Papers

We examine the interplay between ordinal, preference-based solution concepts in games and the long-run behavior of game dynamics, asking in particular to what extent the combinatorial data of a game – its preference graph – determine the outcomes of no-regret learning dynamics – such as follow-the-regularized-leader (FTRL). In one direction, we show that the skeleton of every dynamically stable set (i.e. the set of pure profiles it contains) must also be preferentially stable, that is, it must be closed under profitable deviations. We then ask the converse question: when do preferences determine the long-run behavior of the players' learning dynamics? We begin by showing that preferences characterize asymptotic stability in the case of subgames – i.e. subsets of pure profiles obtained by restricting players' action sets. Beyond this case however, the equivalence between dynamic and preferential stability collapses: concretely, we construct a three-player game with a preferentially stable set whose span is dynamically unstable, showing in this way that preferences do not suffice as a criterion of dynamic stability. We then bridge this gap via the notion of resilience under aggregate deviations, an easy-to-check payoff-based condition that guarantees asymptotic stability of arbitrary spans of pure strategies.

Transition path sampling (TPS) aims to efficiently generate rare molecular transition trajectories between metastable states and is essential for understanding biomolecular mechanisms. Beyond traditional molecular dynamics (MD)-based sampling, machine learning has become central to state-of-the-art TPS. One major class of methods learns control forces during explicit MD rollouts. By preserving the underlying molecular dynamics, these methods tend to produce more physically plausible trajectories than endpoint-conditioned generators that construct paths directly. However, rollout-based control methods have been reported to exhibit unstable and strongly seed-dependent performance. We recast rollout-based control as learning a path-space proposal distribution and investigate stochasticity placement as a design choice for improving exploration and optimization robustness. We develop two stochastic policies: FS-TPS, which directly parameterizes a state-dependent Gaussian distribution over the control policy output, and LaS-TPS, which samples a compact latent control variable and decodes it into structured, cross-atom-correlated force variation. We conduct extensive multi-seed experiments on three biomolecular systems of increasing size: alanine dipeptide, chignolin, and BBL, a fast-folding protein. Stochastic policies consistently improve transition success and path quality over deterministic-policy baselines while substantially reducing sensitivity to random initialization.

This paper presents an event-driven learning and benchmarking framework for the Dynamic Multi-Depot Vehicle Routing Problem with progressively revealed requests and evolving vehicle states. Masked MLP and Transformer policies are trained through behavior cloning and proximal policy optimization. Deterministic feasibility masking prevents invalid vehicle–request assignments, while fixed-prefix/flexible-suffix route commitments protect completed, active, and near-term decisions and separately measure vehicle reassignment and resequencing. The learned policies are compared with dynamic insertion heuristics and time-limited rolling-horizon optimization. In a 20-scenario policy benchmark, all methods completed every request without invalid actions, but nearest feasible achieved the lowest mean objective and outperformed the learned policies in routing quality, waiting time, stability, makespan, and runtime. Across five independent training runs, PPO had little average effect on the MLP and improved the Transformer on average, although with greater seed variability. Under the common protocol, nearest feasible achieved the lowest combined objective and route disruption, whereas rolling horizon achieved the lowest waiting times and makespan at substantially higher computational cost. The learned policies retained millisecond-level decisions and transferred to instances with up to 80 requests without retraining, but did not outperform the strongest heuristic. No single method was best across routing efficiency, service responsiveness, stability, and online computation.

Computational approaches to drug discovery involve multiple sub-problems, and among them, drug-target binding affinity prediction plays an important role. Despite recent advances, accurately predicting binding affinity remains an open research area. The major objective of our paper is to perform a comprehensive review and comparative analysis of recent machine learning methods for drug-target binding affinity prediction, with a focus on identifying strengths, limitations, and research gaps. We review representative recent deep learning approaches that use common benchmark datasets and evaluation metrics, covering a range of neural network architectures and representation strategies. In addition, we analyze seven widely used benchmark datasets and commonly adopted evaluation metrics for drug-target binding affinity prediction. Our analysis indicates that although many methods report strong performance on standard benchmarks, their effectiveness is often influenced by dataset bias and limited evaluation settings. Furthermore, most methods exhibit reduced performance in cold-start scenarios, highlighting challenges in generalization. We identify several limitations of current approaches, including dataset imbalance, the lack of standardized evaluation, limited real-world applicability, and challenges in cold-start scenarios. We also discuss future research directions, including better dataset design, more robust evaluation methods, improved handling of cold-start problems, and the integration of multimodal representations.

