Object_Detection
-
SphereVideo: Prototype-anchored Hyperspherical Boundary for Continual AI-generated Video Detection
2026-08-02 15:54:54Fei Li, Yue Yu, Yuran Wang, Xinghan Li, Jingjing Chen, Yu-Gang JiangAbstract
AI-generated video (AIGV) detection aims to distinguish real videos from AI-generated ones. In practice, detectors trained on existing data often fail to generalize to newly emerging generative models, making this task challenging. Therefore, continual learning (CL) is essential for improving the adaptability. However, CL frameworks for this task remain underexplored. To this end, we propose SphereVideo, a novel CL framework for AIGV detection built on two key observations. First, real videos exhibit a compact feature distribution. Based on this, we encourage real video features to cluster around a real prototype on a hypersphere while repelling AI-generated samples, thereby establishing a decision boundary. This prototype serves as a stable anchor for CL, regulating boundary evolution and mitigating catastrophic forgetting. Second, existing methods tend to rely solely on spatial artifacts as shortcuts. To enhance temporal modeling, we introduce a strategy that models the temporal dynamics of real data at both frame and clip levels. By strengthening real data modeling, this strategy further facilitates learning a real prototype and forming a stable decision boundary. Moreover, we construct a comprehensive and challenging benchmark. Extensive experiments demonstrate that SphereVideo achieves an improved plasticity-stability trade-off, outperforming prior methods by 3.08% on seen data and 4.00% on unseen AI-generated data.
Abstract (translated)
URL
https://arxiv.org/abs/2608.01334
PDF
-
Beyond Symmetric Fusion: Exploiting Task-Dependent Modality Strengths for RGB-Event Small Object Detection
2026-08-02 15:10:54Ziheng Wang, Chaolang Li, Yutong Yang, Xiaohan Xu, Chongxiang Yang, Hengxuan Zhong, Zhen Liang, Pengwen DaiAbstract
State-of-the-art RGB-Event detectors improve the detection of small, fast-moving objects by combining complementary features from RGB and Event data, yet they typically fuse the two modalities into a unified representation for both localization and classification. Such a task-symmetric design is inconsistent with the intuition that the two modalities should play different roles according to their task-specific strengths. To examine this issue, we conduct a modality-specific evaluation and find that the relative advantage of the two modalities reverses across tasks: Event data are substantially more effective for class-agnostic localization, whereas RGB data provide stronger category evidence within localized target regions. Motivated by this task-dependent asymmetry, we propose an Asymmetric Event-RGB Object Detection Transformer (AERODet). During class-agnostic localization, Scale-wise Uncertainty-aware Reliability Estimation (SURE) calculates the relative reliability of the two modalities from their objectness response heatmaps and accordingly calibrates their contributions when the decoder aggregates multimodal features. Once the candidate boxes are obtained, Task-Decoupled Semantic Refinement (TDSR) decouples classification from localization and uses RGB RoI features for fine-grained classification. Extensive experiments on FRED and NeRDD demonstrate that AERODet achieves state-of-the-art performance. In particular, it surpasses the strongest RGB-Event baseline by 10.7 mAP points on the FRED challenging split.
