YOUR RESEARCH, IN SIGHT
覆盖计算机全学科，以 AI 为快捷入口。关注方向、作者和关键词，把发现变成持续的研究积累。
时间范围跟随排序：版本更新、本站收录或首次发表。最新收录可能包含历史论文。部分信源异常，查看状态 →
Johannes Fischer
Given a string $T[1..n]$, its substring complexity is defined as $δ=\max_k d_k/k$, where $d_k$ is the number of its distinct substrings of length $k$ in $T$. We show how to compute it in $O(n+r\log r)$ time, using only $O(r)$ words of space, where $r$ denotes the number of runs in the Burrows-Wheeler transform.
cs.DS
Yunlong Wang, Huizhe Zhang, Haonan Hu, Yudong Li, Bing Wen, Jianchao Tu, Chengxiang Zhuo, Zang Li
Industrial candidate-ranking systems must combine long user behavior histories with heterogeneous user and item features under strict computational budgets. We propose CCFormer, a structured interaction architecture that separates cross-field retrieval from intra-history modeling. Feature-field-separated cross-attention allows user and target queries to read an evolving history hierarchy, while local subspace token mixing contextualizes history tokens independently of candidate targets. Hierarchical sequence compression reduces the number of history tokens and provides multi-scale historical context for subsequent interactions. Experiments on two public benchmarks show higher mean AUC than the evaluated baselines, while experiments on a large-scale industrial dataset show a favorable accuracy-compute trade-off and higher training throughput than HSTU. Online A/B tests across four recommendation and advertising scenarios report system-level improvements of up to 3.57% in click-through rate, 3.86% in page views, and 1.71% in advertising revenue relative to the corresponding production baselines. CCFormer has been fully deployed in all four scenarios, demonstrating its practical scalability and effectiveness in large-scale recommendation systems.
cs.IR
Yuze Li, Fuchen Long, Xinlong Sun, Cheng Yan, Xinyu Zhang, Xi Chen
Compositional motion transfer requires a model to extract motions from multiple reference videos, and apply each motion to a user-specified object in one target scene. Existing methods normally struggle to reliably assign different reference motions to the correct target objects, thus a motion intended for one object may affect another. Meanwhile, some motion-specific tuning approaches are still facing another issue: they must run a new optimization whenever the reference video changes, making repeated transfers inefficient. To address these limitations, we propose RART, a framework that unifies object-wise motion extraction, target-region anchoring, and frame-aligned injection into a frozen video Diffusion Transformer for multi-object motion transfer. For each reference motion, RART extracts temporally ordered features from the masked reference video, and then conditions them on the target first-frame mask to form mask-anchored motion tokens. Frame-aligned motion cross-attention allows each generated frame to read only the motion tokens at the corresponding time slots, composing all object motions in one denoising process. During training, a motion-attention routing loss encourages each target region to attend to its corresponding motion tokens and reduces interference between objects. Therefore, RART can handle novel reference-target combinations without per-video optimization. Experiments show that RART achieves accurate motion preservation, efficient inference, and strong compositional generalization.
cs.CV
Dongyan Sun, Neelesh Gupta, Rajgopal Kannan, Viktor K. Prasanna
Graph analytics generates irregular, topology-dependent memory-access streams via indirect traversals, limiting data prefetching. Accurate prediction enables timely prefetches that hide memory latency. However, ML predictors couple short histories with fixed horizons, fitting the training trace's node layout rather than learning transferable graph patterns. We propose LoCoMAP, a long-context representation-learning framework for memory-access prediction with independently controlled observation history and prediction horizon. We evaluate Connected Components (CC), PageRank (PR), and Single-Source Shortest Paths (SSSP) on six synthetic graph families, with Erdos-Renyi (ER) and Grid as random and deterministic controls. We test two interventions: same-family regeneration changes connectivity, while node shuffling disrupts layout but preserves connectivity. Across 18 graph-workload pairs, extending history from 8 to 128 accesses improves F1 by 7.8%, while extending prediction horizon from 8 to 512 improves F1 by 88.3%. For CC, regeneration is less disruptive than shuffling: without retraining, long context retains 95.4% of original F1 and outperforms short context in six families. Under full shuffle, long context improves CC F1 by 33.6% across ER, Barabasi-Albert, Copying Model, and Stochastic Block Model (SBM), with a 2.06x gain on SBM, but not on Watts-Strogatz or Grid, where near-zero delta mass shifts into heavy tails beyond the predictor's range. Long context outperforms short context on six real-world CC graphs. At full shuffle, long-context PR retains 57.6% of original F1 and averages 11.0% above short context. Both SSSP models retain about 30% of original F1 under full shuffle as distance updates access vertex data in a new order. Within these bounds, long-context predictors are robust across related graphs and layouts without retraining, enabling a generalizable graph prefetcher.