Machine learning (ML) has become an indispensable part of modern engineering design workflows. A crucial step in training an ML model is the selection of the loss function which can be systematically formulated via various techniques such as maximum likelihood estimation (MLE) and cross-validation . While MLE is one of the most popular, effective, and intuitive mechanisms for training ML models, it is brittle: if the assumptions underpinning it are not met, the trained ML model may generalize poorly. This brittleness affects even Gaussian processes (GPs) which are widely used in engineering design and are often (incorrectly) presumed to be very robust to overfitting. In this paper, we fundamentally evaluate the brittleness of MLE in the context of training GPs for probabilistic regression or classification tasks. We compare theoretically grounded metrics against MLE and propose practical solutions. Our extensive studies demonstrate the effectiveness of our solutions in downstream design tasks such as Bayesian optimization and provide a blueprint for practitioners to build accurate and robust GPs that can even outperform tabular foundation models in terms of prediction accuracy, uncertainty quantification, and inference cost. Our contributions are publicly available via GitHub at https://github.com/Bostanabad-Research-Group/GP-vs-TabPFN-vs-GPyTorch.

Macro placement significantly affects a chip's post-route performance, power, and area (PPA). Most placement methods optimize half-perimeter wirelength (HPWL) as the primary objective. However, recent benchmarking shows a near-zero correlation between HPWL and post-route timing metrics such as the worst negative slack (WNS) and total negative slack (TNS). As a result, all six evaluated artificial intelligence (AI) placers degraded PPA relative to the hierarchical baseline. Recent efforts have tried to train cross-stage predictors to close this gap. However, existing methods focus on macro-only representations and use pre-route metrics as training labels. A label fidelity study of ten circuits at four design flow stages reveals that HPWL and pre-route timing poorly reflect final post-route timing rankings. In contrast, post-global-routing achieves the best balance between final timing fidelity and label generation cost-effectiveness. Based on this finding, PPAPlace is a timing-driven differentiable surrogate predicting post-route PPA from macro and standard-cell placements. The surrogate is a dual-stream predictor that combines graph attention over the chip netlist with spatial convolution over the placement grid. It is trained on post-global-routing labels. The predicted WNS and TNS gradients flow end-to-end back to cell coordinates. PPAPlace exploits these gradients in two ways: as a co-objective injected into an analytical placer's optimization loop (PPAPlace-CoOpt), and as a post-placement refinement step that adjusts macro positions via projected gradient descent (PPAPlace-Refine). On five ChiPBench test circuits excluded from training, PPAPlace improves average WNS and TNS by 22% and 51% over the hierarchical baseline while preserving power and routability, using the same predictor without test-circuit retraining. Code is available at https://github.com/ValleyC/PPAPlace.

AI agents increasingly act on their users' behalf, handling tasks such as scheduling meetings, comparing offers, and haggling over prices. These principal-driven tasks routinely place the agent across from a counterpart (another user's agent, a seller, a recruiter) whose goals may conflict with its principal's. Yet the dispositions that make an assistant pleasant can make it a poor delegate: a friendly, helpful frontier model may disclose its principal's private information unprompted and concede at the first sign of resistance. We present SocialRL, a general recipe that trains social reasoning directly, and apply it to a 4B model across six domains: Deal-or-No-Deal, CaSiNo, Craigslist, Job Interview, Calendar, and Marketplace. Every domain is trained in-domain under the same recipe, and every policy is evaluated on all six. We find that (1) in-domain training reaches the frontier: on held-out scenarios the 4B matches or exceeds the GPT-5 family per domain, closing 73-122% of the baseline-to-frontier gap on the negotiation games, with 78% of buyer openings anchoring below target versus 3% untrained; (2) cross-domain transfer follows game structure: structurally paired games lift each other, a broad multi-issue donor lifts nearly all domains, and structurally isolated games transfer nothing; (3) guided by this transfer structure, two strategies, cascade RL and multi-teacher on-policy distillation (OPD), consolidate the per-domain specialists into a single unified 4B that reaches 0.627 average utility across all six environments, matching or exceeding GPT-4.1 (0.625), GPT-5.1 (0.619), and GPT-5.2 (0.613); (4) an explicit theory-of-mind scaffold helps only through training: distilling the ToM trace, rather than actions alone, lifts utility on every environment and generalizes better across them, and of the two ToM skills, only next-action prediction predicts negotiation outcomes.