Abstract (translated)
URL
https://arxiv.org/abs/2608.01302
PDF
-
A Benchmark Dataset for MLLM-Generated Image Detection: GPT Image2 & Nano Banana2
2026-08-02 14:09:17Zirui Zhang, Yinbo Yu, Donghai Guan, Chunwei Tian, Daoqiang Zhang, Qi ZhuAbstract
The realism of images generated by multimodal large language models (MLLMs), such as GPT Image2 and Nano Banana2, has improved rapidly in recent years. Compared with early generative models, current models have made clear progress in text rendering. They can produce high-quality images that closely resemble real-world application scenarios. The enhanced generation capabilities of current MLLMs pose increasingly severe challenges to AI-generated image detection. Detection is no longer limited to identifying obvious artifacts left by early generators. Instead, it requires systematic and realistic benchmarks for the new generation of generated content. However, most existing benchmarks are still built around early generative models and cannot fully evaluate the forensic challenges introduced by high-quality and multi-form generated images. To address this gap, this paper constructs a benchmark dataset for detecting images generated by MLLMs. The benchmark covers several realistic application scenarios and adopts three generation protocols to simulate direct generation, reference-based reconstruction, and local editing. Based on this benchmark, we evaluate detector degradation from traditional scenarios to MLLM-generated images and analyze false positive rates and false negative rates across three sample types, revealing the failure modes of existing methods. We further propose a structural-artifact-prior-guided dual-stream prompt framework (SAP-DSP) as a strong baseline. SAP-DSP uses dual-stream prompt learning and structure-aware routing fusion to improve representation learning. Extensive experiments show that the proposed benchmark exposes the performance degradation of existing detectors on high-quality generated images, while SAP-DSP achieves more stable detection results on this benchmark. Our code and dataset are publicly available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2608.01258
PDF
-
Vulnerabilities, Secrets and Misconfiguration in the Highest-Exposure Docker Hub Images
2026-08-02 13:57:28Cristhian Kapelinski, Beatriz Machado, Diego KreutzAbstract
Docker Hub is the registry underneath most container deployments, and a flaw in a widely reused base image is inherited by every image built on it. Prior ecosystem-scale measurements each rely on a single detector, leaving the tool-dependence of their counts unquantified, while the studies that do compare scanners use samples of tens to hundreds of images. We present ChimangoScan, a pipeline that crawls the Docker Hub namespace (12,716,568 repositories, 663.8 billion cumulative pulls), reconstructs the image layer graph (54.4 million IS_BASE_OF edges), ranks images by an exposure score that folds an image's own pull count and those of its entire downstream subtree into one scalar, and scans the 52,895 highest-exposure repositories (84.7% of all recorded pulls) with six independent scanners, yielding 170.4 million findings. Vulnerabilities are near-universal: 96.3% of images carry a known package vulnerability, 93.4% a critical one, and 98.0% at least one CIS Docker Benchmark misconfiguration. The posture a single tool reports is largely an artifact of that tool: of 80.7 million distinct (vulnerability, package) groups, 66.8% are flagged by only one of the three vulnerability scanners and just 2.7% by all three, and the best single scanner recovers 66.9%. TruffleHog flags a secret in 76.9% of images, yet hand-labeling 1,100 random detections finds 99.7% are non-credentials. A single zlib CVE reaches images carrying 47.3% of total corpus exposure and propagates to 1.13 million distinct downstream images, but exposure does not predict how vulnerable an image is. We release the pipeline and the 283 GB dataset.
Abstract (translated)
URL
https://arxiv.org/abs/2608.02669
PDF
-
Separate-and-Detect: Unified Drum Transcription and Stem Generation via Latent Diffusion
2026-08-02 08:50:13Wei-Han Hsu, Chih-Cheng Chang, Bo-Yu Chen, Li Su, Yi-Hsuan YangAbstract
Automatic Drum Transcription (ADT) is commonly formulated as a direct mapping from a music mixture to symbolic drum events. While effective for transcription, this formulation discards the acoustic stems that are useful for editing, remixing, and production. We revisit an alternative separate-and-detect formulation, where a drum source separation front end first produces five editable drum stems, and a fixed onset detector then converts each stem into symbolic events. The separator is built on a five-stem latent diffusion model that jointly generates kick, snare, toms, hi-hats, and cymbals in a compact VAE latent space. We further study two training-only auxiliary branches--an onset branch (OB) and a timbre branch (TB)--which shape the separator during learning but are discarded at inference. Trained on synthetic drum multitracks and evaluated on MDB Drums and ENST-Drums, the proposed pipeline consistently improves over a strong U-Net-based drum separation baseline in overall transcription F1. It also outperforms a representative end-to-end ADT system on kick and snare F1 under our evaluation protocol, while additionally providing separated audio stems. The ablation results show that OB gives the most stable transcription gains, whereas TB changes the trade-off between reconstruction, acoustic stem quality, and onset detection. These results suggest that generative drum demixing can serve not only as a source separation model, but also as a practical front end for interpretable drum transcription.
