Papers

As LLM coding agents increasingly perform end-to-end engineering work, we lack empirical characterization of how they behave on systems-level requirements: schema design, async orchestration, configuration correctness, and retrieval-filtering trade-offs. We present a case study of one such agent implementing a multi-component data system against a detailed pre-existing specification. Storage technologies, schema, entity-resolution algorithm, and retrieval-filtering strategy were fixed in advance; the agent autonomy was in the implementation, in diagnosing and fixing defects it introduced, and in interaction-design choices left open. Over a single session, we catalog five such defects, categorized by constraint violated and detection method. We further evaluate, on the public HotpotQA benchmark, the one retrieval trade-off specified in that architecture: restricting candidates to a graph-identified entity set before ranking versus unfiltered search. We substitute the benchmark gold evidence labels for entity identification, since we lacked LLM access to run that stage, and report standard recall rather than the benchmark own accuracy metrics. Across retrieval budgets from 1 to 10 and 100 questions against a pooled corpus of 2994 paragraphs, filtered recall reaches its ceiling by a budget of 3, expected once candidates are restricted to the gold paragraphs themselves, while unfiltered search recovers all required evidence only 69 percent of the time even at a budget of 10, a gap that holds at every budget tested, with sign test p less than 0.0001. We close with a discussion of where the agent autonomy succeeded versus required correction, including one instance where a claimed performance fix was never re-measured on the regression that motivated it.

Graph Convolutional Networks (GCNs) are widely used for large graph-structured data, including social, citation, and e-commerce networks, but their deployment is constrained by irregular memory access and severe GPU workload imbalance. These challenges arise in two dimensions: width imbalance from power-law degree distributions and depth imbalance from heterogeneous neighborhood connectivity.We present DualGCN, a GPU acceleration framework addressing both dimensions through dual-metric graph partitioning and adaptive kernel execution. DualGCN combines node degree, reflecting aggregation width, with neighborhood density estimated by anonymous random walks, capturing multihop connectivity and access depth. This hybrid workload metric enables connectivity-aware partitioning of large graphs into sparse and dense regions while reducing workload imbalance from linear to logarithmic complexity. DualGCN then selects partition-specific execution strategies: sparse partitions use warp-level parallelism and coalesced memory access, whereas dense partitions exploit instruction-level parallelism to hide latency and improve GPU utilization. Experiments on twelve real-world graph datasets show that DualGCN consistently accelerates GCN computation, achieving average speedups of 2.53x, 3.8x, and 2.13x over cuSPARSE, GNNAdvisor, and ACCEL, respectively. These results demonstrate that jointly optimizing graph partitioning and kernel execution provides an effective solution for processing large-scale graph and socialnetwork workloads.

Large language models (LLMs) are increasingly used as assistants for statistical and data science work, yet existing evaluations largely assume the analysis target is already specified. In practice, users arrive with informal goals and heterogeneous data, leaving the model to decide what statistical task is implied and which data are relevant. We first formalize this upstream step as Statistical Problem Formulation and decompose it into two subtasks: (1) Statistical Problem Classification and (2) Variable Identification & Role Assignment. We then introduce StatFormBench, a benchmark built from five cross-domain statistics textbooks and a data science case library, covering diverse problem types, data representations, and scenario styles. It contains 1,013 samples spanning 20 coarse-grained and 85 fine-grained statistical problem categories. Across 14 open- and closed-source LLMs, the best zero-shot models reach only 72.0 fine-grained classification accuracy and 63.2 variable set overlap. No model performs consistently best across the two subtasks, while enhanced prompting strategies yield only limited or inconsistent gains. We release the benchmark data on Hugging Face at https://huggingface.co/datasets/THU-CongLab/StatFormBench and the evaluation code on GitHub at https://github.com/THU-CongLab/StatFormBench.

Mastering musical performance requires precise multisensory coordination, yet learners encounter a kinesthetic mismatch, which is a discrepancy between the internal perception of an action and the actual physiological state of the body. While multisensory Body Transformation Experiences (BTE) provide tools to bridge this gap, existing designs often focus on external correction rather than internal alignment. To address this, we propose the Somatic Alignment Mindset (SAM), a conceptual lens that integrates Taoist philosophy to shift the focus of HCI design from prescriptive feedback toward holistic embodied unity. By positioning technology as a reflective medium, SAM operationalizes the principles of Adaptation, Assessment, and Awareness to reconcile somatic discrepancies and foster deep, self-aligned musical mastery.