Neural networks are increasingly deployed in high-stakes applications with growing privacy leakage concerns. We show that this privacy leakage can occur even in the absence of representation imbalances that lead to traditional dataset biases. This poses significant privacy risks when deploying models that process sensitive attributes. In this context, we propose CutClean, a privacy-aware pruning method that allows to reduce privacy information flow through the network, while increasing its sparsity. Our approach employs auxiliary linear privacy heads placed at each network's block to quantify information leakage, and further applies increasing levels of sparsity to remove the private attribute leakage, measured in terms of the accuracy of the privacy head attached to the last block. Experiments on synthetic and real-world datasets demonstrate that our approach effectively minimizes private information flow while achieving high sparsity rates and preserving classification target accuracy.

Which reasoning behaviors are associated with correct answers in reasoning models, and does reasoning-oriented training amplify those behaviors? This distinction is important because reasoning-oriented training can make traces look more deliberative without amplifying the behaviors most tied to model correctness. We quantify this mismatch with Behavioral Lift, a metric that measures how much correctness changes when a behavior is present versus absent in a model's reasoning trace. Across 15 models and 6 benchmarks spanning text-only and vision-language reasoning, we annotate 15,282 traces with a taxonomy whose core behaviors are defined for both LLM and VLM traces. We find evidence for an Amplification-Lift Gap, in which thinking models strongly amplify self-correction, hypothesis testing, and uncertainty acknowledgment, while the highest-lift behaviors are confidence calibration, knowledge alignment, and self-awareness. Confidence calibration is among the strongest positive signals of correctness in both modalities, yet is barely amplified; uncertainty acknowledgment is amplified by 3–7$\times$, yet is weakly or negatively associated with correctness. We find that reasoning-oriented training does not preferentially amplify the highest-Lift behaviors, motivating process-level objectives that reward calibrated and grounded reasoning rather than surface form alone.

Two GPU kernels implementing the same scaled INT8 GEMM interface are usually treated as interchangeable. We test that assumption: holding the checkpoint, prompts, hardware, inference engine, decoding, and quantization configuration fixed, we swap only the INT8 linear kernel (CUTLASS versus Triton) inside vLLM. At 1.7B each arm reproduces itself bit-for-bit across cold restarts, yet the arms agree on no sequence in any end-to-end comparison we ran (0/8, 0/16, and 0/64). What makes this more than a benchmark discrepancy is an integer alibi: for shared INT8 operands under a verified no-overflow bound, the INT32 dot product is exact and order-independent, so the accumulator cannot be the source of any difference. Feeding both kernels identical operands from every linear layer of Qwen3-1.7B and 8B (196 and 252 layers), we find bit-identical outputs under power-of-two scales, confirming a pinned prediction list 196/196 and 252/252 (pre-registered at 1.7B, pinned but not blind at 8B), and observed differences of at most one bfloat16 spacing under the checkpoints' real scales. This localizes the divergence to scale application and output rounding after the exact accumulator. Applied as a probe checkpoint, the same intervention restores end-to-end bitwise agreement (8/8 and 16/16 sequences). Cross-implementation FP8 GEMM shows a different signature: both the prevalence and the magnitude of differences grow with reduction depth, while the INT8 fraction stays at parts per million and within one spacing over a 64x range of K. Teacher-forced replay ties layers to tokens: flips concentrate at small logit margins, which predict flip risk with ROC-AUC 0.94 on 16,384 positions. We will release the pre-registration, per-layer predictions, manifests with kernel-selection evidence, and a conformance procedure that turns these controls into a concrete check for kernel interchangeability.

Neurodegenexrative diseases such as Alzheimer's disease and Parkinson's disease are diagnosed most reliably only after substantial, often irreversible, neuronal loss has already occurred, creating an urgent need for quantitative tools that can detect subtle, early, and individual-specific brain changes from neuroimaging data. This review surveys a broad and rapidly evolving toolkit of data-driven techniques for translational neuroscience and personalized neuro-health, organized around four complementary methodological pillars. Throughout, we emphasize how these methodologically diverse approaches converge on a common translational goal: personalized, mechanistically grounded, and clinically actionable models of individual brain health, and we close by discussing the principal open statistical, computational, and clinical challenges that remain.