Abstract (translated)
URL
https://arxiv.org/abs/2608.01093
PDF
-
DeBERTa-Sentinel: Toward Transparent and Trustworthy Detection of AI-Generated Text
2026-08-02 07:08:20Muhammad Yousaf Rehman, Muhammad IslamAbstract
The rapid spread of large language models (LLMs) across the web raises concerns about misinformation, academic integrity, automated content manipulation, and risks to vulnerable online communities. Existing transformer-based detectors, such as GPT-Sentinel, show promise but struggle to generalize to diverse model outputs and paraphrasing attacks, limiting their role in building trustworthy web ecosystems. This work introduces DeBERTa-Sentinel, a responsible AI-generated text detection framework leveraging DeBERTa-v3's disentangled attention to capture subtle structural irregularities in synthetic content. A central design principle is transparency: unlike black-box commercial detectors, DeBERTa-Sentinel exposes token-level explanations of its decisions, enabling affected stakeholders journalists, educators, and platform trust and safety teams to audit, challenge, and contextualize detection outcomes. Using the GLC-AIText dataset of 28,057 human and LLM-generated samples (GPT, LLaMA, and Claude) with a 60-20-20 split, DeBERTa-Sentinel achieves 98.21\% validation accuracy and surpasses the RoBERTa-Sentinel baseline from NeurIPS 2025, achieving 97.53\% test accuracy, 95.89\% precision, 99.33\% recall, and 99.53\% ROC-AUC, and maintaining a 0.665\% false negative rate. The model's interpretability reveals linguistic markers such as academic phrasing and formal transitions associated with synthetic text, directly supporting stakeholder needs for verifiable, auditable content-authenticity decisions. By advancing responsible detection methods that reduce bias and enhance explainability, DeBERTa-Sentinel promotes trustworthy, ethical, and human-centric AI systems. Code and data are available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2608.01046
PDF
-
GraRe: Grasp Candidate Re-Ranking for Frozen 6-DoF Grasp Detectors
2026-08-02 02:55:32Jibao Yuan, Yuhui Zhao, Yinzhen Lv, Chao Xu, Shun Li, Chenxi Deng, Shaofei ChenAbstract
Existing 6-DoF grasp detectors typically rank grasp candidates by detector confidence. However, our analysis on GraspNet-1Billion shows that detector confidence is often poorly aligned with grasp quality, causing successful grasp candidates to be ranked too low during execution. Motivated by this observation, we formulate grasp candidate re-ranking as a separate task for frozen detectors, aiming to improve candidate ordering without changing the detector or its grasp candidates. We propose GraRe, which estimates grasp quality from candidate attributes, shell-stratified local geometry, and object context. Candidate attributes condition the local geometric and object-context representations, and a Transformer fuses all three feature types. The predicted quality is combined with detector confidence to produce the final ranking. Experiments on GraspNet-1Billion with three frozen detectors show consistent improvements, with gains of up to 13.60 points in Average AP. Real-robot experiments further demonstrate robust grasping in cluttered scenes. These results show that improving candidate ranking provides a practical way to enhance frozen 6-DoF grasp detectors.
Abstract (translated)
URL
https://arxiv.org/abs/2608.00946
PDF
-
Observatorio Lazaro: A self-populating database of anglicism usage in the Spanish press
Abstract
This paper describes Observatorio Lázaro, a language resource that monitors unassimilated lexical borrowings (predominantly English lexical borrowings or anglicisms) in the Spanish digital press. Since April 2020 the system has automatically processed the daily output of a collection of news outlets, detected borrowings with a neural sequence-labeling model, and made the results available through a public web interface and API. The result is a continuously updated diachronic database which, at the time of writing, records more than two million borrowings across 1.88 million articles and 993 million running tokens of text (2020-2026). The paper documents the resource: we describe the end-to-end pipeline (acquisition, detection, post-processing, storage and access), the data model and the terms of availability; we evaluate the resource through the detector's held-out performance (span-level F1=0.86 for the borrowing class), inter-annotator agreement on the training corpus (Cohen's kappa=0.91) and a manual precision audit of 1,000 spans from the deployed data; and we situate it with respect to Spanish borrowing lexicography, annotated borrowing corpora and neology-monitoring observatories. The data shows that unassimilated anglicisms are used in the Spanish press at a frequency of approximately two anglicisms per thousand tokens, and that this rate remains stable. Our statistical analysis over six years reveals that the anglicism vocabulary in Spanish behaves as an open and growing class, with 58.7% of its types attested only once (53.6% after correcting for detection precision), and that its density is highest in the fashion, technology and lifestyle sections and lowest in political and institutional news. The resource is intended to complement static borrowing dictionaries and one-off annotated corpora by providing a continuously updated record of borrowing in the Spanish press.