cs.LG
Sanjay Suresh, Caitao Zhan, Joaquin Chung, Rajkumar Kettimuthu
Distributed quantum computing (DQC) connects multiple quantum processing units to overcome the scale limits of monolithic hardware. We introduce SeQUeNCe.DQC, a new top-level module within the SeQUeNCe package that compiles monolithic circuits into topology-aware execution plans and coordinates their execution using SeQUeNCe's existing quantum-network modules and protocols. We validate SeQUeNCe.DQC by reproducing a two-node trapped-ion Grover experiment in simulation and comparing success probabilities with the reported 71% benchmark. We then evaluate how compiler strategy, topology, and memory capacity jointly govern distributed circuit execution time and success rate. In these experiments, our hybrid compilers successfully identified mixed TeleGate/TeleData execution plans that outperformed pure gate-teleportation strategies. By capturing the physical overhead of distributed quantum primitives, SeQUeNCe.DQC provides a rigorous platform for co-designing DQC software and hardware architectures.
cs.NI · cs.DC · quant-ph
Sitong Shen, Chenshu Wu
Effective coral reef protection and restoration demands continuous, long-term underwater monitoring. However, current solutions largely rely on fragmented, manual labor and costly infrastructure, or are severely limited by constrained battery lifespans. In this paper, we present CoraLive, an ultra-low-power underwater camera system for long-term, continuous coral reef monitoring. CoraLive presents a decoupled underwater-shore co-design that minimizes on-board computation and storage, offloading heavy reconstruction and segmentation tasks to offline shore-side processing. To achieve this, CoraLive introduces a dual-rate sensing scheme combining frequent low-resolution frames with sparse high-fidelity samples and a content-adaptive update mechanism that extracts and stores only compact, task-specific data, eliminating the need to save raw video frames. We implement CoraLive on a custom STM32U575-based dual-camera prototype with device-level computation and storage optimization. Compared with raw recording, CoraLive reduces daily storage by $4.51\times$ and average power by $7.78\times$, extending battery lifetime from 20.4 to 159.5~days while retaining high-accuracy segmentation and reconstruction. Further validated in a real-world underwater deployment, we envision CoraLive as a pioneering step for ubiquitous long-term coral reef underwater monitoring.
cs.NI
Ji Dai, Zhipeng Wei, Zhuang Zhuang, Rongfeng Guo, Jie Chen, Quan Fang, Fei Pan
Generative recommenders predict an item's semantic identifier (SID) one code at a time. Early codes select broad branches, so a routing error can exclude relevant items before fine-grained prediction begins. In multimodal recommendation, these coarse decisions must reflect collaborative preferences as well as item semantics. We introduce RoFiGR, a two-tree recommender that uses collaborative information in both SID construction and generation. It aligns cumulative L1-L2 code vectors with frozen collaborative item embeddings and explicitly encodes item distance and source level in the serialized history. During generation, a sequential teacher supervises legal branches. Fixed coarse-level weights maintain routing supervision, while a bounded adaptive residual allocates additional weight according to teacher-student demand. On Instruments, Arts, and Games, RoFiGR improves all five ranking metrics over the evaluated external baselines. Five-seed paired tests with Holm correction support the main NDCG@10 improvements. Comparisons with alternative collaborative supervision objectives and tokenizers support the proposed design. Matched task and decoding analyses distinguish the effects of training exposure, stream fusion, and candidate budgets.
cs.IR
Declan J. Curran, Sanaa Hobeichi, Hira Saleem, Hao Xue, Flora D. Salim
Downscaling is essential for generating the high-resolution climate data needed for local planning, but traditional methods remain computationally demanding. Recent years have seen impressive results from AI downscaling models, particularly diffusion models, which have attracted attention due to their ability to generate ensembles and overcome the smoothing problem common in other AI methods. However, these models typically remain computationally intensive. We introduce a Hierarchical Diffusion Downscaling (HDD) model, which introduces an easily-extensible hierarchical sampling process to the diffusion framework. A coarse-to-fine hierarchy is imposed via a simple downsampling scheme. HDD achieves competitive accuracy on ERA5 reanalysis datasets and CMIP6 models, significantly reducing computational load by running on up to half as many pixels with competitive results. Additionally, a single model trained at 0.25° resolution transfers seamlessly across multiple CMIP6 models with much coarser resolution. HDD thus offers a lightweight alternative for probabilistic climate downscaling, facilitating affordable large-ensemble high-resolution climate projections. See a full code implementation at: https://github.com/HDD-Hierarchical-Diffusion-Downscaling/HDD-Hierarchical-Diffusion-Downscaling.