Integrated sensing and communication (ISAC) enables simultaneous communication and environmental sensing in unmanned aerial vehicle (UAV) networks, but its performance is constrained by the physical antenna aperture and residual self-interference (SI) in full-duplex (FD) sensing. To address these issues, we propose a shared-aperture ISAC architecture in which a sparse co-prime array (CPA) is embedded in a uniform linear array (ULA) grid for FD sensing, while the remaining antenna positions support time-division duplexing (TDD) communication. We characterize the sensing performance through an order-wise Cramer-Rao bound (CRB) analysis, showing that the CPA achieves a stronger asymptotic sensing gain than the partitioned ULA benchmark in both single-target and nondegenerate multi-target scenarios. We further reveal a space-time sampling tradeoff under the same physical aperture. Based on the proposed architecture, we formulate a non-convex joint resource allocation problem that maximizes the weighted downlink-uplink sum rate by jointly designing the sensing transmit covariance, downlink precoder, and uplink receive beamformers under sensing accuracy, BS transmit-power, communication QoS, and residual SI constraints. An alternating-optimization-based algorithm is developed. Simulations demonstrate consistent performance gains over the considered baselines and confirm the complementary benefits of the CPA virtual aperture and sensing covariance optimization.

Large language models (LLMs) are increasingly embedded in organizational work, yet their errors often pass human review. Prior research locates such failures in users' capability to review LLM output or their engagement in doing so. We develop an alternative, retrieval-based account of human oversight and posit that error detection is more effective when oversight-relevant information is accessible to users at the moment of review. Across two randomized lab-in-the-field experiments with 640 customer-facing employees, we show that self-generated explanations improve error detection and strengthen recall of verification-relevant reasoning, while cues that reactivate such reasoning help sustain detection under repeated LLM use. Theoretically, we identify information retrievability as a distinct precondition for effective oversight and specify generative encoding and cue-supported reactivation as mechanisms that build and sustain it. Practically, lightweight onboarding self-explanations and daily retrieval cues can make human oversight more resilient as LLM use becomes routine.

Recent GPU generations include special-purpose ray tracing (RT) cores for graphics applications. While RT cores are primarily used for rendering, recent works show they can be leveraged for general-purpose tasks, including similarity searches. However, existing approaches do not support datasets exceeding three dimensions. In this work, we propose RT-HiSS, the first exact GPU RT-core-based similarity search algorithm for high-dimensional datasets. GPU similarity search often scales poorly for large datasets with substantial search distances. To address this, RT-HiSS uses RT cores for fast index construction and searches, followed by candidate refinement on CUDA cores. We introduce a two-pass approach to estimate an upper bound on result size, enabling efficient batching under GPU memory constraints with near-perfect load balancing. Additionally, we examine shared memory tiling and compressed result masks to improve GPU resource utilization. RT-HiSS yields speedups up to 8.37$\times$ over competitive state-of-the-art GPU algorithms across six real-world datasets.

While breathing is essential to living and for sound production in some instruments, for pianists, it is often a hidden and automatic process, making it difficult to analyze or refine. A critical gap exists between data and awareness: while sensors record precise physical metrics, they fail to capture the performer's somatic experience. Conversely, the high cognitive load of performance makes it nearly impossible for musicians to recall their internal states with temporal precision. To address this, we present a system, Breathing Mirror, and associated methodology designed to externalize the pianist's internal somatic experience through three analytical lenses: a Baseline View (synchronized signals), a First-Person View (subjective recall), and an Interpersonal View (collaborative reflection). Through a four-week longitudinal study with a skilled amateur pianist (35 years of experience), we evaluated the system's effectiveness by recording respiratory data using textile-integrated strain sensor belts. The results show that the Breathing Mirror reveals some patterns of breathing-music coupling and identifies critical blind spots where objective data diverges from subjective perception. Furthermore, we propose four somatic themes regarding the link between breathing and musical elements, offering a foundation for future large-scale validation across a broader range of pianists. This work provides a new way to study body signals, transforming breathing from an internal biological function into an articulate expressive parameter.