Abstract (translated)
URL
https://arxiv.org/abs/2608.00713
PDF
-
Assistant Placement Aria: A Benchmark for Egocentric Placement Assistance
2026-08-01 13:13:05Amir Belder, Gon\c{c}alo Dias Pais, Refael Vivanti, Omri Carmi, Daniel DeTone, Oren Shrout, Ido Gattegno, Ayellet TalAbstract
Human assistance in robotics spans around several tasks such as navigation, object manipulation, and placement, where a key challenge is selecting target destinations that align with human intentions or preferences. We focus on this challenge in the context of Virtual Placement (VP), the task of identifying all plausible target locations given scene context and human-centric constraints. This differs from traditional placement tasks that typically focus on a single, predefined target location. The VP problem is complex, as it requires both global and local reasoning about the scene's geometry, semantics, and plausibility. To address this gap, we introduce {\bf Assistant Placement Aria}, the first benchmark to explore diverse aspects of VP, including global, local, and human-centric constraints. It contains both synthetic and real indoor scenes annotated for three tasks: (i)~2D Panel Placement, (ii)~Sitting Suggestion, and (iii)~TV Placement. Each scene includes 2D images, a 3D point cloud, and a textual description of the objects within the scene. By contributing this benchmark, we aim to encourage further research in this underexplored and challenging field that is critically dependent on relevant data. We also evaluate several foundation models for object detection and segmentation on our benchmark.
Abstract (translated)
URL
https://arxiv.org/abs/2608.00652
PDF
-
A False Average: Chain-of-Thought Monitors Collapse Where They Are the Only Defense
2026-08-01 10:42:41Shikhar Shiromani, Leo RichterAbstract
Chain-of-thought (CoT) monitoring is meant to catch the reward hacks that look clean in the actions and betray themselves only in the reasoning. We show that this is exactly where an adversary who controls the reasoning can defeat it. Rewriting only an agent's reasoning to read as good-faith engineering, while copying every command and output verbatim so the exploit is unchanged, drops a held-out monitor's catch rate on that subset from about 95% to under 11% in one gradient-free shot. A monitor's aggregate accuracy is a false average: dominated by hacks the actions give away, it hides the near-total collapse this rewrite produces on the subset where CoT monitoring is the only signal. The attack transfers across monitor families and agent models, reproduces with live agents, though against a calibrated monitor evasion concentrates in the strongest agent. Trace-only defenses recover it only partially, even one primed on the attack, because the rewrite stays truthful about what happened and lies only about intent; only information from outside the trace helps substantially. A probe on an open-weight surrogate monitor's activations separates the hacks its verdict misses, but a causal control shows this is a detector, not evidence the monitor secretly knows.
