Online learning from a stream of data is a defining feature of intelligence, yet modern machine learning systems often struggle in this setting, especially under distributional shift. To understand its basic properties, we study the relationship between online and offline learning in the context of kernel regression. We derive a closed-form expression for the function learned by online kernel regression, revealing that online kernel regression is equivalent to offline regression with shifted, inaccurate target outputs. Conversely, we show that by compensating for this effective shift in the teaching signal through target correction, online kernel-based learning can provably learn the same predictor as its offline counterpart. We derive both a closed-form expression for this target correction and an iterative form that can be applied sequentially. Applying this framework to image classification tasks on CIFAR-10 and CORe50, we show that online stochastic gradient descent with iteratively corrected targets outperforms learning with the true targets in continual learning settings. This work therefore provides a basic framework for analyzing and improving online learning in non-stationary environments.

Understanding how molecular alterations propagate across biological systems to drive disease remains a central challenge. Although high-throughput profiling enables comprehensive characterization of tumor states, most models neglect structured biological relationships or lack interpretability across scales. Here we present PPI-Net, a hierarchical graph neural network that integrates protein-protein interaction (PPI) networks with pathway-level representations to model disease from molecular interactions to functional processes. Patient-specific molecular profiles are embedded within a shared interaction network from STRING and propagated through a multi-layer Reactome hierarchy using graph attention, enabling aggregation of gene-level signals into higher-order biological programs. Across RNA-seq data from ten cancer types from The Cancer Genome Atlas, PPI-Net achieves robust predictive performance, with balanced accuracy exceeding 90% in multiple cohorts. Comparative analysis on RNA-Seq data from breast cancer demonstrated that PPI-Net's integration of the Reactome hierarchy improved balanced accuracy by 6.7% relative to a PPI-only model, while hierarchical multi-level supervision improved balanced accuracy by 12.3% relative to using only a single top-level prediction head. Applying a multi-omics approach using RNA-seq and methylation data improves model interpretation, recovering canonical oncogenic modules, including TP53-AKT signaling and stress response pathways, while revealing convergence onto coherent programs such as ion signaling and cellular responses to stimuli. These results demonstrate that integrating interaction networks with pathway hierarchies enables accurate prediction while providing mechanistic insight into cancer biology.

Large language models (LLMs) are increasingly deployed as autonomous agents in offensive cybersecurity. In this paper, we reveal an interesting phenomenon: different agents exhibit distinct attack patterns. Specifically, each agent exhibits an attack-selection bias, disproportionately concentrating its efforts on a narrow subset of attack families regardless of prompt variations. To systematically quantify this behavior, we introduce CyBiasBench, a comprehensive 630-session benchmark that evaluates five agents on three targets and four prompt conditions with ten attack families. We identify explicit bias across agents, with different dominant attack families and varying entropy levels in their attack-family allocation distributions. Such bias is better characterized as a trait of the agents, rather than a factor associated with the attack success rate. Furthermore, our experiments reveal a bias momentum effect, where agents resist explicit steering toward attack families that conflict with their bias. This forced distribution shift does not yield measurable improvements in attack performance. To ensure reproducibility and facilitate future research, we release an interactive result dashboard at https://trustworthyai.co.kr/CyBiasBench/ and a reproducibility artifact with aggregated session-level statistics and full evaluation scripts at https://github.com/Harry24k/CyBiasBench.

Training multimodal large language models has long been limited by the scarcity of high-quality paired multimodal data. Recent studies show that the shared representation space of pretrained multimodal contrastive models can serve as a bridge, enabling models to perform multimodal training with unimodal data. However, the key premise of this paradigm remains insufficiently understood: can representations from different modalities be reliably interchanged? The core obstacle lies in the persistent Modality Gap in the shared space. In this work, we revisit the geometric nature of the modality gap. We find that modality representations already share compatible dominant semantic geometry. What truly hinders modality interchangeability is not a simple global shift, but an anisotropic residual structure concentrated along a small number of dominant directions. Based on this finding, we further propose the principle of anisotropic modality gap alignment: effective modality alignment should align with the target-modality distribution while preserving the semantic structure of the source modality. Guided by this principle, we propose an anisotropic geometric correction framework, AnisoAlign, for unpaired modality alignment. This framework leverages the internal geometric prior of the target modality and performs bounded correction on source-modality representations, thereby constructing substitute representations in the target modality. Experiments confirm its benefits in both geometric diagnostics and text-only MLLM training. Overall, this work recasts the modality gap from an empirical observation into a correctable, structured geometric phenomenon and provides a new representation alignment perspective for training multimodal models with unimodal data.