cs.CV
Yifei Sun
Neural operators are widely used as fast surrogates for numerical PDE solvers, mapping input functions to solution functions. However, their generalizability and robustness are not yet clearly defined in the operator-learning setting, which differs from traditional adversarial robustness definitions. This paper studies the generalizability and robustness of a learned neural operator from a solver-integrated perspective, addressing the challenge that the output of a learned operator and a numerical solver tends to change in tandem under input perturbation. First, we formalize the definition of generalization and robustness through a model-solver error operator, identifying fixed-input model-solver loss as generalization metric, and norm-bounded adversarial attack loss increase and Jacobian-error function norm as robustness metric. Second, we identify the solver-integrated adversarial attack as appropriate for PDE operator learning and show why model-only or fixed-ground-truth attacks can be insufficient when the solver output also changes with the input. Third, we develop solver-integrated adversarial training methods for neural operators. Experiments on representative PDE benchmarks show that this solver-integrated adversarial training clearly improves both generalizability and robustness. Deeper solver integration yields more effective attacks, more informative samples, and more efficient training than less integrated alternatives. These results provide a general framework for robust operator training and automatic sample selection without heavy manual intervention. More broadly, the formulation applies to adversarial regression whenever a ground-truth oracle can evaluate, and ideally differentiate, the true input-output map; PDE operator learning is one such case.
cs.LG
Mohammed Nayeem Khan, K S Sarun Kumar
AI coding agents now do long, multi-step and multi-session software work with little supervision. When a developer comes back to that work, the useful question is not what the model said but what was actually done, by whom, and whether it is finished. We study this in three parts. First, we verify Agenttrail, an open-source local observability tool for coding agents, against its source at commit e4ba2da: nine documented constants and mechanisms hold at file-and-line level, two event pipelines are never unified, and its Map service has an unauthenticated local mutation surface. Second, we run a prototype of our system on 45 real coding-agent sessions (102,233 records, 20,577 tool executions, 2,336 human-to-agent episodes). With a blind-audited classifier (precision 83.3% and 96.7%) we estimate that 23.4% (95% CI 17.1 to 35.7) of code-editing episodes end with no independent check after the last edit, and that 57.8% of to-dos the agent marked complete after editing had no proof in between. A 10-minute timing heuristic for session hand-offs recovers 6 of the 24 session pairs that content lineage links. 42% of sessions run longer than 24 hours, 48.1% of transcript bytes lie past a 2 MiB replay bound, and 1,326 records in 16 sessions contain credential-like strings. Third, we propose Soei (Structured Observability of Evidence and Intent): an acknowledgement-aware event schema, a hash-chained local ledger, a W3C PROV-aligned graph, calibrated inference for attribution and lineage, and a verification layer that keeps a claim of completion apart from proof. On the same corpus the ledger is 20% of raw size, holds no content or credentials, verifies at about 126k events/s, detects tampering and replays deterministically. We position the work against concurrent 2026 proposals and define autonomy levels with evidence gates and harness metrics baselined on our data.
cs.SE · cs.AI · cs.MA
Yanjie Pan, Qingdong He, Zhengkai Jiang, Pengcheng Xu, Chaoyi Wang, Jinlong Peng, Haoxuan Wang, Yun Cao 等 12 位作者
Recent advances in diffusion-based text-to-image generation have demonstrated promising results through visual condition control. However, existing ControlNet-like methods struggle with compositional visual conditioning - simultaneously preserving semantic fidelity across multiple heterogeneous control signals while maintaining high visual quality, where they employ separate control branches that often introduce conflicting guidance during the denoising process, leading to structural distortions and artifacts in generated images. To address this issue, we present PixelPonder, a novel unified control framework, which allows for effective control of multiple visual conditions under a single control structure. Specifically, we design a patch-level adaptive condition selection mechanism that dynamically prioritizes spatially relevant control signals at the sub-region level, enabling precise local guidance without global interference. Additionally, a time-aware control injection scheme is deployed to modulate condition influence according to denoising timesteps, progressively transitioning from structural preservation to texture refinement and fully utilizing the control information from different categories to promote more harmonious image generation. Extensive experiments demonstrate that PixelPonder surpasses previous methods across different benchmark datasets, showing superior improvement in spatial alignment accuracy while maintaining high textual semantic consistency.
cs.CV
Zimeng Wu, Qijian Tian, Penghui Du, Changqian Yu, Jiaxin Chen, Yunhong Wang
Unified Multimodal Models (UMMs) enable powerful multi-image synthesis but incur substantial computation from long visual contexts. While token pruning offers a direct route to accelerating inference, existing methods are primarily adapted from understanding tasks, suffering from insufficient reference image exploration, confused conditional guidance, and pooling-induced ambiguity that degrades generation quality. To address these issues, we propose CATP, the first training-free context-compression method tailored to UMM generation, which \textbf{C}alibrates the discrepancies of general \textbf{A}ttention-based \textbf{T}oken \textbf{P}runing for better quality-efficiency trade-off. Concretely, building on independent layer-wise pruning during denoising, we introduce Guidance-Consistent Pruning (GCP) to synchronize pruning decisions across conditioning branches, and a Taylor-Calibrated Attention Metric (TCAM) that derives a theoretically grounded second-order correction for pooling-induced bias. In addition, a Fast Scoring Operator (FSO) fuses the calibration into tiled online Softmax to reduce pruning cost. Extensive experiments across representative UMMs and benchmarks demonstrate that CATP consistently outperforms state-of-the-art token-reduction methods, achieving up to 2.09$\times$ attention-core speedup.