Abstract (translated)
URL
https://arxiv.org/abs/2608.00583
PDF
-
Test-Time Curriculum for Open-Set AIGC Detection
2026-08-01 09:42:45Yiqian Zhang, Zheyuan Gu, Xiangzhao Hao, Zefeng Zhang, Jingjia Mao, Jiahao Hu, Jiaxu Miao, Jun Yu, Zhenyu Zhang, Shuohuan Wang, Yu SunAbstract
AI-generated image detectors deployed in open-world environments inevitably face distribution shifts as new and stronger generative models continue to emerge. Although existing methods improve cross-generator generalization through better representations or training data construction, they typically follow a static train-once-and-deploy paradigm and cannot adapt after deployment. In this work, we study open-set AIGC image detection from a test-time adaptation perspective. We propose Test-Time Curriculum (TTC), a simple and model-agnostic framework that adapts a detector on unlabeled test data through curriculum-based self-training. TTC starts from highly reliable pseudo-labeled samples and progressively incorporates harder yet informative cases, while enforcing class-balanced selection to reduce biased updates under generator shift. To further improve pseudo-label quality, we introduce Cross-Scale Pseudo-Label Refinement, which aggregates complementary evidence across multiple resolutions for more reliable adaptation, and applies noisy-or fusion at inference to strengthen final predictions. In addition, we construct AIGCGuard, a new benchmark containing 3,100 representative real images and 124,000 generated images from 40 of the most advanced open-source and proprietary text-to-image models. Extensive experiments on five benchmarks show that TTC substantially improves overall detection performance under diverse unseen-generator shifts, establishing a practical and effective test-time adaptation framework for open-set generated image detection.
Abstract (translated)
URL
https://arxiv.org/abs/2608.00559
PDF
-
RadYOLO: Computationally Efficient 3D Object Detection and Segmentation in CT and MRI
2026-08-01 08:02:42Kai Geissler, Laurens M\"uller-Groh, Hans MeineAbstract
Object detection and segmentation in three-dimensional medical images is a very active area of research. However, most proposed deep learning models carry a high computational cost, and only few aim to be broadly applicable, achieve high detection performance, and remain fast to execute on resource-constrained hardware. To address this gap, we present RadYOLO, a 3D extension of YOLO11 tailored to medical images. We compare it with nnU-Net and nnDetection on five datasets comprising CT and MRI data with varying object sizes and prevalence. RadYOLO's detection performance surpasses that of nnDetection on four of five datasets and is comparable on one. Compared to nnU-Net, RadYOLO performs better on lesion detection tasks, while nnU-Net excels at detecting large organs when precise localization is required. When rough object localization is sufficient, RadYOLO matches or outperforms nnU-Net on all five datasets. Regarding inference time, RadYOLO is 8-46x faster than nnU-Net on a GPU. Compared to nnDetection the speedup is even higher. When executed on a CPU, RadYOLO's inference runs within seconds (still faster than nnU-Net on a GPU) offering a significant advantage for clinical and edge-device deployment. RadYOLO repository: this https URL
Abstract (translated)
URL
https://arxiv.org/abs/2608.00508
PDF
-
Hidden-Domain Routing for All-Type Audio Deepfake Detection
2026-08-01 07:32:54Yifan Gao, Yao Tian, Hongbin Suo, Haonan LuAbstract
All-type audio deepfake detection requires authenticity decisions across speech, environmental sound, singing voice, and music, while the audio type is unavailable at inference time. In AT-ADD Track2, this setting creates a hidden audio-domain condition: the binary real/fake label is shared across domains, but representation structure and detector-score behavior vary with audio type. We present a closed-condition routed system that first recovers the hidden audio domain and then interprets detector scores within the selected branch. The AudioType-BEATs-6s Router estimates audio type from a 6-second window; speech inputs are handled by the Speech-XLSR Expert, while sound, singing, and music rely on EAT-based general-audio experts with branch-local score interpretation. Development-set representation analysis, router-family comparisons, and component results show audio-domain separation and complementary detector strengths across audio types. On the official AT-ADD Track2 final evaluation, the system achieves 96.10% Track2 Macro-F1 and ranks first on the final leaderboard, with type-wise Macro-F1 scores of 88.07%, 98.18%, 99.07%, and 99.08% for speech, sound, singing, and music, respectively. These results support recovering the hidden audio domain before interpreting detector scores in all-type audio deepfake detection.