In LLM-based code generation, Non-Functional Requirements (NFRs) are often specified as terse one-line phrases. We ask whether grounding those specifications in ISO/IEC 25010 Quality Model, either as rich natural-language prose (NL-rich) or as structured JSON (Structured), improves code generated on HumanEval/HumanEval-ET compared to a RobuNFR-style one-line baseline (NL-simple). We evaluate four NFRs (performance, error handling, code smell, readability) with ten prompt variations per condition under a fixed model snapshot and paired non-parametric analysis. Primary finding: ISO-grounded enrichment improves static quality proxies (unreadability density falls across all four NFRs (e.g., Performance 0.88 -> 0.69 for NL-rich)) and reduces sensitivity to prompt wording, but does not reliably improve functional correctness; for error handling, extended-test pass rate decreases, suggesting tension between defensive coding patterns and exact-output benchmarks. Secondary finding: when ISO content is held constant, NL-rich and Structured differ negligibly in correctness (|delta| <= 0.023), indicating that semantic content matters more than JSON-vs-prose format. Practitioners should invest in standard-grounded NFR content rather than serialization form. A fully traceable replication package is provided.

Synthesizing time series from natural language is emerging as the most expressive form of controllable time series generation. However, existing text-conditioned generators either take caption embeddings frozen from off-the-shelf text encoders, or adapt the encoder end-to-end, letting the denoising loss shape the embeddings only as a by-product. In either case, the conditioning representation is never deliberately matched to the signal modality, leaving it ill-suited to guide generation. We address this by introducing GALA: Generation-Aware cross-modaL Alignment for text conditional time series generation. GALA is a two-stage approach that first contrastively couples a pretrained text encoder with a time-series foundation model into a shared embedding space with both encoders adapted to generation by an auxiliary generative loss, and then freezes the resulting caption embedding to drive a flow-matching generator. On TSFragment-600K, spanning four domains and three fragment lengths, GALA sets a new state of the art, ranking first in 30 of 36 metric columns and reaching an average rank of 1.08/1.08/1.42 at lengths 24/48/96 against 1.92/2.00/1.75 for the strongest baseline. We further find that generator-internal text encoders force a trade-off between fidelity and caption adherence, whereas conditioning on the aligned embedding breaks it: FID, CTTP, and JFTSD all improve at once. Ablating the auxiliary loss degrades FID, CTTP and JFTSD together, it indicates the generative term is a necessary component of the alignment rather than an add-on.

Empirical reports on the true cost of AI-intensive software development remain scarce, and the few that exist are easy to get wrong in ways that never surface in the final number. We report early results from an ongoing case study: a six-person student team built a full conversational onboarding assistant – RAG-based code chat, guided tours, dependency graphs, technical-debt analysis – over one academic term using pervasive AI assistance. We instrumented development with a three-layer cost model (real AI spend, self-reported human effort, human counterfactual) and initially reported a 19.4x cost ratio. A follow-up pass revealed two independent errors – inferring per-token cost under a flat-rate subscription, and pricing the counterfactual with the wrong regional labor rates – that together had inflated the ratio by roughly 2x; the corrected figure is ~9.9x. We present this correction as an early, generalizable finding in its own right: both errors are easy to make, invisible in the final number, and plausibly common in similar reports. We outline next steps toward a more robust, replicable costing methodology for AI-intensive development.

PLAUD (Performative Latents and Unsupervised DDSP) is a neural synthesizer and Max for Live instrument for live electronic music, built on NoiseBandNet and trained on small personal sound corpora. We present its architecture, combining a variational DDSP synthesis model, latent smoothing, multi-scale spectral and adversarial losses, and an optional transformer prior, alongside a set of bending operations that intervene directly in the synthesis chain: component limiting, waveshaping, and prior feedback. The Max for Live interface exposes control generation, trajectory sampling, and modulation as primary modes of interaction. Throughout, we thread an affordance analysis arguing that the system's performative character follows from architectural decisions rather than being designed on top of them. The paper contributes both a technical account of the system and a situated affordance analysis of its role in live electronic music performance.