cs.CV
Xun Wei
In many real-world classification tasks, datasets exhibit class imbalance, which can degrade the generalization performance of conventional learning algorithms. Data complexity has recently been recognized as a more objective indicator of learning difficulty than the imbalance ratio alone, because class overlap and complex decision boundaries often impair classifiers more severely than imbalance itself. Samples near the decision boundary are inherently difficult to classify, which has motivated many researchers to focus on such instances to improve classifier performance. However, most under-sampling approaches fail to differentiate their sampling of boundary samples across datasets of varying complexity. To overcome this limitation, we propose a complexity-adaptive boundary-aware ensemble (CBE) framework that links the sampling weight distribution to data complexity and dynamically adjusts the boundary preference of under-sampling accordingly. We also introduce an overall complexity measure that combines metrics at multiple levels and use it to divide imbalanced datasets into three categories. Extensive experiments on 24 benchmark datasets from UCI and KEEL show that the proposed method outperforms many existing methods in terms of F-measure and G-mean on Easy and Median datasets, without degrading performance on hard cases.
cs.LG
Ryo Takizawa, Tempei Kabayama, Chiemi Yamazaki, Shun Kitamura, Ryo Matsuoka, Yuta Ando, Satoshi Kodera, Norihiko Takeda
View- and target-specific echocardiography AI models capture useful but complementary information and may not achieve sufficient predictive performance in isolation. A learned ensemble is a natural way to combine these outputs, but heterogeneous models typically require each output to be converted into a binary decision, probability, confidence score, or fixed-length feature vector. We developed a supervised large language model (LLM)-based late-fusion module that directly aggregates view-specific predicted finding statements and predicted measurements in language space and learns a case-level fusion function. In patient-level three-fold cross-validation of 859 examinations from 646 patients, the proposed method integrated heterogeneous outputs within a single framework, performed comparably to the best-performing reference strategy evaluated, and outperformed a general-purpose LLM without task-specific training. These results suggest that a supervised LLM can combine multiple upstream models without model-specific output conversion or feature engineering that requires medical knowledge.
cs.CL · cs.AI
Yikun Hu, Zichen Zhao, Peixiang Qin, Ziyi Zhou, Jiaping Gui, Yuandao Cai, Wensheng Tang
Deep neural networks (DNNs) have become a foundational component of modern computing systems with a wide range of applications, such as computer vision, edge intelligence, etc. For the sake of low latency and data privacy, DNN models are increasingly compiled into executables and deployed on local devices. However, that exposes the models to model theft, enabling adversaries to recover proprietary assets via reverse engineering techniques. While code obfuscation naturally emerges for protecting executables from reverse engineering, existing schemes are primarily designed for traditional programs, focusing on complex control flow structures and integer-based operations. They are fundamentally inappropriate for DNN binaries, which exhibit relatively simple code structures and heavily rely on floating-point computation. In this paper, we propose FLOB, an obfuscation framework that protects floating-point computations in DNN binaries using Mixed Boolean-Arithmetic (MBA) transformations. Our approach lifts floating-point values into a higher-precision binary expansion space and performs MBA transformations at the representation level. The computation is carried out entirely in the lifted space, and the final result is then projected back to the target precision, yielding outputs that are semantically equivalent under the original floating-point format, without introducing additional rounding error. The experimental results show that FLOB outperforms the state-of-the-art obfuscation schemes against existing reverse engineering analyzers and deobfuscators, while preserving computational correctness without introducing additional numerical error, allowing flexible trade-off between protection and performance. Specifically, FLOB reduces the operator recovery rate to 4.51% on average, achieving a 32.82 percentage-point reduction compared to existing methods.
cs.CR
Anand Kumar, Aditya Shekhar Nittala
In many real-world scenarios, users interact with digital systems while their hands are partially occupied, such as when holding tools or everyday objects. Prior work on busy-hand interaction has focused on microgestures and motion-based techniques, which rely on finger articulation or relative movement. In situations where a stable grasp is required, such as handling fragile, heavy, or precision objects, such motion can be undesirable or impractical. Pressure-based input offers a complementary modality that enables interaction without changing finger pose, yet its use while holding objects has received limited empirical attention. To explore this, we conducted two within-subject experiments examining users' ability to perform controlled pressure selections and analyzing finger-specific performance across different grasps and geometries. Results indicate that participants can reliably perform pressure-based selections in these conditions, achieving high success rates (92.7% on average) with low selection times (1.74s on average). Our findings characterize the potential of pressure input as a complementary microinteraction modality under grasp-constrained, busy-hand scenarios.