Abstract (translated)
URL
https://arxiv.org/abs/2608.00493
PDF
-
The Gate, Not the Cache: Gate Provenance Bounds the Closed-Loop Reliability of Training-Free VLA Token Skipping
2026-08-01 01:59:12Qi Luo, Shuaijun Liu, Hao Zhao, Kunlin Li, Xiaobo Wang, Ningxing Su, Dongsheng Wang, Yun ChenAbstract
Token skipping is a widely used training-free way to accelerate vision--language--action (VLA) models by bypassing computation for most visual tokens at each control step according to a gate. When the next gate is harvested from the previous accelerated forward, however, the tokens skipped at one step are also the ones least visible to the next gate, and the damage can compound across control steps until the task fails. We study the two mechanisms this class is built on, reuse and deletion, crossing each against where its gate signal comes from on identical episodes. At a skip ratio of 0.9 on LIBERO-Object, both collapse when the gate comes from the model's own accelerated forwards, to 0.68 under reuse and to 0.31 under deletion against a dense 1.00, and the collapse is invisible to the action-level detectors we evaluate. What separates collapse from dense-level operation is not the mechanism but whether the gate is clean, computed by a forward that skipped nothing. We therefore propose actuation-slack refresh, one dense pass run while the robot executes its current action chunk, off the critical path, that hands the next step a clean gate and a fresh KV base. Since the measured detectors do not reliably reveal the failure, the refresh is unconditional rather than triggered. Both mechanisms then recover to 0.98, keeping the speed of skipping and the information of a dense pass. We then integrate the refresh into state-of-the-art caching and pruning methods across two VLA policies, 4 LIBERO suites, and 4 SIMPLER tasks, where it repairs every collapse caused by using a self-harvested gate. Serve latency drops 18--22\% below dense, measured both in simulation and on a physical robot. Where the gate signal comes from, not how tokens are skipped, decides closed-loop reliability for accelerated VLAs.
Abstract (translated)
URL
https://arxiv.org/abs/2608.00391
PDF
-
Belief-Space Perception Routing under Coupled Sensor Faults and Compute Contention
2026-07-31 22:17:12Sparsh Roy, Vihan Aggarwal, Davin YinAbstract
A robot that has to see and react on a fixed clock runs into two problems at once. Its cameras degrade in rain, mud, fog, and darkness. And the single onboard processor it runs on is shared with planning and control, so the compute left over for perception moves around from second to second. Most systems model the two separately. We present a perception router that tracks probabilistic estimates of sensor-fault state and compute- contention state, couples them with a noisy-OR term, and uses the coupled estimate to pick one of four detector configurations (YOLO11x/n at 1280 or 640 px) so that the frame finishes before its deadline. Where the two stressors co-occur, the coupled policy cuts the deadline-miss rate by 1.1 to 9.4 percentage points against a policy that treats them independently. The interval excludes zero in five of six conditions, the pooled effect over 10 sequences and 6 conditions has sign-test p = 0.001, and every uncoupled control and the fault-free trajectory sit at exactly 0.0 pp. Routing costs tens of microseconds per frame. We then asked whether the coupling the method exploits arises on its own. Across eight real RADIATE adverse-weather sequences and three workload proxies independent of the fault signal, after Benjamini-Hochberg correction and a replication run, none of 24 tests found it. We report that null and scope the routing result as a proof of mechanism. Whether such coupling occurs in the field is still open, and the released evaluation pipeline lets a deployment settle it on its own traces.
Abstract (translated)
URL
https://arxiv.org/abs/2608.00322
PDF
-
MDWD: A Street-Level Dataset for Municipal Solid Waste Detection in Dense Urban Environments
2026-07-31 19:54:25Andrea Filiberto Lucas, Mark Bugeja, Carl James Debono, Dylan SeychellAbstract
Automated visual monitoring of urban environments is a growing Computer Vision research area, but municipal solid waste detection remains under-represented in dedicated benchmark resources. Existing waste-related datasets predominantly address individual litter detection, aerial imagery, or image-level classification, and none simultaneously provide street-level imagery, instance-level localization, and categorization of domestic waste streams within a structured municipal collection context. This paper introduces the Maltese Domestic Waste Dataset (MDWD), a street-level benchmark comprising 3,697 high-resolution images and 11,461 manually annotated instances across five domestic waste categories representative of Malta's municipal collection system. The dataset captures substantial variation in location, illumination, object scale, occlusion, and urban context. To establish reproducible baselines, a cross-architecture benchmark is conducted across multiple generations of the YOLO family and a transformer-based detector. On the test set, RF-DETR-M achieves the strongest overall performance with an mAP50 of 94.49% and an F1-score of 93.56%, whilst smaller-capacity variants maintain competitive accuracy at substantially reduced parameter counts. These results indicate that MDWD supports effective training across both compact real-time detectors and transformer-based models, establishing a benchmark for future research in vision-based municipal waste monitoring.