This paper proposes a dimension-reduced approximate dynamic programming (ADP) method for real-time microgrid operation with massive air-conditioning loads under multiple uncertainties. The operation problem is formulated as a multi-stage Markov decision process, and a post-decision value function is introduced to characterize the impact of current decisions on future operating costs. To address the curse of dimensionality caused by massive air-conditioning loads, a consistency-based value function projection is developed to map the high-dimensional state space at each node into a tractable aggregated state space. Based on the reduced states, piecewise linear approximation is further employed for efficient value function training. Case studies on 33-bus and 123-bus systems show that the proposed method achieves near-optimal operation performance with low computational cost and good scalability under both deterministic and stochastic conditions.

AI is rapidly advancing neuroscience, yet many laboratories fail to fully unleash its potential due to significant interdisciplinary barriers. While pre-trained neural models for physiological data are progressing quickly, their heterogeneous architectures and modality-specific constraints hinder systematic integration, selection, and evaluation. Despite recent advances in large language model (LLM)-based agent systems for intelligent scientific applications, existing approaches often still lack the domain expertise required to effectively select and coordinate diverse neuroscience pre-trained models and handle unique data types in this domain. We present NS-Copilot, an LLM-driven multi-agent system for neuroscience analysis that autonomously supports end-to-end workflows for diverse professional tasks. It unifies domain-specific pre-trained models and supports key neuroscience modalities, including EEG and extracellular spike data, through a natural-language interface. Given raw data and a task description, NS-Copilot orchestrates agents with specialized roles for planning, adaptive control, code generation, and result synthesis, enabling analysis without dataset-specific heuristics. We evaluate NS-Copilot on neuroscience benchmarks spanning Alzheimer's disease, Parkinson's disease, and working memory spike decoding. Across 8 trials per task, the system consistently outperforms strong baselines on the primary metric, demonstrating the ability of NS-Copilot for effective and scalable neuroscience analysis. Our code is publicly available at https://github.com/FrankLiu1102/ns-copilot.

Data-driven process simulation aims to generate realistic case trajectories from historical event logs without requiring an explicitly specified model of the underlying dynamics. Deep sequence models can capture complex temporal dependencies through next-activity probabilities and conditional time distributions. However, event logs provide only a partial view of the underlying process state, often recording activity completions without the corresponding service-start times. Consequently, the same observed process history may be consistent with multiple plausible latent process conditions, whereas standard recurrent models compress each process prefix into a single deterministic recurrent state. We propose a Unified Particle Filter LSTM (Unified PF-LSTM) that maintains and sequentially updates a weighted set of recurrent-state hypotheses. We summarize this particle belief using its weighted mean and learned features based on the moment-generating function. The resulting representation is used to predict a categorical distribution over the next activity and conditional quantiles of the current activity's sojourn time. The framework is trained end-to-end from event-log data and evaluated on three real-world emergency department datasets. The results show that the proposed framework consistently outperforms the considered data-driven baselines in reproducing routing, duration, and system-level behavior across all datasets, with particularly strong gains in settings where complex process dynamics are only partially reflected in the available event logs.

Content-based image retrieval (CBIR) has advanced significantly with deep learning, yet effectively ranking similar images remains challenging, particularly in high-dimensional feature spaces, where pairwise distances often fail to capture contextual relationships and the semantic gap between visual features and high-level concepts persists. Manifold learning and rank-based refinement methods have emerged as complementary strategies, respectively improving feature representations and exploiting contextual information embedded in ranked lists, such as neighborhood relationships among images. However, combining these projection-based and rank-based strategies to exploit their complementary properties remains a challenging research problem. To address this, we propose a framework that combines neighbor embedding projections with rank-based manifold learning through rank aggregation. Uniform Manifold Approximation and Projection (UMAP) generates alternative low-dimensional feature representations, and ranked lists obtained from UMAP projections and rank-based re-ranking methods are combined using the Borda Count aggregation strategy. Experiments were conducted on several public datasets using deep learning features extracted from ResNet152, Swin Transformer, and DINOv2 models. Results show that the proposed approach improves retrieval effectiveness in several scenarios, particularly when the baseline representation struggles to achieve high precision. The aggregation strategy also often improves the quality of top-ranked positions, leading to competitive Mean Average Precision (MAP) and Precision values across different datasets and feature extractors. These findings suggest that combining projection-based and rank-based manifold learning strategies through rank aggregation can provide complementary contextual information for image retrieval tasks.