cs.HC
Fail Gafarov
We study coordinated groups of wallets on a bonding-curve launchpad (pump.fun on Solana), using a complete record of 200\,443 token launches, 15\,854\,121 curve trades and 51\,025\,958 SOL transfers over 88.3 hours. A prospective, multi-channel discovery procedure (co-buying, shared funding, joint signing, transfers, consolidation) yields 23\,710 wallet groups. Without ground-truth labels we validate them by out-of-sample link prediction, independent non-trading links and a comparison with naive co-buy clustering at equal numbers of flagged tokens: raw clustering collapses into one giant component, yet it ranks tokens at least as well as the groups, so the value of the structure is descriptive. Group activity is highly concentrated and splits into serial creators and long-lived rings of early buyers. Known members buy earlier than other buyers of the same token and extract more cash than an activity-matched placebo set, while late buyers' net cash flow is negative; tokens with members show a pump-like price path, but the contrast vanishes when controls are restricted to comparable tokens, so we report it as an association. A follower entering at the earliest group signal loses money out of sample in all 32 configurations and does no better than random entry. A few pools of tens to over a hundred fast wallets are associated with a two- to four-fold higher chance of reaching 20 SOL, also within creators, with a smaller excess at 60 SOL and none for graduation; on a final hold-out the team rules keep a precision of about one half to three quarters, yet many signals arrive when the curve already holds 20 SOL, one team supplies most of them, and they add no information beyond early curve state. We also document evaluation errors that turn the same data into apparent profit.
physics.soc-ph · cs.SI · q-fin.TR
Mengxiao Zhang
We study first-order and best-of-both-worlds regret guarantees in multi-armed bandits. Pogodin and Lattimore (2019) proposed an algorithm that combines minimax-optimal worst-case regret with a first-order guarantee whose loss-dependent term is $\mathcal{O}(\sqrt{K L^\star}\log T)$, where $T$ denotes the horizon, $L^\star$ denotes the smallest expected cumulative loss among fixed arms, and $K$ denotes the number of arms. However, a logarithmic gap remains between this loss-dependent term and the optimal first-order scale. In this paper, we remove this logarithmic factor by proposing an efficient and parameter-free algorithm that achieves the minimax optimal first-order regret guarantee $\mathcal{O}(\sqrt{K L^\star})$. Technically, our algorithm is built upon FTRL with a Tsallis entropy regularizer (Zimmert and Seldin,2021), with a learning rate adapted to the learner's cumulative loss and a sampling distribution determined by a scalar fixed point. This construction controls the combined estimation and stability cost in terms of the learner's loss. Moreover, we show that the same algorithm achieves a best-of-both-worlds guarantee, with $\mathcal{O}(\sum_{i\ne i^\star}\frac{\log T}{Δ_i})$ regret for i.i.d. losses with a unique best arm $i^\star$, where $Δ_i$ denotes the suboptimality gap of arm $i$.
cs.LG
Zhaoxin Wang, Zihang Ding, Handing Wang
Deep neural networks achieve strong performance across many tasks but remain vulnerable to adversarial perturbations. Adversarial training (AT) is one of the most effective defenses, yet it suffers from high memory cost on large models and the persistent clean-robust accuracy trade-off. Recent work introduces parameter-efficient fine-tuning, such as LoRA, into AT, but this constraint induces a notable robustness gap relative to full-parameter AT. In this work, we revisit parameter-efficient adversarial fine-tuning from a parameter space perspective. Through gradient analysis, we find that adversarial optimization is highly concentrated on a small subset of parameters, which we call Robustness-Critical Parameters (RCPs), suggesting that robustness is encoded in a sparse subspace. Building on this observation, we propose Sparse Adversarial Fine-Tuning (SAFT), which identifies RCPs via adversarial gradient saliency and performs AT by updating only these parameters while freezing the rest. Across architectures and datasets, SAFT outperforms LoRA-based methods in both clean and robust accuracy with fewer trainable parameters, and approaches or surpasses full parameter AT in robustness while updating only about 5\% of parameters.
cs.LG · cs.AI
Yilong Li, Chengpo Yan, Yijing Zeng, Hanlei Wu, Mingjun Song, Suman Banerjee
Edge SoCs combine CPUs, GPUs, and NPUs, yet runtimes for multimodal large language models (MLLMs) keep generation on one engine and leave others idle. Streaming workloads allow ingestion and answering to overlap, but backend-local KV state binds queries to the engine that built their history. Stage performance and co-running interference vary, so no single placement wins across SoCs and workloads. \name separates logical history from engine-specific KV state so any language engine can continue a stream. Stateful Streaming Generation overlaps ingestion and answering, while Concurrent Sequence Generation serves ready queries in parallel. Phase-level state continuation lets one engine prefill a query and another decode the answer, freeing the first to resume ingestion. State-Driven Execution Planning (SDEP) chooses paths by predicted completion time and sets each engine's share of queries using arrival rates and measured service times. Across M4 Max, RK3588, and QCS6490, \name beats the strongest single-engine baselines, even when they reuse context. On M4 Max, SDEP cuts live-stream p95 time to first token by $76\%$, with every first token within one second, and raises throughput under backlog by up to $35\%$ over Metal-only execution with context reuse. On two RK3588 streams, SDEP delivers $19.6\%$ higher request throughput than the best single engine with context reuse and $27.2\%$ more than parallel engines without shared state. Stateful Streaming Generation reaches up to $2.48\times$ serial throughput on the same engines. Every answer matches its serial reference.