Abstract (translated)
URL
https://arxiv.org/abs/2608.00257
PDF
-
A Few Neurons Reveal When LLMs Misuse Tools: Sparse Detection and Selective Steering for Reliable Tool Use
2026-07-31 19:03:44Yutong Ke, Ming Yin, Chongwen Zhao, Kaizhu HuangAbstract
Agentic LLMs exhibit three consequential tool-use failures: invalid arguments (validity), unnecessary calls (over-calling), and omitted calls when tools are needed (missing). We find that a small, failure-specific set of MLP neurons could distinguish such failures with linearly separable decision boundaries. Building on this observation, we introduce PRISMS (Probing Representations In Support of Monitoring and Steering), a closed-loop framework that shares a failure-specific neuron basis between sparse detection and activation steering. PRISMS selects contribution-critical MLP neurons and fits an L1-regularized detector on their activations. Across six models from the Qwen3, Llama, and Gemma families, over-calling and missing are detected at the pre-generation prompt boundary with ROC-AUC 0.90-1.00, while validity is detected from the generated tool-call span with ROC-AUC 0.86-0.90. These results are achieved with highly sparse readouts: only 1-2 MLP neurons for missing, 2-16 for over-calling, and approximately 128 for validity. These sparse detectors match or outperform dense residual-stream baselines using 23-627 times fewer features. The shared neuron basis also supports bidirectional control over tool-calling behavior, suppressing unnecessary calls and eliciting omitted ones. PRISMS therefore gates intervention on predicted failure risk to mitigate the collateral effects of unconditional steering. Across all six models, PRISMS reduces pooled over-calling rate by 80% (from 0.131 to 0.026) while increasing tool-required accuracy by 14.2 percentage points (from 0.689 to 0.831). PRISMS thus provides lightweight failure detection and selective intervention across model families.
Abstract (translated)
URL
https://arxiv.org/abs/2608.00218
PDF
-
TOOD: Task-Aware Out-of-Distribution Score Calibration for Continual Learners
2026-07-31 16:18:30Mostafa ElAraby, Samer B. Nashed, Liam PaullAbstract
The primary challenge of continual learning (CL) systems is to learn new tasks while remaining performant on previously learned tasks. A similarly important though less well-studied aspect of CL systems is their ability to distinguish inputs that are unlikely to come from within the set of tasks the system has already encountered, often called out-of-distribution (OOD) detection. This paper presents several findings related to the dynamics of OOD detection in CL systems, causes of performance degradation over time which we call OOD forgetting (OODF), and proposed mitigation strategies for this degradation. Chiefly, we find the unintuitive result that OODF is only weakly anti-correlated with classification performance on previous tasks, suggesting that the underlying mechanisms producing OODF are distinct. Moreover, this effect is observed for both energy-based and feature-based OOD detection methods. Energy-based detectors suffer a drop in logit scale as additional tasks are learned, which we term the Confidence Gap, while feature-based detectors also degrade under a complementary effect we call Manifold Crowding. Motivated by these observations, we propose TOOD, a training-free post-hoc method that decomposes logits into per-task energy scores and re-calibrates them using replay-buffer statistics. Experiments on CIFAR-10, CIFAR-100, and a 100-task ImageNet-1K stream show that TOOD improves OOD detection performance over uncalibrated energy in most settings and ranks first or second in nine of ten CIFAR configurations, with the largest gains when the confidence gap is most severe. These results suggest that a substantial portion of OOD deterioration in continual learning arises from score miscalibration rather than from a complete loss of discriminative structure.