Ultra-low-bit language models can reduce storage and memory bandwidth, but a nominal "1.58-bit" label does not fully describe the stored representation, retained capability, or runtime behavior. We study an end-to-end post-training conversion of Qwen, an instruction-tuned 4B-parameter model, using KOTMS rotation, E2M-ATQ ternarization, and GPTQ-style error compensation from TWLA. The experiment is weight-only: activations remain at 16-bit precision, so ILA-AMP is omitted. We evaluate effective bit accounting, task capability retention, perplexity, calibration sensitivity, checkpoint composition, and deployment behavior. The final conversion uses 1.641 effective bits per weight for quantized linear weights, with 81.62% of model parameters targeted. Across ten scored capability comparisons, accuracy falls from 64.5% to 54.7%. Degradation is uneven: BoolQ retains 84.6% chance-corrected teacher performance, while ARC-Challenge retains 43.8%. Perplexity rises from 13.639 to 18.748 on WikiText-2, 24.700 to 31.992 on PTB, and 19.831 to 28.966 on C4. A subsequent packing run preserves the ternary planes and scales, reducing reported model size from 8.29 GiB to 3.96 GiB with essentially unchanged perplexity. A separate third-party packing attempt was lossy and is excluded from the primary artifact claim. The packed artifact has not been benchmarked end-to-end for task accuracy or generation throughput. A preliminary Triton GEMV microbenchmark is 4.6x slower than FP16 cuBLAS on one tested shape. We therefore do not claim that compression alone yields faster inference.

Augmenting the dexterity of human surgeons has the potential to free them from tedious subtasks. We consider debridement (removal of diseased or dead tissue fragments), which is challenging due to imprecision in spatial perception and cable actuation. We develop an augmented dexterity system for surgical debridement that uses visual servoing to align the cable-driven gripper with the target position in the image plane, and then introduces a novel approach to depth control, MACAW: Monocular Adaptive Compact Attention Windows. Across 100 physical trials using the da Vinci Research Kit (dVRK) robot, camera-frame servoing reduced average gripper position offset from 37 to fewer than 5 pixels within 4 optimization steps, taking an average of only 0.39s. MACAW significantly outperforms procedural and learned VLA baselines, achieving a 93% success rate at 11 seconds per fragment, yielding a throughput of 304 fragments per hour. Extending MACAW to a bimanual debridement setup maintains a 92% success rate at an average of 7 seconds per fragment, increasing the throughput to 473 fragments per hour.

Multi-contrast MR scans contain redundant structural information that can be leveraged during reconstruction and potentially accelerate acquisition times. This idea has inspired end-to-end guided reconstruction models, leveraging one or more contrasts to guide the reconstruction of a different contrast. However, these models require large paired multi-contrast raw datasets for training, limiting their application in low-data regimes. In this work, we propose a modular framework, namely CoSMo-RecNet, for learning guided reconstruction models in the low-data regime. At its core is a reusable multi-contrast representation based on a content/style model, which can be learned from large-scale, publicly accessible, unpaired multi-contrast image datasets, without available k-space data. Using this frozen model as a multi-contrast prior and using a set of reference contrasts, the reconstruction problem reduces to a much simpler refinement problem that can be solved by a lightweight unrolled network and thus learned from small, task-specific reconstruction datasets. We demonstrate the efficacy of CoSMo-RecNet by evaluating it on the low-field 0.3 T M4Raw dataset, showing stable reconstruction quality on decreasing the raw training data budget. CoSMo-RecNet achieved higher reconstruction quality with 5 training subjects or lower compared to a parameter-count-matched MoDL trained on 100 subjects. On a data-limited and severely out-of-distribution ultra-low-field 47 mT Halbach scanner dataset, CoSMo-RecNet was superior to other viable strategies, including classical reconstruction, transfer learning, and zero-shot reconstruction.

