Science & Mathematics

We present PnP-Transformer, a neural pose solver designed as an iterative-free, differentiable alternative to classical PnP-RANSAC algorithms. PnP-Transformer regresses the 6D object pose by processing predicted semi-dense 2D-3D correspondence fields alongside confidence scores, while mapping geometric inputs into a continuous latent space. To evaluate its practical utility, we integrate PnP-Transformer into a zero-shot 6D pose estimation pipeline, where the correspondence module and pose solver are trained separately. PnP-Transformer achieves a 41.2% mean AR score with low latency (0.46 s/obj) on seven core BOP benchmark datasets, surpassing all prior baselines (such as RayPose with 39.1% AR) but one (Co-op with 58.4% AR). When the pose solver is replaced with EPnP-RANSAC (similar to Co-op) while keeping the correspondence module unchanged, the RANSAC alternative outperforms PnP-Transformer on six of the seven datasets by reaching 52.2% mean AR at a cost of an additional 60 ms per object, revealing an accuracy-latency trade-off. Code and pretrained models are available at https://github.com/ebylmz/pnp-transformer.

A game ships when a player can launch it, finish it, and return to a saved world, none of which a successful build guarantees, yet game-creation agents are mostly tested on one engine or on one-prompt browser games. We present CEGame-Bench, the first long-horizon benchmark for complete-game creation across five production engines: Defold, Godot, Roblox Studio, Unity, and Unreal Engine. Agents build each game from an empty project and a design document in runs averaging up to 295 steps and 65M tokens; the document doubles as a contract, and a hidden evaluator accepts a game only by real-input play. Under four scaffolds, agents earn most integration points but usually under half the gameplay points, and their games reach the ending in only 16% of runs, while references built by engineers with coding agents score near the maximum, which places this artifact-to-play gap in the agents. No two engines rank configurations alike (Kendall's tau 0.14-0.62), and the same game usually scores higher on Unity and Unreal Engine, which check more before play. We release Game Completion Boost, an agent skill for requirement-driven playtesting and repair that raises the execution-based score of three models by 14.1-24.3 points.

Distributing quantum circuits across a network of quantum devices is a recent approach to improve the scalability of quantum computation. Entangled pairs of qubits distributed across network nodes are used as a communication resource for executing inter-node (remote) gates. However, establishing these entanglement pairs incurs significant latency, and doing so during the circuit's execution may lead to the decoherence of logical qubits. Instead, it may be beneficial to distribute certain entanglement pairs before the circuit is executed. Thus, we consider the problem of distributing a quantum circuit across a quantum network containing some pre-distributed entanglement pairs and seek to minimize the execution time of the circuit. We address this problem in three steps. First, given the quantum network and (optionally) a probability distribution over training circuits, we determine a set of pre-distributed entanglement pairs under a specified budget. Second, we compute a qubit-to-memory allocation that minimizes the expected execution time given these pre-distributed resources. Finally, we identify any additional entanglement pairs required during execution, generate them concurrently, and execute the circuit gates. We evaluate our methods on a broad range of benchmark circuits.

Answering machine detection (AMD) is a fundamental task in telephony that classifies whether an answered call is picked up by a human or a voicemail (machine). Despite its widespread use across cloud communication platforms, existing approaches rely on proprietary systems with undisclosed architectures, or require computationally expensive models that cannot operate in real time. We propose a lightweight AMD system that combines pretrained audio embeddings with gradient boosting machines, trained to prioritize the correct detection of human-answered calls. We evaluate seven models across two datasets: three models that we propose, a proprietary industry model, two prior academic approaches, and a large language model baseline that relies on automatic speech recognition. Furthermore, we release OpenAMD, the first publicly available AMD dataset comprising of more than 500 recordings in 7 languages. Our best model achieves the highest precision across both datasets while maintaining low inference latency. We assess the robustness of models to additive noise and packet loss, and perform tests of statistical significance to discover per-language and per-gender model performance. Finally, we evaluate our model on an out-of-domain task of detecting robocalls, where it achieves 94.34% recall without any additional training, which is comparable to the industry standard.