cs.LG
Chenyan Wu, Yandong Li, Xianfeng Tang, Zhe Huang, James Wang
Reconstructing multi-human body meshes from a single monocular image is a challenging problem, requiring both individual body meshes and relative 3D positions to produce a coherent scene representation. We introduce MUG (Multi-hUman Graph network), which uses a single graph neural network to construct coherent multi-human meshes from only 2D poses. Our approach demonstrates that pose-based methods can jointly perform depth reasoning and multi-human mesh generation. Existing image-based methods often rely on lab-collected datasets with accurate 3D labels, introducing domain gaps when applied to in-the-wild or art images due to differences in appearance and context. In contrast, MUG leverages the geometric consistency of 2D poses across diverse datasets to mitigate such discrepancies. MUG operates in three phases. First, it constructs a novel heterogeneous graph from multi-human 2D poses, connecting nodes across people and within individuals to capture inter-human interactions and body geometry, including skeletal and mesh structures. Second, a dual-branch graph neural network predicts inter-human depth relations and root-joint-relative mesh coordinates. Finally, these outputs are combined to construct complete multi-human 3D meshes. Despite using only 2D poses and a streamlined architecture, MUG outperforms existing multi-human mesh estimation methods across datasets from diverse domains, including RH, MuPoTS-3D, and 3DPW. Qualitative results show that MUG can handle art images, through-wall scenarios, and poor lighting conditions when paired with advanced 2D pose networks. Both qualitative and quantitative evaluations highlight its generalization ability in open-world scenarios.
cs.CV
Daeyong Kwon, Soyoung Yoon, Seung-won Hwang
Multi-hop QA benchmarks often reward Large Language Models (LLMs) for spurious correctness, where models reach correct answers through invalid intermediate reasoning. We propose SAFE, an LLM-as-verifier framework for evidence-grounded multi-hop QA. Rather than judging only the final answer after generation, SAFE verifies reasoning during generation by checking intermediate steps against the provided passages and previous reasoning trajectory. To make this process checkable, SAFE decomposes reasoning into atomic, evidence-grounded units represented with Knowledge Graph (KG) triples. At train-time, SAFE verifies benchmark supervision under KG-grounded constraints and constructs reliable verifier training data. At inference-time, an external verifier checks each generated step, identifies invalid reasoning, and provides correction feedback before errors propagate. Across three multi-hop QA benchmarks, SAFE improves accuracy by 8.8 pp on average. These results show that evidence-grounded multi-hop QA benefits from shifting LLM-based evaluation from post-hoc answer judgment to stepwise reasoning verification.
cs.CL · cs.AI
Francisco Martín Rico, Esther Aguado González, Francisco Miguel Moreno Olivo, Rodrigo Pérez Rodríguez, José Miguel Guerrero Hernández, Francisco J. Rodríguez Lera
Planning at run time, as in robots that re-plan while they act, needs a planner that reads the PDDL its users write (durative actions, numeric resources, time windows, continuous change), answers quickly, and delivers only validated plans. Specialised planners are strong on individual fragments of PDDL, but they differ in coverage, latency and in what their failures mean. We study whether selecting heterogeneous engines from the semantics of each task can give broad coverage of PDDL, from STRIPS to PDDL+, while remaining competitive with specialised planners and making explicit what each answer guarantees. Kairos profiles each task to decide which of five engines apply (ground heuristic search, compressed and snap-action temporal search, time steps for hybrid dynamics, lifted search) and schedules them under one budget of time and memory. Every candidate plan is checked by a validator independent of the planner's transformations, and "unsolvable" is reported only when a search whose exhaustion is a proof exhausts its space. On 1,756 tasks from nine comparative suites, Kairos solves more tasks at the full limits than the strongest evaluated planner of each suite, with the largest differences on numeric, temporal, PDDL3 and PDDL+ planning, and it also solves the most tasks at 100 ms and 1 s. Resampling over domains, the gain is above zero in six of the nine suites, while on classical planning against LAMA it is within the noise. A further 450 feature-oriented tasks and 26 one-feature probes show that Kairos handles all seventeen features probed, where no other planner evaluated handles more than eight. Ablations identify compressed temporal search and the novelty fallback as major contributors, and a comparison with VAL traces every disagreement to a PDDL+ convention or to a documented VAL defect. Planner and benchmark are open source.