Advanced persistent threat (APT) attacks remain difficult to detect due to their stealth, adaptability, and use of legitimate system components. Provenance-based intrusion detection systems (PIDS) offer a promising defense by capturing detailed relationships between system components and actions. However, current PIDS rely on predefined or subset-determined thresholds, which limit detection stability and the ability to detect any anomalous behavior in general. Furthermore, related work often neglects the role of process executables, which describe system activity by interacting through a process with files, network components, and other processes. We introduce GRASP, a PIDS based on masked self-supervised classification. GRASP masks the executable information of processes and learns to infer it from their two-hop provenance graph neighborhood, marking misclassified processes as anomalies. It captures behavior patterns for the learned executables without thresholding, making it robust against interference and unknown activities. Evaluations on the DARPA TC and OpTC datasets demonstrate that GRASP consistently detects anomalous behavior, including known attack-related activities, outperforming existing systems. Our PIDS identifies all documented attacks on datasets where the behavior of executables is learnable. In addition, compared to existing systems, GRASP uncovers potentially malicious anomalous behavior not labeled as an attack in the documentation.

Optical neural networks are emerging as powerful machine learning and information processing tools because of their potential advantages in speed and energy efficiency. The training methods of these physical models, however, remain underexplored compared to their digital counterparts and are leading to suboptimal performance. This paper reports a pre-training-driven approach that leads to snapshot image denoising with substantially improved quality. We demonstrated effective free-space optical denoising by a diffractive network optimized by a two-step process including (1) pre-training using a massive dataset of 3.45 million diverse but simple images and (2) fine-tuning with the corresponding task-specific datasets. Compared to conventional Fourier-domain filtering and directly trained diffractive networks, such a transfer learning process exhibited prominent advantages for denoising images degraded by severe noise, peak signal-to-noise ratio (PSNR) below 8 dB, while preserving fine image features and improving the PSNR to above 18 dB. Importantly, the same pre-trained optical network could be consistently fine-tuned to process degraded images from highly diverse styles ranging from handwritten digits (MNIST) and chest X-rays (ChestMNIST) to CIFAR-10 images and human faces (CelebA). We further demonstrated the critical role of our optical denoisers in vision-based applications, including face detection, plate recognition, and localization of UAVs in noisy conditions.

We investigate interactive trajectory planning subject to uncertainty in the decisions of surrounding agents. To control the ego-agent, we aim to first learn the decision distribution and solve a Stochastic Model Predictive Control (SMPC) problem. To account for errors in the learned distribution, we show that it is possible to utilize Probably Approximately Correct (PAC) learning in combination with Distributionally Robust (DR) optimization to obtain a solution which accounts for the errors induced by the learning model. The results indicate that our PAC learning-based DR-MPC framework provides a method to interpolate between a robust MPC and an omnipotent SMPC, based on the available number of samples.

We present an SMT-based active learning algorithm for nondeterministic weighted automata (WFAs) as a practical and robust alternative to Hankel/L*-style methods. Our algorithm is parametric in a given semiring and, if it terminates, guaranteed to produce minimal WFAs. We prove partial correctness and provide a sufficient termination condition, which in particular implies termination for all finite semirings. Our extensive experimental evaluation shows that our algorithm is capable of learning numerous minimal WFAs over both finite and infinite semirings, vastly outperforms a naive baseline, and is competitive with a state-of-the-art algorithm while producing significantly smaller automata and requiring less interaction with the teacher.

