AI-native systems that support intelligent, environment-aware services, including positioning and mapping. Current research has already integrated AI into several aspects of communication, particularly wireless channel estimation. However, most learning-based channel estimators focus on communication-oriented objectives rather than explicitly estimating multipath parameters such as delay, angle of arrival (AoA), and angle of departure (AoD). These parameters provide essential geometric information for downstream tasks including positioning and radio simultaneous localization and mapping (SLAM). In this work, we propose a Spectral-Convolutional Factorized Transformer (SCFT) for joint multipath component (MPC) estimation from noisy complex channel estimates and physical beam-pair descriptors. SCFT combines local spectral convolution, permutation-equivariant beam-set attention, and a query-based Transformer decoder to detect and classify line-of-sight and non-line-of-sight paths and estimate their observed delays, AoAs, and AoDs. Experiments on realistic ray-tracing data show that SCFT reduces the generalized optimal sub-pattern assignment error by 13.1% compared with ResNet, the strongest competing baseline. We further demonstrate this advantage in independent snapshot positioning and radio SLAM, where SCFT reduces position RMSE by 18.5% and 65.0%, respectively, relative to ResNet. These findings highlight the importance of explicitly recovering multipath geometry, as improvements in MPC estimation can translate into substantial gains in downstream tasks.
Deterministic multi-agent LLM deliberation can exactly re- play a fixed transcript, yet exact replay alone does not test local resistance to controlled internal interventions. Because each response becomes context for later agents, a changed greedy token can redirect a transcript-mediated execution. We present a white-box audit of this local failure mode in a five-agent committee under greedy decoding. We prospectively select low- and high-margin token positions, require exact no-op re- play, inject norm-relative residual displacements, and measure first transcript divergence within ten generated tokens. Low- margin positions are more branch-sensitive than high-margin controls, while matched margin-reducing gradient directions branch substantially more often than margin-increasing con- trols. We then continue selected realized branches through the committee to distinguish textual persistence from preference and final-decision changes. The contribution is a controlled audit for local branch susceptibility in this specified protocol, not an estimate of deployment frequency, decision correct- ness, or a phenomenon unique to multi-agent systems.
Floating point computing has dominated scientific computations since the 1950s because of its versatile ability to approximate real numbers with both sufficiently high precision and sufficiently great range. Since 1985, two specific formats have become standard across the industry, one with 32 bits which is adequate for many purposes and one with 64 bits which allows numbers with both much higher precision and much greater range for applications where these are needed. However, during the past decade, deep learning has come to dominate computation, and narrower floating point formats with lower precision and reduced range have steadily become more common. In this article we discuss this phenomenon, why it is happening, how various 16-bit, 8-bit and even 4-bit floating point formats have been developed, and efforts to standardize these formats.
Multiple local updates are widely used in decentralized optimization to enhance communication efficiency, often in conjunction with stochastic gradients when exact gradient computation is costly. However, such algorithms generally converge only to a neighborhood of the optimum, whose size depends on the stepsize and the variance of gradient noise. Variance reduction has been incorporated into ADMM-based methods to achieve exact convergence in expectation, but their stepsizes typically scale inversely with both the square root of the network size and the cube of the number of local updates, leading to slow convergence as the network size or the number of local updates increases. In this paper, we propose a novel primal-dual-based decentralized optimization algorithm that integrates variance reduction with multiple local updates, with a stepsize independent of the network size and inversely proportional to the number of local updates (rather than its cube), thereby enabling significantly larger stepsizes and faster convergence. Moreover, to the best of our knowledge, we establish the first linear convergence guarantee in expectation for decentralized optimization with multiple local updates under stochastic gradients in the strongly convex setting. Beyond the theoretical guarantee, the proposed method exchanges only one variable per round, halving the communication cost compared with gradient-tracking-based methods. Numerical experiments further demonstrate its advantages in convergence speed and accuracy.
A score-based diffusion model is trained on the two-dimensional classical Heisenberg model, the lattice representation of the $O(3)$ nonlinear sigma model. Neural networks are trained on three different coarse-grainings of the same training set to implement a multiscale sampling scheme that modestly improves both the accuracy and computational cost of direct sampling on the finest lattice. Stein (or Schwinger-Dyson) diagnostics for the two-point correlation function are introduced to assess the quality of the learned score without relying on reverse diffusion sampling. These diagnostics are also incorporated into an additional training loss. We discuss several technical considerations relevant to implementing diffusion models on $S^2$. These include a consideration of lifting the network input to $S^3$, as in the $CP^1$ representation, and developing a boundary-layer approximation to the short-time spherical heat kernel that extends its usefulness to larger diffusion times.