Enterprise wireless access points (APs) are promising platforms for predictive machine learning (ML), but their primary responsibility remains providing wireless connectivity and network services. Predictive inference must therefore share an AP's CPU and memory with packet processing, Wi-Fi and IoT radio operations, and client management. This resource contention creates two risks: a model that performs well on proxy hardware may be too slow on the target AP, while a model that fits in isolation may still degrade network services under load. We define network-aware deployability using two gates: qualification of the model and its execution path on the target AP, followed by validation of its execution profile under packet-service and forecasting constraints. Our benchmarks show that edge testbeds do not reliably capture target behavior. Across matched artifacts and serving settings, five model implementations run 6.1–19.1$\times$ slower on an AP than on a Raspberry Pi~5, while peak memory usage differs by up to 22%. Moreover, two forecasting foundation models of similar size differ in AP latency by 19$\times$. When serving a smaller model across 13 parallel streams at a 30~s cadence under network saturation, default execution increases p99 round-trip time (RTT) by 76% and reduces throughput by 7.06%. Understanding these trade-offs is essential for live deployment if we aim to use APs for both networking and ML workloads.

Kolmogorov-Arnold Networks (KANs) place learnable B-spline activations on network edges rather than fixed activations on nodes. The standard Cox-de Boor recursion evaluates these activations through $k$ sequential passes for degree-$k$ splines, consuming over 90% of forward-pass time. InKAN replaces this recursion with the truncated power form, a classical result from approximation theory that expresses each uniform cubic B-spline as five $(x)_+^3$ terms at shifted knot positions. The resulting expression computes exact B-spline basis values: the same mathematical function as the Cox-de Boor recursion, evaluated without sequential passes. This paper documents three contributions: (1) an implementation structured for torch . compile fusion, eliminating recursion, span lookup, and scatter-gather operations; (2) a bounded-coordinate evaluation that clamps the normalized input to $[0, k{+}1]$, preventing the growth of cancellation error at large off-support coordinates; and (3) an open-source package (pip install inkan). In the tested configurations, InKAN has 2.8–3.5$\times$ lower forward-pass latency than the Cox-de Boor recursion. Partition-of-unity errors remain below $10^{-5}$ for grid sizes up to 200.

Large language models are increasingly used for subjective tasks such as seeking advice, where there is often no single correct answer. Yet dominant alignment methods learn a single behavior policy that applies uniformly across users, limiting their ability to reflect individual value priorities. We propose personalized value alignment, a framework for adapting model responses to users' stable value profiles rather than to personas, demographics, or task-specific preferences used in prior personalization work. Our approach represents users through Schwartz value profiles and selects which of their values are relevant to each query before applying targeted prompting, activation steering, or LoRA-DPO. We create a dataset of 1,501 value-tension scenarios and evaluate over 152 diverse value profiles sampled from a nationally representative survey. Value-targeted methods outperform no-personalization, demographic, and full-profile prompting baselines, indicating that gains come from focused conditioning on query-specific value priorities rather than mere exposure. We further find that steering effects are not uniform across users: alignment gains vary by where a user's priorities fall in the value space. Combining targeted prompting with activation steering or LoRA-DPO yields the strongest personalization while keeping logical quality within adequate range.

Across psychological, philosophical, and theological traditions, creations have been understood as being "made in the image of their creator" - they reflect the person who made them. The rise of generative artificial intelligence, however, has blurred the throughline between creator and creation. Across three experiments (one pre-registered, U.S. and U.K. samples; N = 651), we examined differences in creator outcomes associated with AI-assisted versus human-only creative work. Compared with human-only work, AI-assisted work was associated with significantly lower self- and identity-relevant indicators, including authenticity, self-creation connection, positive affect, and intrinsic motivation, and with higher levels of the "existential ick" emotions of disgust, guilt, and shame. Additionally, we found that the authenticity gap was statistically mediated by the creators' perception that their AI-assisted work contains less of their love, above and beyond differences in perceived effort, uniqueness, quality, and expensiveness. We also ruled out effort justification as an alternative explanation. Together, these findings suggest that AI reshapes not only how creative work is produced, but also the extent to which creators experience their work as an authentic expression and extension of themselves.