cs.AI · cs.RO
Shota Nomura, Takayuki Osa, Yasuhiro Kato, Kazuki Ota, Naoki Shitanda, Tatsuya Harada
Imitation learning has become a widely adopted approach for vision-based robot policies, owing to its conceptual simplicity, scalability, and ability to leverage large-scale datasets. As these policies have grown more capable, action chunking has become their standard interface. At each decision step the policy outputs several future actions at once. Backbones from diffusion policies to large vision-language-action models all share it. Whatever backbone produces the chunk, its temporal structure is learned only implicitly. The commanded trajectory is therefore jerky both within a chunk and between successive chunks. On real hardware this appears as visible vibration and as repeated saturation of the follower's velocity limit. We address this issue by incorporating a movement-primitive-based trajectory representation into action chunking. Specifically, a trajectory segment is represented as a linear combination of smooth basis functions, yielding smooth trajectories by construction while leaving the policy architecture, training objective, and inference procedure unchanged. The same representation also yields a closed-form, inference-time correction for the second source of jerk. We condition the predicted coefficients so that the new chunk passes through a via-point at the previous chunk's last commanded action. This correction needs no retraining and, unlike asynchronous Real-Time Chunking, applies to policies queried in one shot. We instantiate it on a one-shot (ACT) and an iterative (DiT) backbone and evaluate in simulation and on a real bimanual ALOHA platform. The representation lowers within-chunk jerk by an order of magnitude on the one-shot backbone and by 2.4 times on the iterative one, with task success statistically non-inferior. The conditioning then removes 73-84% of the boundary discontinuity, at negligible per-step cost.
cs.LG · cs.RO
Haozhe Yang, Zhiyang Dou, Zekai Gu, Cheng Lin, Wenping Wang, Yuan Liu, Taku Komura
Generating high-quality and controllable camera motion is essential for AI-assisted cinematography, video synthesis, and 3D scene understanding. We introduce TKCAM, a Text- and Keyframe-conditioned CAMera-motion synthesis framework based on generative masked modeling. We represent camera dynamics using a 12-dimensional kinematic feature comprising position, velocity, and a continuous rotation representation and discretize them into hierarchical motion tokens via a Residual Vector Quantizer (RVQ). A two-stage masked transformer architecture then learns to reconstruct and refine these tokens, utilizing explicit self- and cross-attention modules for multimodal conditioning. A central feature of our framework is sparse visual keyframe conditioning: users can provide free-form text prompts together with RGB observations at selected timestamps, which provide temporally localized visual guidance for generating coherent in-between trajectories. Furthermore, to advance evaluation standards, we curate RealEstate10K-Cap, a large-scale text-camera dataset, and establish a cross-domain benchmark with a Universal CLaTr Evaluator. Extensive experiments demonstrate that TKCAM surpasses recent state-of-the-art baselines on Fréchet distance (FID), text-motion matching scores, and retrieval metrics (R@K), while additional analyses evaluate temporal smoothness and cross-domain generalization. Code and RealEstate10K-Cap dataset are available at https://github.com/linearalgebrayhz/TKCAM.
cs.CV
Cenwei Zhang, Suncheng Xiang, Lei You
Medical segmentation foundation models such as SAM and MedSAM provide strong prompt-driven segmentation, but their image encoders are still too large for many clinical settings. Compression is also risky in medicine because a model can keep high Dice while losing boundary fidelity. We propose MedCore, a structured pruning framework for MedSAM. The main idea is to preserve two kinds of structures: structures that became important during SAM-to-MedSAM adaptation, and structures that have high boundary leverage. We identify the first type by a dual-intervention score comparing zeroing a group with resetting it to its original SAM weights. We identify the second type by boundary-aware Fisher estimation. We also introduce a boundary leverage principle, which relates first-order boundary displacement to boundary logit perturbation divided by the spatial gradient norm. This principle explains why boundary metrics can degrade even when Dice remains high. On polyp segmentation benchmarks, MedCore reduces parameters by 60.0% and FLOPs by 58.4% while achieving Dice 0.9549, Boundary F1 0.6388, and HD95 5.14 after post-pruning fine-tuning. It also reaches 86.6% parameter reduction and 90.4G FLOPs, retaining Dice and HD95 near the fine-tuned MedSAM reference, with lower BF1. Our analysis shows that MedSAM lies in a head-fragile boundary regime: the median 95th-percentile boundary leverage is 2.887 times as high for head-pruning as for MLP-pruning steps, consistent with BF1 and HD95 degradation. Our code is available at https://github.com/cenweizhang/MedCore.