LLM agents can turn natural-language financial research requests into executable and backtested strategies, but a successful run may still implement the wrong rule, use information unavailable at decision time, or obscure how repairs and candidate selection changed the strategy. Because these failures arise throughout development, final code and performance metrics alone are insufficient to establish either fidelity to the request or the validity of the development process. We therefore address the reliability of financial strategy development at the trajectory level: the system must control how intermediate outputs become downstream state and preserve the evidence needed to justify the terminal outcome. We introduce QuantHarness, a two-layer agent harness in which a finance-specific Workflow structures specification, data grounding, construction, testing, and review, while Harness Core enforces stage-specific information, action, and commitment boundaries. We also introduce FinRelBench, a benchmark that evaluates strategy-development reliability through execution success, requirement satisfaction, and runtime evidence. QuantHarness achieves the highest Task Success at 90.72\%, compared with 74.23\%--84.54\% for the comparison systems. Controlled stress tests show that its boundaries contain targeted violations, while structured run records improve evidence retrieval and support decision replay under the evaluated protocol.

Adversarial vulnerability persists across modern vision architectures from CNNs to vision language models (VLMs), yet existing detection methods rely on heuristics without theoretical guarantees. We address the fundamental question of when adversarial perturbations can be provably detected from a geometric perspective. Our key insight is that adversarial perturbations cannot simultaneously preserve geometric structure across spaces with fundamentally different properties. Accordingly, we construct two such complementary metric spaces. First, we use a standard CNN embedding space $Z$, where adversarial samples exhibit significant displacement patterns. Second, we build a novel prime-quantized space $P$, that absorbs small perturbations through number-theoretic discretization, resulting in minimal displacement, while preserving discriminability. We then leverage the geometric discrepancies across spaces $Z$ and $P$ to detect adversarial samples. To the best of our knowledge, we establish the first rigorous separation theory for adversarial detection, proving that adversarial samples create unavoidable geometric inconsistencies across both spaces. Our framework provides theoretical guarantees including pixel-level absorption bounds, neighborhood diameter concentration, Gromov-Wasserstein (GW) separation theorems, and practical risk control. Extensive experiments validate our theoretical predictions and achieve consistently strong detection performance across a wide range of attack types and model families.

Classical inverse bandit methods typically aim to recover true reward parameters. In contrast, we infer a demonstrator's subjective beliefs, specifically its internal estimates of suboptimality gaps. We consider no-regret demonstrators that follow randomized policies, such as Thompson sampling, and characterize a family of algorithms for which per-step inversion is possible. This enables gap estimates to be recovered directly from the arm-selection distribution. Our analysis identifies an inherent stability-coverage trade-off: inversion becomes ill-conditioned as the demonstrator concentrates its probability mass on a single arm, while the number of informative rounds depends on the rate at which its policy converges. We establish conditions under which inversion is reliable and provide theoretical guarantees on the resulting estimation error. Experiments on simulated data support the analysis and illustrate how these conditions affect gap recovery in practice.

Graph Neural Networks (GNNs) have proven effective for graph analytical tasks, yet their "black-box" nature complicates the understanding of their layer-wise behavior and hinders their application to large graphs. We demonstrate SliceGXQ, an end-to-end declarative and interactive system for querying and exploring layer-wise explanations for GNNs over large-scale graphs. In particular, SliceGXQ provides (1) an interactive graphical user interface for dataset, model, and explanation configuration; (2) a lightweight SPARQL-like parser that allows users to specify target nodes, layers, explanation modes, and constraints; and (3) an integrated execution and visualization workflow that invokes SliceGX as its underlying explanation engine. We demonstrate how users can diagnose a misclassified node, identify problematic layers and influential messages, and perform targeted model debugging, together with large-scale graph interpretation and fraud-detection analysis. Our demonstration video and source code are available at https://github.com/jinzhaoqet/SliceGXQ.