Abstract (translated)
URL
https://arxiv.org/abs/2607.29592
PDF
-
Explaining AI-Image Detection: What the Heatmap Actually Shows
2026-07-31 16:07:05Leonid Kuturin, Ilya Sotnikov, Mark Khusnutdinov, Mikhail Potemkin, Pavel Baranas, Aleksandra Korepanova, Alexander KalashnikovAbstract
A marketplace review photograph is a document: platforms approve refunds on it, and generative models drove the cost of forging one to zero. We study that detection problem, so we build a detector and attach an attribution map as its evidence, then measure what that pair delivers on 186,527 images under controls designed to change our conclusions when something is wrong. Compression history, not synthesis, drives naive evaluation: our strongest model reaches 0.9999 PR-AUC (area under the precision-recall curve) on a product-disjoint split, yet falls to 0.7254 once we re-encode synthetics into the real class's format, while five public detectors move by at most 0.07. Aligning one class relocates the cue rather than removing it, and the repaired model then assigns native files a median probability of synthesis of 0.0004. One identical final encode for both classes repairs that, and a three-seed factorial credits the encoding change with the whole gain (+0.176 +- 0.009 PR-AUC). That encode equalises the last stage only: forensic features alone still separate the classes at 0.7145 against a base rate of 0.254. For evidence we test maps causally, against controls that never consult the detector. Whether an attribution ranking exists at all depends on whether the detector reacts to the image. On our first-fix detector, which calls 96 of 100 edited frames real, no map beats a random one. On the detector we selected, twelve of seventeen maps clear that control on edited images and eight on generated ones; perturbation leads both axes and no gradient-CAM variant shows a positive advantage. The trivial controls never clear it, and on generated images the centre prior is worse than random. Our ensembled regional map clears both axes and takes the top pixel AP at 12.4 s per map against 44.9 for occlusion. Clearing a detector-blind control is not yet a faithful explanation, and we demonstrate none.
Abstract (translated)
URL
https://arxiv.org/abs/2607.29581
PDF
-
ARB: A Matched Authorship-Rewriting Benchmark Dataset for AI-Text Detector Evaluation
2026-07-31 15:35:57Gaetano Perrone, Simon Pietro RomanoAbstract
Standard AI-text detection benchmarks compare human-written text against text generated directly by large language models (LLMs). While prior work has shown that rewriting and paraphrasing can degrade detector performance, it remains unclear whether performance measured on this conventional benchmark predicts detector behavior when human-authored content is rewritten by an LLM. To address this gap, we introduce Authorship-Rewriting Benchmark (ARB), built from 1,800 human source texts (600 each from XSum, WritingPrompts, and OpenWebText) and four open-weight generators (Llama-3.2-3B, Qwen2.5-7B, Mistral-7B, Gemma-2-9B). Each source item yields four matched variants: human-written (HUMAN), direct LLM generation (Free-LLM), LLM-rewritten human text (H2L), and same-generator LLM-rewritten LLM text (LLM2L). We evaluated five detectors (FastDetectGPT, Binoculars-falcon-7b, RADAR, BERT-Defense, RoBERTa-Defense) at a strict 1%-false-positive operating point (TPR@1%FPR). FastDetectGPT and Binoculars-falcon-7b detected 91.2% and 93.5\% of direct LLM text, but only 30.8% and 15.1% of human text an LLM had rewritten, a drop of 60-78 percentage points. The same detectors retained 78.3% and 83.0% recall when LLM text was rewritten by the same model, a much smaller decline of 10-13 points. RADAR followed the same pattern (66.8% to 12.2%), while BERT-Defense and RoBERTa-Defense stayed below 3% recall across all regimes. These results show that detector performance measured on the conventional human-vs-LLM benchmark does not transfer to human-authored text revised by an LLM, even though the same detectors remain largely robust to LLM-only rewriting.
Abstract (translated)
URL
https://arxiv.org/abs/2607.29539
PDF