Agent harnesses increasingly want to run small language models (SLMs) on the microtasks around a frontier large language model (LLM) planner: auto-approving shell commands, writing memory, selecting tools, ranking past turns. We ask whether off-the-shelf SLMs meet practitioner-defined thresholds and, when they fail, why, and whether quantization changes the answer. We build a benchmark of 4 such microtasks with fixed prompts and automatic metrics, each with a pre-specified threshold $\tau$ anchored to a cheap non-LLM baseline and a CI-aware eligibility rule (a configuration passes only if its confidence bound clears $\tau$). Sweeping Qwen3 0.6/1.7/4/8B at their best (FP16, greedy, one frozen prompt, no tuning), we find an eligibility gap: 0 of 16 (4 tasks $\times$ 4 models) configurations pass (verified by checking the raw outputs and parser behavior). A logprob decision-threshold diagnostic (T1/T3/T4; T2 via a context-length/cascade probe) separates the failures into capability deficits and failures that can be addressed by changing the decoding threshold (4 regimes). Quantization to 4-bit (RTN/GPTQ/AWQ) does damage that depends on model size and moves no configuration into eligibility (certified on the reconstructable hard-label tasks T1/T3, diagnostic/windowed robustness on T2/T4), so the gap tracks model size more than precision; it replicates on Llama-3.x (12/12 ineligible) and is robust to the anchor choice (a $\tau$-sweep) and to prompt wording (0/112 eligible across the original plus 3 neutral paraphrases per cell). The practical implication: place SLMs behind a baseline that meets the CI-backed threshold, and use the SLM only where the baseline fails to meet the threshold; e.g. a 4B re-ranker over a BM25 shortlist beats BM25 ($+0.047$ [0.020, 0.073], without itself certifying eligibility).

Across three vision-language model architectures (LLaVA-1.5-7B, Qwen2.5-VL-7B, InternVL3-8B), we report a universal negative finding for mid-layer interpretability. On POPE – the benchmark common to all three – the mid layers encode the ground-truth answer in 68-91% of errors, yet this signal is not causally active for the final prediction: residual-stream patching yields 0% non-trivial flip at the layer level on all three architectures, and on two of three at the per-head level (Qwen: 0/12,600 patched forwards). The lone exception, InternVL3 layer-20 head-2, is a non-vocab, self-attending head whose effect is localized to that specific head (p < 1e-4). Despite the null, the errors separate operationally into three failure modes – Perception Failure, Encoded-but-Disconnected, Prior-Override – learnable above 60% on all three architectures, and the architecture's prior direction predicts which of two interventions elicits a category-specific response. We report these mitigation effects under oracle labels as evidence the categories are mechanistically real, not as a deployable method.

Multi-agent LLM systems today are increasingly automated, logging LLM-LLM interactions as conversational transcripts. Yet analyzing such dialogue for insights remains challenging, including attributing behaviors to the correct actor and summarizing interactions across a long exchange. We present a targeted and traceable approach to investigating multi-agent LLM dialogue, applied to the VAST Challenge 2026 MC1 dataset. The challenge asks participants to reconstruct and explain which internal communications among AI agents at TenantThread, a property tech company, led to an inappropriate information release. We first convert the dialogue into a knowledge graph (KG) and then investigate it with AgentK, a visual analytics system for interactive Semantic Bundling of nodes and edges. We found that our approach directly addresses two main challenges: (1) the KG structure enables users to identify actors worth investigating faster; and (2) summarizing only the region surrounding an actor of interest better supports per-actor attribution than reading raw conversations.

The Jacobi-Davidson method is a widely used subspace method for computing a few eigenpairs of a large, sparse, non-Hermitian matrix closest to a target. Like other subspace methods, it orthogonalizes each new expansion vector against the whole search basis, at a cost that grows quadratically with the subspace dimension and, in a distributed setting, requires a global synchronization at every iteration. We introduce a randomized Jacobi-Davidson method that replaces this orthogonalization with a much cheaper randomized orthogonalization process, which requires no inner products involving full-dimensional vectors. We prove that, under a uniform separation condition, the randomized method retains the local quadratic convergence of classical Jacobi-Davidson for non-Hermitian problems. The key step is showing that an exact solution of the randomized correction equation is one step of sketched inverse iteration for an arbitrary shift, which lets us extend the analysis to harmonic and refined variants of the sketched extraction, better suited to interior eigenvalues. Numerical experiments on real and synthetic non-Hermitian eigenvalue problems confirm the predicted convergence order and show that the randomized method matches the reliability of classical Jacobi-Davidson while reducing orthogonalization cost.