Early 2025 we ran a series of vibe coding challenges across four different student cohorts. The cohorts included 54 ICT students, 24 digital marketing students, and 7 journalism students at Fontys University of Applied Sciences (Netherlands), and 22 BA Communication students at North-West University (South Africa). From the student reflections, five major patterns emerged. Students reported that AI tools shifted their focus from syntax to higher-order thinking; they also described a skill shift from memorizing to evaluating; they viewed AI proficiency as career-essential; they framed their relationship with AI as partnership rather than replacement; and finally non-technical students showed the strongest appreciation for the accessibility these tools provide. This practitioner report documents what we observed during the classroom experiments, we reflect on how the landscape has shifted in the year since, and shares practical lessons for educators considering similar experiments. We present the observations as what they are: patterns from practice, not proven conclusions, in the beleif that sharing early stage experiences contributes to the overall field of AI and education.

High-dimensional count data arise in applications such as single-cell RNA sequencing and neural spike trains, where mapping between distributions across successive batches or time points form critical components of data analysis. The recent success of diffusion- and flow-based deep generative models for images, video, and text motivates extending these ideas to count-valued settings, but many existing methods either treat each count as a categorical state or transform counts into a continuous space, neither of which is natural or efficient when the count range is large. We propose count-FM, a flow-matching framework for count data based on a continuous-time birth-death process with local unit jumps. Count-FM learns marginal transitions efficiently in count space through simulation-free training of conditional transition rates, allowing transport between arbitrary count-distributed source and target populations. In simulation, count-FM achieves better sample quality than representative baselines while using substantially fewer parameters. We further apply count-FM to scRNA-seq and neural spike-train data for unconditional generation, transport, and conditional generation. Across these tasks, count-FM yields improved sample quality, greater modeling efficiency, and interpretable transport paths.

Physics-informed operator learning is an attractive candidate for surrogate modeling of microstructures, especially in multiscale finite-element simulations. Its practical use, however, is often limited by the high cost of loss evaluation. We address this bottleneck by combining the Equilibrium Neural Operator (EquiNO) with the QR-based discrete empirical interpolation method (Q-DEIM). EquiNO learns only the modal coefficients of reduced displacement-fluctuation and first Piola-Kirchhoff stress representations built from periodic and divergence-free bases, thereby enforcing periodicity and mechanical equilibrium by construction. Q-DEIM then identifies a small set of spatial points through a column-pivoted QR factorization of the stress basis and restricts constitutive evaluations during training to these points alone. This makes full-batch second-order optimization practical for three-dimensional representative volume elements (RVEs). Homogenized first Piola-Kirchhoff stresses are recovered directly from the offline-averaged reduced stress modes, without the need to reconstruct the full stress field at inference time. We validate the framework on two three-dimensional finite-strain hyperelastic RVEs. Q-DEIM reduces the per-step training cost by roughly three orders of magnitude relative to full-field loss evaluation, while reduced homogenization achieves speed-up factors of order $10^3$ to $10^4$ over direct full-field computations. Despite relying on only a small number of offline snapshot loading paths for basis construction, the method accurately interpolates and extrapolates both microscopic stress fields and homogenized stresses, with prediction quality improving systematically as more snapshots are added.

Large language models (LLMs) are known to generate plausible but false information across a wide range of contexts, yet the real-world magnitude and consequences of this hallucination problem remain poorly understood. Here we leverage a uniquely verifiable object - scientific citations - to audit 111 million references across 2.5 million papers in arXiv, bioRxiv, SSRN, and PubMed Central. We find a sharp rise in non-existent references following widespread LLM adoption, with a conservative estimate of 146,932 hallucinated citations in 2025 alone. These errors are diffusely embedded across many papers but especially pronounced in fields with rapid AI uptake, in manuscripts with linguistic signatures of AI-assisted writing, and among small and early-career author teams. At the same time, hallucinated references disproportionately assign credit to already prominent and male scholars, suggesting that LLM-generated errors may reinforce existing inequities in scientific recognition. Preprint moderation and journal publication processes capture only a fraction of these errors, suggesting that the spread of hallucinated content has outpaced existing safeguards. Together, these findings demonstrate that LLM hallucinations are infiltrating knowledge production at scale, threatening both the reliability and equity of future scientific discovery as human and AI systems draw on the existing literature.