In reasoning supervised fine-tuning, candidate responses for the same instruction can differ substantially in how well they match the student's current distribution. Recent likelihood-based response selection methods suggest that responses closer to the student distribution provide more effective supervision, motivating the hypothesis that high-likelihood responses may generally be preferable for fine-tuning. In this paper, we revisit this intuition and show that the value of likelihood-based data selection depends critically on model capacity and training duration. Through controlled experiments on mathematical reasoning, using students ranging from 1.5B to 8B parameters and supervision generated by stronger teacher models, we observe a clear capacity-dependent “{\color{SMALLCOLOR}Fast-Fit} / {\color{LARGECOLOR}Slow-Gain}” pattern. High-likelihood data provides faster and more stable early improvements, especially for smaller models, but low-likelihood data becomes increasingly beneficial for larger models when training is allowed to continue longer. To explain this phenomenon, we analyze learning dynamics, showing that small models often fail to absorb low-likelihood supervision and instead fall into shallow or repetitive behaviors, while larger models are better able to move toward the teacher distribution under such data. We further provide a capacity-constrained theoretical view of distillation that clarifies how data difficulty, data span, and student capacity jointly govern transfer. Overall, our findings show that effective data selection for reasoning should be aware of model capacity and computing budget rather than based on a single universal preference for high-likelihood supervision.

Computer-aided detection (CADe) systems for colonoscopy promise to reduce clinical miss rates, yet reliable real-world deployment remains elusive. This translational gap stems in part from a structural flaw in model development: the reliance on curated datasets that under-represent the long negative stretches and procedure-related artifacts characteristic of routine examinations. Training and evaluating architectures strictly on these lesion-centric benchmarks creates an illusion of success, since such benchmarks cannot capture clinically crucial metrics. To expose this gap, we establish TRUE-Colon, a standardized benchmarking protocol that measures key deployment characteristics alongside localization accuracy, and evaluate four real-time architectures (Faster R-CNN, YOLOv8, YOLOv11, RT-DETR) across curated benchmarks (SUN, PICCOLO) and 60 unedited, full-length procedures (REAL-Colon). We observe a consistent transfer asymmetry: models trained strictly on curated clips suffer a severe performance collapse when evaluated on full procedures, whereas procedure-trained models substantially improve rejection of non-polyp content on REAL-Colon, and largely retain their accuracy on curated benchmarks. Beyond transferability, we find that the Transformer detector attains the strongest sensitivity and the earliest, most persistent detections, while the convolutional detectors stay competitive at a higher throughput. Together, these results indicate that both training and benchmarking for deployable CADe should shift from curated, lesion-centric clips toward full-procedure data and deployment-relevant operating points. Source code is available at https://github.com/sdoerrich97/true-colon.

Spiking neural networks (SNNs) offer an energy-efficient alternative to conventional deep neural networks by exploiting sparse event-driven computation, but their training remains challenging because the non-differentiable spike function requires surrogate gradients whose fixed shape may be suboptimal across layers and training stages. In this work, we introduce SAGE, an uncertainty-modulated surrogate-gradient mechanism for Transformer-based SNNs. SAGE estimates block-level uncertainty from normalized self-attention entropy and uses this signal to adapt the surrogate-gradient slope during training while leaving the inference model unchanged. By modulating only the training-time surrogate parameter, the proposed method preserves the original architecture and deployment cost while improving optimization flexibility. Experiments on CIFAR-10/100 demonstrate that SAGE achieves improved accuracy over fixed-surrogate baselines, with results up to 1-2% consistent gains across multiple simulation time steps. These results highlight the potential of attention-derived uncertainty as a lightweight training signal for adaptive surrogate-gradient learning in transformer-based SNNs.

Reinforcement Learning with Verifiable Rewards (RLVR), often optimized with Group Relative Policy Optimization (GRPO), has become a central recipe for improving the reasoning capabilities of pretrained language models but current studies remain heavily English-centric. We conduct a large-scale empirical study of multilingual and non-English GRPO across a wide range of base models, training languages, and different reasoning language rewards. We find that training to reason in the native language often leaves only a small gap to training for English reasoning. We further observe strong crosslingual transfer: training in one language often improves performance in many others. However, specific trends are highly model- and language-dependent. In some cases, training in a particular language induces severe regressions on out-of-domain capabilities in other languages. Our analysis shows that RLVR beyond English can provide broad crosslingual gains, but also requires broad evaluation to detect language-specific regressions.