Generative retrieval assigns each item a discrete semantic identifier (SID) and retrieves items by autoregressively generating SID sequences. In dynamic environments such as e-commerce, new items arrive continuously, requiring periodic updates to the SID-to-Item mapping and model to maintain retrieval timeliness. Each update, however, can invalidate search paths already learned by the retriever, and the quantitative relationship between mapping change and retrieval quality has not been systematically characterized. We study this problem with extensive controlled experiments over 67 days of large-scale e-commerce search logs and obtain three core findings. (1) SID semantic drift causes severe retrieval degradation that may be masked by large beam sizes: mapping updates reduce HitRate by 41.78-62.05% at beam size 1 but only 8.33-10.85% at beam size 1000, as redundant candidate generation conceals the true severity of drift. (2) The fraction of changed SIDs is strictly monotonically related to retrieval quality: as the remapped fraction increases from 0% to 72.75%, HitRate decreases monotonically, dropping up to 42.66% at beam size 1. Full remapping remains substantially worse than the preservation policy under warm-start training, indicating that short-budget continued training alone cannot repair topology disruption. (3) The SID topology-preserving principle makes even the simplest warm-start strategy highly effective: fixing old-item SIDs, inserting new-items, and warm-starting from the previous model improves old-item HitRate by 10% and new-item HitRate by 109%, validated across two non-overlapping update cycles. These findings distill into a concise deployment principle: insert new-items, freeze old-items; warm-start when possible, otherwise freeze; never change SIDs the model has already learned. A three-week online A/B test on TmallAPP shows UCTCVR +0.29% and GMV +1.16%.

We present a complete, GPU-resident pipeline for multi-robot 3D inspection of underwater structures. Given a triangle mesh and a fleet of $K$ robots, each carrying a short-range, limited field-of-view camera, the pipeline computes 6-DOF trajectories that jointly cover a target fraction of the surface while minimizing a linearly interpolated makespan/total-cost objective. The joint problem is NP-hard, so we decompose it into five core stages - candidate viewpoint generation, visibility computation, set cover, min-max vehicle routing solved as a mixed integer program (MIP) with lifted Miller-Tucker-Zemlin subtour elimination, and priority-based multi-agent path finding with continuous-time inter-robot conflict checking - flanked by mesh loading and surface sampling as preprocessing and by trajectory smoothing as postprocessing. Our contributions are: an end-to-end GPU-accelerated integration of the five stages, with its kernel design documented; a GPU batch adaptation of Lien's $\varepsilon$-visibility with four custom CUDA kernels; a $β$-aware per-vehicle tour upper bound derived from the interpolated objective, supplementing the Desrochers-Laporte coefficient lift with forbidden-pair cuts; a 4D space-time extension of Zhou and Zeng's GPU parallel-frontier A*; and an experimental evaluation on 14 meshes (up to $7.2$M faces) that ablates the routing cuts and the parallel-frontier search, audits collision freedom and waypoint attainment \emph{after} trajectory smoothing, and measures the end-to-end gap against analytical and LP-relaxation lower bounds. Code and raw results are publicly available.

Audio fingerprinting methods identify audio based on its content, even when the content has been possibly corrupted. Despite their applicability and usage in the speech domain, research and advancements in audio fingerprinting traditionally focus on music and are almost exclusively evaluated on music. We provide insights into speech-oriented audio fingerprinting by benchmarking the best performing audio fingerprinting models on speech data while taking into consideration the effects of noise, reverberation, temporal changes, pitch modifications, and lossy compression. Our findings indicate that audio fingerprinting methods based on self-supervised learning outperform classical approaches on speech data, while the performance of these models on speech is inferior compared to the performance that they achieve on music. Results also include an analysis of false positives, showing that fingerprinting methods are more likely to match with an audio of the same speaker rather than an audio of the same text, when mismatches occur.

Time series foundation models (TSFMs) have achieved notable progress in classification tasks through large-scale pretraining. However, their post-training practice mainly focuses on implicit parameter fine-tuning, where task-relevant information is entangled with abundant irrelevant knowledge in vast parameter spaces. This may hinder TSFMs from precisely activating useful knowledge for downstream classification. To this end, we propose Retrieval-to-Recognition (R2R), a plug-and-play post-training paradigm that adapts pretrained TSFMs to time series classification. Rather than pretraining a new foundation model, R2R improves how existing TSFMs are post-trained: it introduces a decoupled query encoder to extract discriminative patterns and retrieve instance-level numerical knowledge from a constructed knowledge base. The retrieved explicit knowledge is then fused with the TSFM's implicit parametric embeddings, enabling joint knowledge transfer at both the parameter and feature levels. In this way, R2R allows TSFMs to recall task-relevant patterns learned during pretraining, improving the efficiency of knowledge utilization in downstream classification transfer. Experiments on 158 time series datasets show that R2R consistently improves the classification performance of state-of-the-art TSFMs. Ablation and knowledge-enhanced studies further validate the effectiveness of the proposed framework.