We give a novel logical characterization of encoder-decoder transformers, the foundational architecture for LLMs that also sees use in various settings that benefit from cross-attention. We study such transformers over text in the practical setting of floating-point numbers and soft-attention, characterizing them with a new temporal logic. This logic extends propositional logic with a counting global modality over the encoder input and a past modality over the decoder input. We also give an additional characterization of such transformers via a type of distributed automata, and show that our results are not limited to the specific choices in the architecture and can account for changes in, e.g., masking. Finally, we discuss encoder-decoder transformers in the autoregressive setting.

Single-channel speaker distance estimation has recently achieved centimeter-level accuracy in simulated environments, yet it remains unclear which components of the room impulse response (RIR) the model exploits and how performance depends on the recording conditions. In this work, we decompose simulated RIRs into four variants (full, direct-only, no-late, and no-early) using the mixing time estimated from the echo density function as the boundary between early reflections and late reverberation. We define four calibration scenarios, from fully calibrated (synchronised capture, known source level) to fully uncalibrated (arbitrary onset, unknown level), and evaluate all combinations on a matched dataset. Results show that without time calibration, mean absolute error (MAE) increases to $1.29$ m and the model extracts reverberation-based cues, with early reflections emerging as the most informative component. Further analysis against DRR, $C_{50}$, and $T_{60}$ confirms that estimation accuracy improves with stronger early energy and degrades in highly reverberant environments. When time calibration is available, the model achieves a MAE of $0.14$ m by extracting the propagation delay alone, regardless of the RIR content.

Tourism is a high-stakes setting for conversational recommender systems (CRS): a plausible-sounding suggestion can waste real money and trip time once a traveler acts on it. Existing CRS benchmarks primarily evaluate systems with a single Recall@k score over entity mentions, and tourism-specific resources add spatial or knowledge-graph context, yet none of them couple multi-turn recommendation with verbatim review-span evidence and rejection recovery. This leaves an evaluation gap for tourism recommendation that is simultaneously trustworthy, verifiable, and adaptive: recommend the right point of interest (POI) for multi-aspect preferences (such as cuisine, price, atmosphere, walking distance), justify each suggestion with verifiable evidence from prior visitors so the traveler can act without trial and error, and recover when the first recommendation is rejected mid-dialogue. We introduce TRACE, where each item is a multi-turn tourism recommendation dialogue with review-span citations and explicit rejection turns: 10,000 dialogues over 2,400 Yelp POIs and 34,208 reviews across eight U.S. cities, paired with 14 retrieval, planning, and LLM baselines, along with 25 metrics organized under Accuracy, Grounding, and Recovery. Across these baselines, TRACE reveals the Three-Competency Gap: LLM Zero-Shot leads in closed-set Recall@1 and rejection recovery but cites less densely than retrievers; non-LLM retrievers achieve surface-verbatim grounding but with low accuracy; Multi-Review Synthesis fails at recovery. The Grounding Score agrees with human citation precision (Spearman rho=+0.80, p<10^-20), and paired t-tests reproduce the per-baseline ranking (p<0.01 on the dominant contrasts). TRACE reframes accountable tourism recommendation as a joint target (right POI, verifiable evidence, adaptive repair) rather than a single-axis leaderboard.

Regulatory audits of AI systems increasingly rely on differential privacy (DP) to protect training data and model internals. We study audit design when the audited developer can strategically respond to the privacy-constrained audit interface. We formalize privacy-constrained auditing as a bilevel Stackelberg game, in which an auditor commits to a query policy and DP budget allocation across harm dimensions, and a strategic developer reallocates mitigation efforts in response. We introduce the welfare-weighted under-detection gap $B_w$, the welfare-weighted true residual harm the audit fails to detect at the developer's strategic best response, and prove that naive DP auditing (uniform or harm-proportional allocation) induces a strictly larger $B_w$ than any non-strategic mitigation baseline whenever effective detectability is heterogeneous, the welfare weights are not comonotone with detectability, and the developer's optimum is interior. We characterize the optimal auditor allocation as a four-factor balance of welfare weight, audit miss-probability, detectability elasticity, and mitigation-cost curvature, and provide a single-level reformulation of the bilevel problem via the developer's KKT system. We propose Strategic Private Audit Design (SPAD), a projected-gradient algorithm with hypergradients computed through the developer's best response.