Large language model providers routinely cite multilingual safety benchmarks spanning a dozen or more languages as evidence that their models are safe for non-English-speaking users. We show that these collection-level coverage claims frequently do not survive inspection at the level of an individual language. Auditing 21 resources across 25 language slices, of which 20 count as datasets under our counting rules, spanning three languages chosen to represent low- (Hausa), mid- (Swahili), and high-resource (French) tiers, we find that gaps in provenance, annotation reliability, access, harm-taxonomy coverage, and data reuse recur in patterns that partially, but not fully, track resource level. Using a controlled within-pipeline comparison, we show a Hausa-language slice falling below its own paper's translation-quality acceptance threshold while the same pipeline's Swahili output clears the same bar comfortably; this is evidence that these gaps are measurable and addressable, not inherent. We further show that self-harm and sexual-content categories have no native-language coverage in either African-language tier we studied, a total rather than gradated gap that a purely resource-level account does not predict. We connect these findings to a documented, persistent asymmetry in multilingual jailbreak robustness (single-turn attacks largely mitigated, multi-turn attacks still effective), arguing that this asymmetry is structurally consistent with where our audit finds training and evaluation data thinnest. We contribute a reusable slice-level audit methodology, a cross-tier empirical comparison, and concrete recommendations for dataset creators, model providers, and venues aiming to make “multilingual coverage” claims verifiable rather than merely stated. Dataset: https://huggingface.co/datasets/ChialukaOnuoha/safety-slice-audit

Translating C code into safe, idiomatic Rust is a longstanding software-engineering goal because it can eliminate entire classes of memory-safety vulnerabilities while preserving the functional behavior of legacy systems. Large language models (LLMs) have shown promise for this task but typically underperform when applied off-the-shelf, since general-purpose pretraining rarely emphasizes idiomatic Rust generation, cross-language semantic equivalence, or the ability to reason about and repair compiler/runtime feedback. In this report we describe a three-stage fine-tuning curriculum applied to Qwen3-27B that is designed to progressively specialize the model for the C-to-Rust (C2Rust) translation task: (1) continued pretraining on Rust-centric corpora to strengthen the model's prior over idiomatic Rust syntax and standard-library usage; (2) supervised fine-tuning (SFT) on the microsoft/Verus_Training_Data dataset to instill debugging and self-repair behavior over Rust code; and (3) task-specific SFT on paired C/Rust solutions derived from LeetCode problems to teach direct semantic translation. We evaluate the resulting model using the agentic, static-analysis-guided verification framework of SACTOR, which performs structure-aware, two-phase (unidiomatic to idiomatic) translation with foreign-function-interface (FFI)-based end-to-end (E2E) testing. We report success rate, idiomaticity (Clippy lint counts, unsafe-code fraction), and failure-mode analyses, and compare our fine-tuned model against baseline Qwen3-27B and other LLMs evaluated under the same framework.

Executing Linear Temporal Logic (LTL) instructions with learned robot policies faces two practical challenges: demonstrations may cover individual behaviors without containing the temporal compositions requested at deployment, and semantic success predicates often provide little useful motor guidance through quantitative robustness. We address these challenges by separating behavior learning from temporal composition. From the same offline demonstrations, we train an unconditioned multimodal diffusion policy and a semantic predictor that estimates which semantic outcomes are likely to follow a proposed action sequence. These predictions provide a learned guidance signal without requiring hand-designed robustness measures for semantic task objectives. At deployment, an LTL automaton tracks instruction progress and scores the policy's proposed actions according to their predicted semantic outcomes. Neither learned component receives the specification during training, allowing demonstrated behaviors to be reused in longer, previously unseen temporal compositions without retraining. For additional runtime safety constraints with informative continuous margins, an optional robustness-based controller locally refines the selected actions. Experiments in a navigation environment, CALVIN manipulation, and on a real robot demonstrate reliable execution of complex temporal instructions and additional safety constraints, with substantial improvements over temporal-logic baselines.

Objective: Foundation models represent the next advancement in AI for EEG analysis; however current explainable AI techniques provide attribution scores in the time-channel input space, which is mismatched to clinical intuition about EEG. Thus, there is a critical need for a universal method that can extend the interpretability of any foundation model to alternative and physiologically relevant domains without modifying or retraining the underlying model. Methods: EEG-PRISM leverages linear transformations and established backpropagation rules to map time-channel attribution scores into alternative domains. We derive mappings to the frequency domain via an invertible DFT and to the source domain via an approximately invertible EEG generative model. We evaluate EEG-PRISM in simulated and real data, assessing recovery of ground-truth phenomena across domains with five foundation models and four AI explainers. Results: In simulation, EEG-PRISM achieves near-perfect spectral recovery and 69.2% spatial accuracy. In epilepsy, EEG-PRISM correctly determines that delta-theta activity is most salient and correctly localizes the seizure onset region with 50% accuracy. In autism, EEG-PRISM localizes the predictive delta-alpha biomarkers to frontal and temporal regions, consistent with prior work. Conclusion: EEG-PRISM is a theoretically-grounded post-hoc attribution method with accurate mapping into the spectral and spatial domains. It supports window-level analysis of transient events (e.g., seizures) and group-level identification of clinically relevant biomarkers (e.g., autism), thus advancing interpretable EEG foundation models. Significance: This work enables physiologically-grounded interpretation of EEG foundation models and supports clinically relevant insights such as event localization and biomarker identification.