Low reward variance reduces estimation cost, but does not eliminate monitoring of arms that may improve while unobserved. We refine the variance--variation tradeoff and characterize how both costs depend on supplied variance and variation budgets in bounded, non-stationary multi-armed bandits. Let $A$ be the number of arms, $T$ the horizon, $B$ a bound on consecutive mean variation excluding the initial vector, and $Q$ a bound on the sum of the largest per-round reward variances. For every admissible pair $Q,B$, the minimax expected dynamic regret is, within a single logarithmic factor, \[ \min\{T,\;A+\sqrt{AQ}+(AQBT)^{1/3} +\sqrt{AT}\min\{B,\sqrt B\}\}. \] The deterministic boundary $Q=0$ is characterized without logarithms. For small positive budgets, a safe calibration phase identifies a common interval containing all future means of surviving arms. Clipping observations to a wider interval reduces the cost of restarting a prior hybrid static-regret algorithm; an original-mean analytical reference controls its residuals without charging the variation of clipped means. Balancing clipping bias and monitoring closes the small-budget noisy regime. Fixed oblivious constructions give the matching lower bound. We also quantify adaptation costs: on a disclosed deterministic one-improvement family, a stationary-regret budget has a matching worst-case change-regret frontier, including a finite improvement pulse. This implies an $Ω((T/A)^{1/4})$ loss when one policy competes with all supplied-budget deterministic oracle rates. The budget-free low-noise scheduling problem remains separate.

Multi-teacher on-policy distillation (MOPD) aggregates the capabilities of multiple domain-expert teachers into a single student model. However, all domains in MOPD share the student's finite budget of optimization steps and rollouts. Fixed uses constant domain sampling ratios based on corpus sizes; by contrast, D$^3$-MOPD adjusts the ratios using the teacher-student reverse-KL divergence and its recent decrease. However, favoring large recent decreases can direct more training toward domains that already improve quickly, while domains that improve more slowly may receive too few examples. To address inflexible allocation and the risk of giving insufficient training to domains that improve more slowly, we propose Nash-MOPD, an adaptive domain scheduler based on a local minimax game using existing reverse-KL statistics. The equilibrium maximizes the smallest predicted proportional reduction under the local model and gives more prompts to domains that need more examples for comparable improvement. Consequently, Nash-MOPD responds to changing learning progress and supports domains that improve more slowly without extra teacher queries or rollouts. At step 100, Nash-MOPD exceeds both baselines' final mathematics and code checkpoint scores on DeepSeek-R1-Distill-Qwen-1.5B and reaches Fixed's on Qwen3-8B-Base. Furthermore, it leads eight of nine reported metrics and is best or tied best on all six mathematics and code accuracy metrics, respectively.

Part I [arXiv:2606.30202] and Part II establish which trust-region radius update mechanisms can fail and under which hypotheses they must succeed. Neither says where the thresholds sit once a rule runs with real constants, or how the rule interacts with the model Hessian and the subproblem solver. This paper turns them into nine falsifiable claims and settles what it can, on four designed problems where the theory's quantities are computable and on CUTEst, where only aggregate cost is visible. Every mechanism runs in one Julia package written for this study, so two runs differing in one component differ in nothing else. The rule is worth less than the model and more than nothing. Crossing the mechanisms with four model Hessians puts $96.5$ per cent of the variation on the model and $1.6$ on the rule, and the rules separate most under the most accurate model rather than least. Uncapped R-grad is the robust default among the criticality-anchored rules, though the retrospective rule is ahead of it on this collection. No constant escapes the side conditions. For every cap there is a minimiser at which the constraint stays active, and for every anchoring constant a flatness on which escape takes arbitrarily long. The second of these is proved here as well as measured, and the proof uses no property of the model or of the subproblem solution beyond $\|s_k\| \leq Δ_k$. The radius rule governs the length of the step, and so the cost of a run. The model decides the search direction, and with it whether the run converges at all.