Eliciting truthful reports from autonomous agents is a core problem in scalable AI oversight: a principal scores the agent's report using a strictly proper scoring rule, but the agent also benefits from the report through a non-accuracy channel (approval for autonomous action, allocation share, downstream control). The same structure appears in classical mechanism-design settings such as marketplace operation. Our main result is an endogeneity: the principal's optimal oversight necessarily uses a non-affine approval function to screen types, yet any non-affine approval makes truthful reporting suboptimal under the combined objective whenever deviation is undetectable. The principal cannot avoid the perturbation that undermines calibration. This impossibility holds for all strictly proper scoring rules, with a closed-form perturbation formula. A constructive escape exists: a step-function approval threshold achieves first-best screening for every strictly proper scoring rule, because the agent's binary inflate-or-not choice creates a type-space threshold regardless of the generator's curvature. Under the Brier score specifically, the type-independent inflation cost yields a welfare equivalence between second-best and first-best; we prove this equivalence is unique to Brier (the welfare gap under smooth $C^1$ oversight is bounded below by $Ω(\text{Var}(1/G'') (γ/β)^2)$ for every non-Brier rule). Two instances develop the framework: AI agent oversight (the lead motivating setting) and marketplace operation (a parallel mechanism-design domain). The message for AI alignment is direct: smooth scoring-based oversight cannot elicit truthful reports from a strategic agent; sharp thresholds are the calibration-preserving design.

Estimating counterfactual distributions under interventions is central to treatment risk assessment and counterfactual generation tasks. Existing approaches model the counterfactual distribution as a standalone generative target, without exploiting its relationship to the observational data. In this work, we show that under standard assumptions, observational and counterfactual outcome distributions are tightly linked: they have identical support and tail behavior, remain statistically close under weak confounding, and share any features of high-dimensional outcomes which are invariant to confounders. These properties motivate learning counterfactual distributions not from scratch, but via a deconfounding flow from the observational distribution. We formulate this problem via flow-matching and derive a semiparametrically efficient estimator based on a novel efficient influence function correction. We subsequently extend our estimator to target minimal-energy flows in high-dimensions, which we show can be especially simple targets between observational and counterfactual distributions. In experiments, deconfounding flows outperform existing debiased counterfactual distribution estimators, while also mitigating known failure modes of flow-based methods.

Data valuation methods allocate payments and audit training data's contribution to machine-learning pipelines; however, they often assume passive contributors. In reality, contributors can split datasets across pseudonymous identities, duplicate high-value examples, create near-duplicates, or launder synthetic variants to inflate their share. We formalize this as false-name manipulation in ML data attribution. Our main construction is the quotient semivalue mechanism: compute Shapley-, Banzhaf-, or Beta-style values over evidence-backed attribution clusters instead of raw identities, using a canonical-representative operator to absorb within-cluster duplication. We prove an impossibility: on a fixed monotone data-value game, exact Shapley-fair attribution over reported identities is incompatible with unrestricted false-name-proofness, even on binary-valued instances, and characterize the split-gain of a general semivalue on a unanimity counter-example. The mechanism is exactly false-name-proof under two structural conditions: false-name-neutral within-cluster allocation and quotient-stable manipulations. Under imperfect provenance, when these conditions hold approximately, manipulation gain and fairness loss are bounded by three measurable quantities: escaped-cluster mass, value-estimation error, and clustering distance. We instantiate the mechanisms in DataMarket-Gym, a benchmark for attribution under strategic provider attacks. On synthetic classification tasks, quotient semivalues with example-level evidence reduce manipulation gain on duplicate and near-duplicate Sybil attacks from $1.74$ under baseline Shapley to $0.96$, near the honest level. The cosine-threshold and (false-merge, false-split) rate sweeps trace the corresponding fairness--Sybil frontier.