cs.CV · cs.LG
Kai Sheng, Liuyi Wang, Jinlong Li, Haojie Dai, Chengju Liu, Qijun Chen
Giving embodied agents the freedom to act autonomously, initiate interactions, and understand their surroundings through experience is a step toward bringing them to life in the physical world. Realizing this vision requires action, interaction, and memory to work together, so that each encounter can inform current decisions and contribute experience for future tasks. We present EvoLife, a unified navigation harness that enables a multimodal embodied agent to choose how to move, when to ask, and what experience to carry forward. Its hybrid action space supports waypoint navigation, local discrete adjustments, pixel-based navigation, and object-directed navigation. Active interaction allows the agent to seek clarification or assistance when it encounters ambiguity or difficulty, while persistent spatial memory binds observations and dialogue to places, retaining useful experience to guide subsequent visits. We evaluate eight benchmarks across four task families: vision-language navigation, object navigation, interactive instance navigation, and continuous multi-task navigation. With an average of only 1.87 active interactions per episode, our method achieves a first-visit success rate (SR) of 100% across four 100-episode sets spanning vision-language navigation and object navigation. Complete-sequence SR reaches 98% on 50 LHPR-VLN sequences and 94% on 50 GOAT-Bench sequences. In the R2R revisit setting, success weighted by path length (SPL) increases from 82.23% to 93.31% with retained experience. Multi-day physical navigation further demonstrates exploration, proactive help-seeking, and revisiting remembered places. Together, these results advance the vision of bringing embodied agents to life through autonomous action, proactive interaction, and the accumulation and reuse of experience.
cs.RO
Jaewoo Jeong, Taesoo Kim, Sangdon Park
The hallucination of code generation models hinders their applicability to systems requiring higher safety standards. One critical bottleneck in addressing code hallucination is the difficulty of identifying the functional correctness of generated code, due to its unnatural form. We address this core bottleneck by automatically generating unit tests using dynamic code analysis tools, leveraging the \emph{executable nature} of code. Accordingly, we propose a \emph{selective code generator} that abstains from uncertain generations -- based on the functional correctness evaluated by generated unit tests -- to theoretically control the correctness among non-abstained answers, \ie the false discovery rate. Finally, we propose to use generated unit tests in evaluation as well as in learning for precise code evaluation, calling this paradigm \emph{FuzzEval}. We demonstrate the efficacy of our method along with the controllability of code hallucination and reasonable selection efficiency.
cs.SE · cs.LG
Fareed Sheriff
Masked diffusion models train a denoiser on masked cross-entropy and generate by unmasking many positions in parallel. Existing MDM sampling bounds assume an exact denoiser or leave its error uncontrolled while training bounds cover only the masking distribution on which the model was trained. We present an end-to-end guarantee for a learned denoiser under conditional mixing (CM): stationary sources (e.g. floored order-$r$ Markov and two-sided-Doeblin hidden-Markov sources) for which revealing distant visible tokens moves a masked token's posterior on average exponentially little in the distance with correlation length $τ_c$. We show the empirical risk minimizer trained under any masking distribution attains per-token excess $\varepsilon'$ on that distribution from $\varepsilon'^{-\mathcal{O}(τ_c\log|\mathcal{V}|)}$ training sequences, independent of sequence length $N$ -- one sample budget serves every decoding schedule when each schedule's denoiser is trained under its own weighting. We further show that the exponent's order $Θ(τ_c\log|\mathcal{V}|)$ is minimax-optimal up to constants; more efficient rates require added structure. With this, we bound the end-to-end KL as $\mathrm{KL}(π\Vert p_{\mathrm{alg}})\leq \varepsilon+N\varepsilon'$ under a dilated schedule of $T=\mathcal{O}(τ_c\log(N|\mathcal{V}|/\varepsilon))$ denoising steps versus $Ω(N)$ steps for a fixed random order on an explicit CM source. Finally, we show that block-autoregressive samplers chain across blocks with no added error and that a $Λ$-looped order-$r$ message-passing class on order-$r$ sources achieves polynomial sample complexity, with approximation error improving per loop until saturation at $Λr\approx τ_c\log(1/\varepsilon')$. Experiments on released and retrained models support our results' qualitative predictions.
cs.LG
Shengze Xu, Maolin Deng, Tieyong Zeng
Softmax routing in mixture-of-experts models can concentrate probability on a subset of experts, suppressing access to the others. We study how this concentration forms, why it persists, and how suppressed probabilities recover. For fixed experts, a stationary-loss barrier identifies the face supporting the optimal routed output. In a common-target square-loss model with freely trainable expert outputs, slow adaptation tracks fixed-expert dynamics on finite horizons, while every fixed positive adaptation rate yields an interior routing limit. An exact normal-velocity decomposition separates complementary-expert covariance from relative gradient advantage. Under positive two-sided power bounds along recovery paths, uniform over initial masses, the resulting coefficient's order of vanishing gives logarithmic or power-law recovery times. Integrations of constructed fields reproduce these orders. An exact discrete log-odds identity relates neural updates to threshold crossing. In paired Transformer-MoE training, removing a class-conditional router-gradient contrast delays confirmed recovery and reduces accumulated log-odds on fixed tokens, both with and without capacity constraints.
cs.LG