Between October 2018 and July 2026 AI models progressed from simple systems like BERT to massive agents that solve complex math and write software. The ability to resolve real coding issues improved by nearly six times per year since late 2024. During this time costs dropped sharply with OpenAIs budget model GPT 5 point 6 Luna matching flagship capabilities for just one to six dollars per million tokens beating older versions at a fraction of the price. Top performance is now split across specialized models as Claude Opus 5 leads in frontend coding Claude Fable 5 excels at repository level coding and GPT 5 point 6 Sol dominates terminal tasks. In a grade school math test using the Qwen 2 point 5 model basic methods solved 58 of 100 problems while advanced sampling solved up to 79. A confidence ranking tool correctly identified 47 right answers in its top 50 choices proving highly useful for sorting tasks with all research materials made fully public.

Linear attention keeps a state of fixed size. At long context, many stored items share this state, and interference between them degrades retrieval. Gated DeltaNet-2 (GDN-2), like every delta-rule model before it, derives its erase vector from the key of the current token. However, the interference in its reads is measured through the query, and the erase step cannot reach it. We introduce the Query-derived Erase Direction (QED). QED adds a second erase direction derived from the query and orthogonal to the key. In the fast-weight view, a key-directed delta edit cannot change the key-orthogonal part of a read. It uses the editable part to cancel old-state content measured along the query. It also improves retrieval at every length past the training window, and it about doubles the usable context length on S-NIAH-1.

The clinical management of gynecological diseases often relies on medical imaging for diagnosis, treatment planning, and follow-up. Segmentation in this setting is challenging because successive tasks may differ in imaging modality, target anatomy, pathology, and annotation structure. Continual learning allows models to adapt to new tasks without simultaneous access to previous datasets. However, when successive tasks differ substantially, learning a new task can degrade performance on earlier ones, a problem known as catastrophic forgetting. Understanding where adaptation disrupts previous knowledge can help guide the design of more targeted continual-learning strategies. We investigate how forgetting changes as different parts of an encoder–decoder network are allowed to adapt. We progressively expand the trainable region of a 3D nnU-Net backbone from the bottleneck toward input- and output-proximal blocks. Under a shared learning rate, adaptation near the bottleneck largely preserves previous-task performance but provides limited current-task learning, whereas broader adaptation improves current-task performance but sharply increases forgetting. This trade-off persists even when the average change in trainable backbone parameters is approximately comparable. Assigning different learning rates to different blocks substantially reduces forgetting when part of the backbone is trainable, although this changes both the size and location of the updates. Forgetting still increases as more blocks are trained and remains severe when the full backbone is updated. These results show that forgetting depends not only on how much the model changes, but also on which parts of the model are allowed to change.

Generic event-level anomaly detection for collider physics has two recurring problems: anomaly scores are hard to interpret, and they correlate strongly with energy scale and object multiplicity. We present Organized Representation via Contrastive learning for Anomaly detection (ORCA), a two-stage framework that first learns an embedding space via supervised contrastive learning across a diverse set of physics processes, then runs a standard autoencoder in that space to generate event-level anomaly scores. On a simulated dataset consistent with conditions at the High-Luminosity Large Hadron Collider, ORCA delivers significant gains in both breadth and depth of sensitivity to new physics signals with respect to a baseline autoencoder architecture. Beyond improved sensitivity, the contrastive embedding makes the anomalous sample interpretable: because known processes occupy distinct regions of the space, a maximum-likelihood template fit to the embedding distributions can attribute events in an anomalous sample to template physics processes with quantified uncertainties. We demonstrate that the fit accurately recovers injected signal yields, including for signals excluded from the training of the embedding, and characterizes signals absent from the template library through the known processes they most resemble. These results establish ORCA as a route to interpretable anomaly detection-based searches at colliders, where the embedding geometry carries higher dimensional physics information compared to standard one-dimensional output fits, enhancing downstream statistical analysis.