Large Language Models often improve accuracy on reasoning tasks by sampling multiple Chain-of-Thought (CoT) traces and aggregating them with majority voting (MV), a test-time technique called self-consistency. When we truncate a CoT partway through and regenerate the remainder, we observe that traces with correct answers reproduce their original answer more often than traces with wrong answers. We use this difference as a reliability signal, prefix consistency, that weights each candidate answer by how often it reappears under regeneration. It requires no access to token log-probabilities or self-rating prompts. Across five reasoning models and four math and science benchmarks, prefix consistency is the best correctness predictor in most settings, and reweighting votes by it reaches Standard MV plateau accuracy at up to 21x fewer tokens (median 4.6x). Our code is available at https://github.com/naoto-iwase/prefix-consistency.

We consider a first order stochastic optimization framework where, at each iteration, $K$ independent identically distributed (i.i.d.) data point samples are drawn, based on which stochastic gradients can be queried. We allow gradient noise to be heavy-tailed, with possibly infinite variances. For the considered heavy-tailed setting, many algorithmic variants have recently been proposed based on gradient clipping or other nonlinear operators (e.g., normalization) applied over noisy gradients. In this paper, we take an alternative approach and propose a novel stochastic first order method dubbed Robust Stochastic Gradient Descent with medoid mini-batch gradient sampling, R-SGD-Mini for short. The core idea of R-SGD-Mini is to split the $K$-sized data batch into $M$ distinct data chunks, form for each chunk the stochastic gradient, and update the solution estimate with respect to the stochastic gradient direction of the chunk that is medoid of gradients of all data-chunks. Under a general class of symmetric heavy-tailed gradient noises and a standard non-convex setting, we establish explicit bounds on the expected time-averaged squared gradient norm. More precisely, we show that the latter quantity converges at rate $\mathcal{O}(T^{-1})$ to a small neighborhood of zero; we explicitly characterize this neighborhood in terms of noise and algorithm's parameters. Moreover, if the time horizon is known in advance, we establish the rate of $\mathcal{O}(T^{-\frac{1}{2}}).$ Furthermore, when clipping is incorporated, we obtain convergence guaranties in the high-probability sense and recover the same rate. Experimental results indicate that R-SGD-Mini and its clipped variant consistently perform favorably compared to SGD, clipped SGD and Median-of-Means based methods.

Detecting stealthy malicious communications from flow logs under benign-only training remains a critical challenge in network security. Malicious communications often camouflage as normal traffic like standard HTTPS flows. Conventional intrusion detectors rely strictly on known labeled attacks. Alternatively, they score flows completely independently. These approaches fail against sparse and context-dependent suspicious activity. To capture this essential context, graph anomaly detectors have been introduced to add valuable relational information to the analysis. However, existing methods fail to test the structural consistency of specific communication edges. To overcome these fundamental limitations, we present GESR, a novel graph-based framework for detecting suspicious communications and anomalous hosts under a benign-only training setting. GESR models complex network activity as attributed communication graphs. It cleverly reconstructs edge semantics entirely from local structural context rather than isolated features. This non-intuitive design forces the framework to predict expected communication patterns from neighborhood topologies. Attackers cannot easily manipulate this deep structural dependency. The model then converts the resulting structural inconsistencies into host-level anomaly scores. It utilizes robust Median Absolute Deviation (MAD) calibration for this final step. We evaluate GESR extensively on CTU-13 and CICIDS2017 datasets. These evaluations strictly impose tight false-positive operating constraints. On CICIDS2017, GESR achieves an outstanding ROC-AUC of 0.9753. It also yields a high TPR of 0.8569 at a strict 5% FPR threshold. GESR consistently outperforms existing methods across both evaluated benchmarks. The results prove that structure-conditioned edge reconstruction is a credible direction for practical intrusion detection.