In many settings where LLM post-training is desirable, high-quality answers are recognizable but difficult to score with explicit rewards. Reinforcement learning with verifiable rewards (RLVR) is limited to tasks where an explicit correctness signal is available. Prior adversarial approaches to training reasoning LLMs either complement verifiable rewards with co-trained critics or require reference reasoning traces. We ask whether a co-trained LLM judge can turn reference answers alone into the sole reward for on-policy post-training of reasoning LLMs. We call this method Reinforced Adversarial Learning (RAL). We first validate RAL on three verifiable tasks (integer multiplication, code-based arithmetic-expression synthesis, and SMILES-to-formula translation), using the verifiers only for evaluation. With sufficiently capable judges and enough training, RAL matches direct RLVR. We then apply RAL to scientific abstract generation in two settings. In the first setting, RAL guides the generator to produce abstracts in the scientific domain of the reference abstracts, even though the prompt leaves that domain unspecified. In the second setting, we generate abstracts for U.S. National Institutes of Health grant proposals conditioned on the grant title. Human scientists and an ensemble of LLM judges find our generations harder to distinguish from real abstracts than those of frontier-LLM baselines. These results suggest a path toward post-training reasoning models in scientific and other high-value domains, where increasingly capable judges infer hard-to-specify reward functions from example answers.
Many open mixture-of-experts (MoE) language models train with Muon instead of AdamW. Most balance the load with a bias on each expert's routing score, adjusted after every step. Their recipes disagree on whether Muon should also train the router. In models with 55M and 108M active parameters, we measure how the optimizer changes the span of these biases, the largest minus the smallest in a layer. The span grows when all tokens favor the same experts, so it shows how hard a layer is to balance. At learning rates tuned on shorter runs, Muon raised the span of the first MoE layer at every seed, roughly doubling it on average, while the other layers changed much less. The rise came from Muon on the rest of the network, whichever optimizer trained the router. In those runs, the inputs to the first router shared a direction, and soon after the warmup its busiest expert averaged four to seven times its fair share. Over the first 3,000 steps, subtracting the batch mean from these inputs removed this overload while barely changing the loss. On the smaller model, halving Muon's learning rate removed most of the rise and slightly improved the loss. Moving only the router to AdamW made the loads shift more during the warmup and slightly raised the loss, so at this scale the results favor keeping the router on Muon.
Synthetic text generation offers a promising approach to alleviating data scarcity in low-resource languages. However, its effectiveness is inherently limited for languages that are scarcely represented, or entirely absent, in an LLM's pretraining data. At the same time, many such languages have human-curated vocabularies that provide rich lexical and semantic information, often accompanied by grammatical information such as part-of-speech labels. In this work, we ask whether LLMs can leverage these resources to generate coherent text in extremely low-resource and unseen languages. We introduce a framework that combines human-curated vocabularies with a small number of human-written exemplars that provide syntactic signals, while sampling candidate words according to a Zipfian distribution. Across languages with varying levels of resource availability, we evaluate the generated text using lexical diversity, off-target ratio, and human judgments of fluency. Our results show that reasoning-enabled LLMs can generate fluent text even in extremely low-resource languages, with particularly strong results at the sentence level and promising performance at the paragraph level. Our findings highlight a practical pathway for bootstrapping synthetic data in low-resource languages, expanding the amount of usable text even when existing digital resources are extremely limited. Our source code is available at https://github.com/michaelsinanta/vocab2text.
Retrieval-augmented generation (RAG) systems ground large language models (LLMs) by retrieving external evidence and conditioning generation on the retrieved context. Although dense retrieval is effective for local semantic matching, independently ranked retrieval can return redundant passages, miss cross-chunk dependencies, and provide incomplete evidence for multi-hop reasoning. We propose SpectRAG, a spectral graph-augmented retrieval framework that combines hybrid candidate seeding, weighted chunk graph construction, graph-based expansion, spectral relevance propagation, and token-budgeted context selection. Document chunks are represented by dense semantic embeddings and graph-spectral coordinates derived from a weighted chunk graph whose edges encode semantic similarity, document adjacency, section hierarchy, shared entities, citation links, and similarity between LLM-generated utility questions. At query time, SpectRAG builds a high-recall seed set using exact inner-product search and hybrid relevance signals, expands it over the chunk graph, re-scores candidates using graph proximity and spectral information, and assembles the final context under a fixed token budget. The goal is not merely to retrieve individually similar chunks, but to construct a context that is relevant, diverse, structurally connected, and sufficient for grounded reasoning. Experiments on MultiHopRAG, UltraDomain, and long-document question-answering settings show that SpectRAG improves broad evidence recovery and achieves competitive generation quality against dense, hybrid, HyDE, LightRAG, and NaiveRAG baselines. The results further indicate that graph-spectral structure is particularly useful for inference-oriented and cross-document retrieval, while highlighting trade-offs in early precision, computational cost, and domain-dependent performance.