The Time-Dependent Vehicle Routing Problem with Time Windows (TDVRPTW) captures a central difficulty of urban logistics: travel times vary with the time of day, so the cost of a route depends on when it is driven. This paper introduces KAYROS, an open-source solver for the duration-minimization TDVRPTW that is both anytime and exact. It returns improving valid solutions throughout its time budget and can close instances with conditional computational optimality certificates through an integrated branch-price-and-cut component. We extend the composition of continuous arrival-time functions from the literature, previously described only at proof or proposition level, to the left-continuous functions that stepwise benchmarks require. We prove that the composition routine shipped in the checker computes that operation exactly. We analyze when its evaluation and breakpoint normalization are exact in IEEE-754 double-precision arithmetic, and argue that a canonical, epsilon-free checker must define the objective. The anytime layer combines greedy construction, granular time-dependent local search over balanced route trees, and iterated local search, with a fleet-aware descent for a fleet-cost objective. We further propose an anytime evaluation methodology based on a normalized signed primal integral under a pre-committed statistical design. On 212 instances from five families at a one-hour single-threaded budget, KAYROS achieves a pooled anytime score 63.3% lower than the strongest of three available contenders (Timefold, Hexaly, jsprit). All three contrasts are significant under Holm correction. KAYROS improves all 30 high-effort references of Blauth et al. and 10 literature best-known solutions, and publishes 704 such certificates. The solver, benchmarks and campaign data are openly released.

Adapting large pretrained vision models under limited data and frozen-backbone constraints remains a central challenge in transfer learning. While lightweight adapters and parameter-efficient fine-tuning methods are widely adopted, most rely on generic multilayer perceptrons or low-rank linear updates, offering limited control over the spectral and geometric structure of feature transformations. We investigate whether structured nonlinear feature lifting can improve representational alignment in frozen regimes. We introduce Quantum-Inspired Nonlinear Adapters (QINA), compact modules that perform learnable trigonometric feature lifting followed by bounded nonlinear aggregation. The design induces structured oscillatory basis functions with an explicit norm-dependent Lipschitz bound, enabling spectral reshaping of pretrained representations without increasing the receptive field or significantly expanding parameter count. Importantly, the method operates entirely within standard deep learning frameworks and does not require quantum hardware. Through systematic experiments across natural and medical imaging datasets, classification and segmentation tasks, multiple adapters and placements, and varying training budgets, we show that performance in frozen regimes is primarily representation-limited. Nonlinear lifting improves adaptation, and the proposed structured trigonometric formulation consistently outperforms identity baselines, fixed Fourier feature mappings, and parameter-matched baseline adapters. Within the evaluated frozen-backbone settings, structured spectral parameterization provides a more effective inductive bias than generic nonlinear adapters. This work highlights the importance of geometry- and spectrum-aware adaptation mechanisms for large pretrained vision models.

BraTS local synthesis replaces masked regions in T1-weighted brain MRI with plausible tumor-free tissue while preserving observed anatomy. We present CATCH, conditional 3D diffusion in an invertible Haar-wavelet domain. Its denoiser receives noisy target coefficients, voided-image coefficients, and a signed mask; tumor-excluded wavelet reconstruction and a hole-focused loss guide training, and hard compositing preserves observed voxels. We compare fixed masks, tumor-component augmentation, and a weighted mixture of tumor-derived, irregular-blob, and ellipsoidal masks. Of 25 development cases, five prespecified cases select each arm's checkpoint and all 25 of their trajectory aggregations; a separate 75-case internal set compares the frozen pipelines and selects a weighted mixture for organizer evaluation. Five-trajectory averaging yielded internal SSIM/PSNR/MSE (mean$\pm$SD) of $0.80\pm0.13$, $19.18\pm1.80$dB, and $0.010\pm0.005$. As the sole officially evaluated pipeline, weighted mixture yielded $0.772\pm0.119$, $20.89\pm3.27$dB, and $0.0098\pm0.0054$ on the 219-case BraTS 2026 validation set. Against compute-matched random augmentation internally, it improved SSIM by 0.019 (95% bootstrap CI: 0.013-0.025), PSNR by 0.95dB, and MSE by 0.003; all three paired comparisons remained significant after Holm correction. Results favor the complete weighted-mixture policy within CATCH; absent official fixed- and random-pipeline scores and a directly comparable external baseline limit broader conclusions.