We solve exactly a fundamental problem of adaptive control against adversarial disturbances: regulate the scalar system $x_{t+1} = ax_t + u_t + w_t$, $x_0=0$, $\|w\|_\infty \le 1$, where the constant pole $a \in [-\Delta, \Delta]$ is unknown in sign and magnitude and $\Delta$ is arbitrarily large. Elementary as the system looks, the least worst-case peak $\|x\|_\infty$ that a causal controller can guarantee against an adversarial pair $(a, w)$ (the value of this game) has, to our knowledge, never been determined for any adaptive control problem with parametric uncertainty of arbitrary size under this criterion; existing theory supplies stability certificates, gain bounds, and regret rates, not the value. That value is $\gamma^\star(\Delta) = 1 + \Delta$ for every $\Delta>0$. The summand $1$ is the irreducible price of the disturbance, and $\Delta$ the exact price of a single, unavoidable identification spike. The optimal policy is certainty-equivalent deadbeat control at the midpoint of the set-membership consistent interval, an instance of the robust oracle $\times$ consistent model chasing architecture. The architecture is forced, not merely sufficient: writing $\theta_t := -u_t/x_t$ exhibits every causal controller as an oracle-selector composition, and optimality pins the selector to the midpoint at the critical histories. The standard tools, classical and modern, each fail quantifiably: probing is punished before it pays, commitment is fatal at sub-disturbance excitation once adaptation is necessary, optimism degenerates to tie-breaking or pays asymptotically at least twice the optimum, and regret certificates are blind to the worst-case peak in both directions. The optimal law contains no exploration mechanism, its learning purely passive. These results give the first exact optimality certificate for consistent model chasing as a design principle for adversarial adaptive control.

Machine learning is now a central tool for solving inverse problems in particle physics and astronomy. Models are trained on simulation and deployed on real data, raising the question not just of whether they fit, but of whether they are wrong in ways we did not anticipate: the unknown unknowns. This challenge of model misspecification is not unique to machine learning. In physics, misspecification is sometimes exactly what we want to find: new discoveries appear as failures of existing models. At other times, we want such effects absorbed into the analysis without biasing the measurement. A robust analysis is one that absorbs the misspecifications we are not interested in, while preserving sensitivity to the ones we are. Machine learning can both amplify misspecification and provide new tools to address it. We discuss the challenges of model misspecification, diagnostics for detecting it, and strategies for mitigation. No single diagnostic can confirm that a model is correctly specified: detection and mitigation are two halves of an iterative loop, in which a battery of complementary diagnostics is applied, the model is updated, and the process repeated. Robustness against unknown unknowns is ultimately less about any single technique than about a disposition: a willingness to suspect one's own model, and to design analyses that can survive being wrong in ways one did not anticipate.

Clinically actionable, patient-specific hemodynamic assessment, specifically wall shear stress, vortex structure and pressure distributions, is critical for determining risky or unfavorable evolution in Abdominal Aortic Aneurysms (AAA). While Physics-Informed Deep Operator Networks (PI-DeepONets) show promising results in complementing established 5 tools such as Computational Fluid Dynamics (CFD), a persistent architectural challenge remains for complex 3D flows. In this direction, we propose a Modified Multi-Input Multi-Output PI-DeepONets (M3PI-DeepONet) designed for predicting unsteady flows in an idealized AAA geometry. Central to our model is the Aggregated Injection strategy, where latent representations from multiple input branches are fused prior to trunk injection, allowing the coordinate basis to adapt to multiple physical constraints. To the best of our knowledge, this is the first architecture to combine the layer-wise gating mechanism with a multi-branch operator-network topology, yielding an input-adaptive trunk basis. Additionally, we integrate the 3D Navier-Stokes equations as governing physical laws, so the model is trained based on physics-informed residuals, initial and boundary conditions, and only 0.3% of the labeled internal data together with the selected branch-conditioning signals. The M3PI-DeepONet simultaneously predicts unsteady 3D flow velocity and pressure fields with an average relative L2 velocity error below 4% and pressure error around 5% while achieving a conservative retained-cycle inference speedup of approximately 36x compared to reference CFD simulations once the branch inputs used for conditioning are available. This work advances the application of deep learning in cardiovascular disease modeling, marking step toward real-time, non-invasive clinical diagnostics.