In the envisioned future dense urban airspace, multiple companies will operate heterogeneous fleets of small unmanned aerial systems (sUASs), where each fleet includes several homogeneous aircraft with identical policies and configurations, e.g., equipage, sensing, and communication ranges, making tactical deconfliction highly complex for the aircraft. This paper aims to address two core questions: (1) Can tactical deconfliction policies converge or reach an equilibrium to ensure a conflict-free airspace when companies operate heterogeneous fleets of homogeneous aircraft? (2) If so, will the converged policies discriminate against companies operating sUASs with weaker configurations? We investigate a multi-agent reinforcement learning paradigm in which homogeneous aircraft within heterogeneous fleets operate concurrently to perform package delivery missions over Dallas, Texas, USA. An attention-enhanced Proximal Policy Optimization-based Advantage Actor-Critic (PPOA2C) framework is employed to resolve intra- and inter-fleet conflicts, with each fleet independently training its own policy while preserving privacy. Experimental results show that two fleets with distinct, shared PPOA2C policies can reach an equilibrium to maintain safe separation. While two PPOA2C policies outperform two strong rule-based baselines in terms of conflict resolution, a PPOA2C policy exhibits safer interaction with a rule-based policy, indicating adaptive capabilities of PPOA2C policies. Furthermore, we conducted extensive policy-configuration evaluations, which reveal that equilibria between similar policy types tend to favor fleets with stronger configurations. Even under similar configurations but different policy types, the equilibrium favors one of the heterogeneous policies, underscoring the need for fairness-aware conflict management in heterogeneous sUAS operations.

This report documents the preparedness assessment of Code World Model (CWM), a model for code generation and reasoning about code from Meta. We conducted pre-release testing across domains identified in our Frontier AI Framework as potentially presenting catastrophic risks, and also evaluated the model's misaligned propensities. Our assessment found that CWM does not pose additional frontier risks beyond those present in the current AI ecosystem. We therefore release it as an open-weight model.

Reward models (RMs) have become an indispensable fixture of the language model (LM) post-training playbook, enabling policy alignment and test-time scaling. Research on the application of RMs in code generation, however, has been comparatively sparse, with existing work largely focusing on execution feedback. This choice constrains post-training to optimizing functional correctness over self-contained executable code. In this work, we examine the training and evaluation of multilingual, multi-criteria code RMs. To this end, we first compile Themis-CodeRewardBench, a benchmark to evaluate code RMs across five preference dimensions (i.e., criteria) and eight programming languages, on which we profile 50+ code, math, and general-purpose RMs. Observing the limited proficiency of current RMs beyond scoring for functional correctness, we develop Themis-CodePreference, the largest open-source collection of code preferences to date (more than 350k preference pairs), and use it to train Themis-RM, a suite of multilingual code reward models for flexible multi-criteria scoring, ranging in size from 600M to 32B parameters. Our experiments and ablations demonstrate positive scaling trends, strong cross-lingual transfer when training on diverse preferences, and the importance of multi-criteria training for reliable code reward modeling.

We present an empirical case study of cinema SDR-to-HDR mapping using ASC StEM2, a rare common-source dataset containing EXR scene-referred images and matched SDR/HDR cinema release masters from the same ACES-based mastering workflow. Based on pixel-wise statistics over all 18,580 frames of the test film, we construct a three-domain comparison involving EXR source data, SDR release masters, and HDR release masters to characterize their luminance and color structural relationships within this controlled workflow. In the luminance dimension, SDR and HDR masters exhibit a highly stable global monotonic correspondence, with geometric structure remaining largely consistent overall; sparse and structured deviations appear in self-luminous highlights and specific material regions. In the color dimension, the two masters remain largely consistent in hue, with saturation exhibiting a redistribution pattern of shadow suppression, midtone expansion, and highlight convergence. Using EXR as a scene-referred anchor, we further define a pixel-level decision map that operationally separates EXR-closer recovery regions from content-adaptive adjustment regions. Under this operational definition, 82.4% of sampled image regions are classified as EXR-closer recovery, while the remainder require localized adaptive adjustment. Rather than claiming a universal law for all cinema mastering pipelines, the study provides an interpretable quantitative baseline for structure-aware SDR-to-HDR analysis and for designing learning-based models under shared-source mastering conditions.