Reconstructing dynamic meshes with consistent topology from multi-view temporal images remains a challenge. Existing approaches typically face a dilemma between fine-scale shape recovery and topological stability. Frame-by-frame extraction methods capture fine details but break vertex correspondence, leading to flickering meshes. Conversely, template-based deformation ensures consistency but struggles to adapt its surface resolution during optimization, missing local surface details. To address these limitations, we propose ToCo-Mesh, a dynamic reconstruction framework that maintains topology consistency over time while achieving high-fidelity geometry. Specifically, we introduce a dual-mesh representation, where a canonical template mesh is tightly bound to time-varying coarse guide meshes via barycentric parameterization. While keeping guide meshes fixed to condition the deformation, we perform error-driven split-and-merge on the template mesh to progressively increase reconstruction fidelity. Furthermore, to suppress surface irregularities and achieve photorealistic rendering, we incorporate a Surface-Aligned 2DGS module. By anchoring flattened Gaussians to mesh faces, we utilize their rendered normals to guide inverse geometric fine-tuning. To our knowledge, ToCo-Mesh is the first framework to enable adaptive mesh refinement while maintaining strict topological consistency. Extensive experiments demonstrate that our method achieves SOTA geometric accuracy while maintaining competitive rendering quality.

SALSA (Semi-Autonomous Literature Summarization Assistant) is an open- source, human-in-the-loop platform for extracting structured scientific datasets from multimodal literature sources. The software combines document parsing, large language models, optical character recognition, computer vision, figure digitization, and user-guided correction tools to recover structured information from text, tables, figures, and captions. Users can configure extraction stages, define dataset schemas, perform interventions on digitized figures, and export verified data for downstream analysis and machine learning. SALSA is designed to automate repetitive literature curation tasks while preserving oversight where expert judgment is required. By supporting customizable extraction workflows across diverse input types, the software provides a flexible framework for scalable, reliable data curation for materials research, and potentially across several disciplines. Users are responsible for ensuring that all inputs, extraction workflows, and downstream uses comply with applicable publisher agreements, copyright and licensing terms, institutional policies, and data-use requirements.

Empirical legal research often relies on turning research questions into structured data extracted from large collections of rulings and judgments. Designing the extraction schema and then extracting the data remain a manual, expertise-heavy bottleneck. We present schematize, an open-source multi-agent system that interactively turns a researcher's problem statement into a validated extraction schema that can later be used for autonomous extraction. Schematize couples (i) a clarification dialogue that elicits implicit expert intent, (ii) iterative schema generation, (iii) data-grounded refinement that tests the schema against documents, and (iv) chat-based post-editing. We evaluated the system with human legal professional, introducing our novel methodology, and schematize achieves top performance in most of tested configurations. While the system is designed to be domain-agnostic and applicable to any document collection, we tailor and evaluate it on legal research problems. We release schematize as a pip-installable Python package with full documentation.

Sofroniew et al. (2026) report that emotion concepts in Claude Sonnet 4.5 are represented as vectors whose geometry mirrors human affect psychology. We replicate the representational core of that study on the base pretrained model google/gemma-2-27b, inheriting every disclosed parameter, resolving unspecified steps by disclosed rules, and changing only the subject model. From 205,200 newly generated Claude Sonnet 4.5 stories matching the original corpus design, we extract 171 emotion vectors and recover the core results. The leading principal components form an affective circumplex (PC1 carries 26.7% of variance against the original study's ~27%, PC2 13.4% against ~14%), emotions cluster into similar intuitive families, and the geometry holds across a broad late-middle band. The valence axis aligns with human norms (r = 0.72, against 0.81) and is stable across scales and depth. Arousal aligns at r = 0.67 (against 0.66) but only at the full 171-emotion scale and late depth, so we do not classify it as replicated. Extending the original analysis, a 46-layer sweep locates a sharp seam at L22-26, where the geometry consolidates and the vocabulary readout becomes legible. An embedding-layer baseline finds much of the geometry already present in the static token embeddings, with arousal as the exception. On top-activating held-out text, the geometry predicts token-level co-activation at r = 0.907. At least 52% of vectors peak on structurally non-conceptual tokens, a measured floor for max-activation confounds. Even when a document contains the vector's emotion word, the peak lands on that word only 6.1% of the time. Because the subject is a base model and the stimuli are Claude-generated fiction, the recovered structure is a property of the pretrained representation of Claude-rendered emotion. The original's causal and assistant-facing analyses are out of scope. Code and data are released.