Deformable image registration relies on accurate anatomical correspondences. However, explicitly coupling local matching reliability with spatial context selection and deformation update strength remains insufficiently explored in learning-based registration. This challenge is particularly pronounced in homogeneous, weakly textured, and severely deformed regions, where ambiguous matches may induce erroneous deformation updates. We propose ReliMorph, a reliability-routed symmetric framework for 3D deformable registration that couples correspondence assessment with spatial context selection and update strength.Local correlation distributions characterize correspondence ambiguity using shared anatomical features with modality-specific adaptation. These reliability cues guide spatial expert fusion and gate residual velocities during coarse-to-fine refinement. Updated inverse-inconsistency feedback guides routing and gating at finer pyramid levels. To our knowledge, ReliMorph is the first 3D deformable registration framework to jointly condition spatial expert routing and residual velocity gating on correspondence reliability. Extensive experiments demonstrate that ReliMorph consistently outperforms competing methods in anatomical overlap, highlighting its effectiveness across unimodal and multimodal registration tasks.
Machine learning models deployed on edge devices are vulnerable to model stealing attacks, in which an adversary replicates a model using only its predictions. Existing defenses add mechanisms that are costly on resource-constrained hardware. Spiking neural networks (SNNs) offer energy-efficient inference at the edge and have been shown to leak less information about their training data than artificial neural networks (ANNs). This work investigates whether SNNs also resist model stealing and identifies the source of any such resistance. We evaluate four model stealing attacks against SNN and ANN models on three image datasets and introduce a metric that accounts for the stochastic outputs of SNNs. SNNs exhibit greater resistance than ANNs: on average, surrogate models extracted from SNNs agree with their targets 38% less often outside the training distribution, and targeted adversarial examples crafted on these surrogates transfer 54% less often. This resistance, however, arises mainly from the randomness of their input encoding rather than from their spiking computation, and an adversary who repeats each query can largely overcome it. SNNs therefore protect against model stealing without additional defense mechanisms, but this protection does not withstand adaptive adversaries.
The Persona Selection Model (PSM) posits that post-training selects a persona the model learned during pre-training. Work on subliminal learning has shown that models can impart covert signals (like "love of owls") into semantically unrelated outputs like strings of numbers, and that students trained on these outputs inherit those traits. In this work I show that by evaluating the likelihood of such outputs under the base pre-trained model with an appropriate header, a method I call persona-header likelihood evaluation (PHLE), one can decode these subliminal signals and classify between them. For example, PHLE distinguishes number sequences generated by a post-trained model prompted with "You love owls" from those prompted with "You love trains", and distinguishes answers from a model with a secret desire to harm the user, filtered to remove any noticeable misalignment, from those of a model with no system prompt. In both cases prompted classifiers, including the generating model itself, perform far worse. The signal partially persists across model families, with the Qwen base model reading signals generated by Olmo and vice versa. This is evidence for a non-trivial prediction of the PSM, and the method may be practically useful for auditing the hidden objectives of models.
Aviation authorities hold decades of accident reports whose narratives record how investigators reconstructed each accident. Computational work on these archives has mostly sorted reports into fixed taxonomies. In this work, we propose a hybrid framework in which a text-based corpus identifies the leading factors of a failure mode. Physics-based models then quantify how likely that failure is at a stated operating point. Ordered causal chains extracted from 56,202 American accident reports with a local large language model aggregate into a causation knowledge graph. On an evaluation sample of 2,000 reports, the chains recover 90.8% of investigator-flagged causes. Each physical factor whose drivers the record carries receives an occurrence model calibrated to the accident population. A leaky noisy-OR fitted to the corpus combines the factors into a failure mode probability that supports intervention. For engine power loss, the carburetor icing model separates 1,205 icing accidents from other accidents using temperature and dewpoint alone, reaching 0.61 in area under the receiver operating characteristic curve against a density ratio ceiling of 0.64. Routine weather hours give no valid control, because flying concentrates in fair weather. On later years the physics adds 0.03 to that area for occurrence and nothing for severity, so its value lies in interpretability. On four named accidents, abduction gives a probability of necessity of 0.29 to 0.50 for the heat omission, under monotonicity. The Fast Causal Inference algorithm does not license the largest backdoor adjustment on the learned networks.
When a user talks over a full-duplex dialogue model, it must decide whether the user is interrupting (stop) or backchannelling (carry on). Post-training can improve this discrimination or change how readily it stops. In our GRPO runs the second dominates: no configuration lowers both error rates on more than one model, and two released checkpoints discriminate no better than their bases. We trace this partly to how the reward is estimated. The probability of stopping inside a short window is a known function of a few stop probabilities the policy emits, yet GRPO estimates it by sampling. A group of rollouts that all agree gives no gradient, even when all fail; the push on the stop threshold is strongest where the model is unsure, whichever outcome is right; and whole-rollout credit reaches decisions the reward never judges. Confining the credit to the window removes only that leakage. Window likelihood (WL) samples the words but integrates over the stop decisions, so the probability of the correct outcome is computed exactly and its logarithm maximised. On HumDial-FDBench, across four open models, WL gives the best F1 and lowers both error rates by more than their confidence intervals, without talking into silence.