Gradient normalization stabilizes deep-learning optimization, and spectral normalizations are especially natural for matrix-shaped parameter blocks; Muon is the motivating example. We study an idealized deterministic, continuous-time, vanishing-momentum version of this idea in the mean-field regime, where wide models are represented by probability measures on parameter space. Starting from normalized matrix flows, we introduce Spectral Wasserstein distances indexed by norms $γ$ on positive semidefinite matrices: the trace norm gives classical $W_2$, the operator norm gives the Muon geometry, and Schatten norms interpolate between them. We develop the static Kantorovich formulation, a max-min robust-cost representation, Gaussian reductions extending the Bures formula, and for monotone norms, prove equivalence with a Benamou--Brenier formulation. This yields a gradient-flow interpretation of the mean-field normalized training dynamics. We illustrate these findings by numerical experiments on MMD flows, Gaussian reductions, two-layer ReLU models, and shallow attention.

Artificial intelligence is transforming mathematics at a speed and scale that demand active engagement from the mathematical community. We examine five areas where this transformation is particularly pressing: values, practice, teaching, technology, and ethics. We offer recommendations on safeguarding our intellectual autonomy, rethinking our practice, broadening curricula, building academically oriented infrastructure, and developing shared ethical principles - with the aim of ensuring that the future of mathematics is shaped by the community itself.

There is an extensive literature that studies how to find optimal policies in resource allocation problems, taking the underlying design parameters that define the allocation, such as what data is collected, how many people can be served, and quality of service as fixed constraints. Yet, from a planner's perspective, these design parameters are themselves optimization variables that are just as important in determining overall welfare as selecting the optimal targeting rule for a given set of constraints. This realization motivates a rich set of meta-design questions exploring how planners should make principled decisions about investments in prediction, capacity constraints, and treatment quality, all of which lie upstream of classical policy optimization. Building on initial theoretical work in this space, our paper has three main contributions. First, we formally define the broad meta-design space of resource allocation problems. Second, we develop empirical tools that enable practitioners to reliably navigate it. Third, we demonstrate the framework in two real-world case studies on German employment services and targeted cash transfer programs in Ethiopia.

Setting the learning rate (LR) for a deep learning model is a critical part of successful training. Choosing LRs is often done empirically with trial and error. In this work, we explore a solvable model of optimal LR schedules for a powerlaw random feature model trained with stochastic gradient descent (SGD). We consider the optimal schedule $η_T^\star(t)$ where $t$ is the current iterate and $T$ is the training horizon. This schedule is computed both as a numerical optimization problem and also analytically using optimal control theory. Our analysis reveals two regimes which we term the easy phase and hard phase. In the easy phase the optimal schedule is a polynomial decay $η_T^\star(t) \simeq T^{-ξ} (1-t/T)^δ$ where $ξ$ and $δ$ depend on the properties of the features and task. In the hard phase, the optimal schedule resembles warmup-stable-decay with constant initial LR and annealing performed over a vanishing fraction of training steps. We investigate joint optimization of LR and batch size and find batch ramps can improve the wall-clock time in the easy phase. Beyond SGD, we derive optimal schedules for momentum parameter $β(t)$ and show that it improves the loss-scaling exponent in the hard phase. We compare our optimal schedule to various benchmarks including (1) optimal constant learning rates $η_T(t) \sim T^{-ξ}$ (2) optimal power laws $η_T(t) \sim T^{-ξ} t^{-χ}$, finding that our schedule achieves better rates than either of these. Our theory suggests that LR transfer across training horizon depends on the structure of the model and task. For ResNet image classification on CIFAR-5M, the learning curves exhibit hard-phase behavior where optimal base LRs are constant under sufficient annealing. GPT-2 style transformers trained in language modeling exhibit easy-phase behavior where optimal LRs shift even under annealing.