# Research Papers — 2026-04 · bundle page 127 of 233

> Full-text markdown of papers #6301–6350 of 11612 published in 2026-04, newest first. AI-readable mirrors of the full 11612-paper 2026-04 archive (50 papers per file).

Canonical page: https://aipod.dev/papers/archive/2026-04/page/127
[← Previous bundle](https://aipod.dev/papers/archive/2026-04/page/126.md)
[Next bundle →](https://aipod.dev/papers/archive/2026-04/page/128.md)

### Deepfakes at Face Value: Image and Authority

Authors: James Ravi Kirkpatrick
Published: 2026-04-14
Categories: cs.CY, cs.AI
arXiv: https://arxiv.org/abs/2604.12490v1

Deepfakes are synthetic media that superimpose or generate someone's likeness on to pre-existing sound, images, or videos using deep learning methods. Existing accounts of the wrongs involved in creating and distributing deepfakes focus on the harms they cause or the non-normative interests they violate. However, these approaches do not explain how deepfakes can be wrongful even when they cause no harm or set back any other non-normative interest. To address this issue, this paper identifies a neglected reason why deepfakes are wrong: they can subvert our legitimate interests in having authority over the permissible uses of our image and the governance of our identity. We argue that deepfakes are wrong when they usurp our authority to determine the provenance of our own agency by exploiting our biometric features as a generative resource. In particular, we have a specific right against the algorithmic conscription of our identity. We refine the scope of this interest by distinguishing between permissible forms of appropriation, such as artistic depiction, from wrongful algorithmic simulation.

---

### KG-Reasoner: A Reinforced Model for End-to-End Multi-Hop Knowledge Graph Reasoning

Authors: Shuai Wang, Yinan Yu
Published: 2026-04-14
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2604.12487v1

Large Language Models (LLMs) exhibit strong abilities in natural language understanding and generation, yet they struggle with knowledge-intensive reasoning. Structured Knowledge Graphs (KGs) provide an effective form of external knowledge representation and have been widely used to enhance performance in classical Knowledge Base Question Answering (KBQA) tasks. However, performing precise multi-hop reasoning over KGs for complex queries remains highly challenging. Most existing approaches decompose the reasoning process into a sequence of isolated steps executed through a fixed pipeline. While effective to some extent, such designs constrain reasoning flexibility and fragment the overall decision process, often leading to incoherence and the loss of critical intermediate information from earlier steps. In this paper, we introduce KG-Reasoner, an end-to-end framework that integrates multi-step reasoning into a unified "thinking" phase of a Reasoning LLM. Through Reinforcement Learning (RL), the LLM is trained to internalize the KG traversal process, enabling it to dynamically explore reasoning paths, and perform backtracking when necessary. Experiments on eight multi-hop and knowledge-intensive reasoning benchmarks demonstrate that KG-Reasoner achieves competitive or superior performance compared to the state-of-the-art methods. Codes are available at the repository: https://github.com/Wangshuaiia/KG-Reasoner.

---

### Elastic Net Regularization and Gabor Dictionary for Classification of Heart Sound Signals using Deep Learning

Authors: Mahmoud Fakhry, Ascensión Gallardo-Antolín
Published: 2026-04-14
Categories: cs.SD, cs.AI
arXiv: https://arxiv.org/abs/2604.12483v1

In this article, we propose the optimization of the resolution of time-frequency atoms and the regularization of fitting models to obtain better representations of heart sound signals. This is done by evaluating the classification performance of deep learning (DL) networks in discriminating five heart valvular conditions based on a new class of time-frequency feature matrices derived from the fitting models. We inspect several combinations of resolution and regularization, and the optimal one is that provides the highest performance. To this end, a fitting model is obtained based on a heart sound signal and an overcomplete dictionary of Gabor atoms using elastic net regularization of linear models. We consider two different DL architectures, the first mainly consisting of a 1D convolutional neural network (CNN) layer and a long short-term memory (LSTM) layer, while the second is composed of 1D and 2D CNN layers followed by an LSTM layer. The networks are trained with two algorithms, namely stochastic gradient descent with momentum (SGDM) and adaptive moment (ADAM). Extensive experimentation has been conducted using a database containing heart sound signals of five heart valvular conditions. The best classification accuracy of $98.95\%$ is achieved with the second architecture when trained with ADAM and feature matrices derived from optimal models obtained with a Gabor dictionary consisting of atoms with high-time low-frequency resolution and imposing sparsity on the models.

---

### Social Learning Strategies for Evolved Virtual Soft Robots

Authors: K. Ege de Bruin, Kyrre Glette, Kai Olav Ellefsen, Giorgia Nadizar, Eric Medvet
Published: 2026-04-14
Categories: cs.RO, cs.AI
arXiv: https://arxiv.org/abs/2604.12482v2

Optimizing the body and brain of a robot is a coupled challenge: the morphology determines what control strategies are effective, while the control parameters influence how well the morphology performs. This joint optimization can be done through nested loops of evolutionary and learning processes, where the control parameters of each robot are learned independently. However, the control parameters learned by one robot may contain valuable information for others. Thus, we introduce a social learning approach in which robots can exploit optimized parameters from their peers to accelerate their own brain optimization. Within this framework, we systematically investigate how the selection of teachers, deciding which and how many robots to learn from, affects performance, experimenting with virtual soft robots in four tasks and environments. In particular, we study the effect of inheriting experience from morphologically similar robots due to the tightly coupled body and brain in robot optimization. Our results confirm the effectiveness of building on others' experience, as social learning clearly outperforms learning from scratch under equivalent computational budgets. In addition, while the optimal teacher selection strategy remains open, our findings suggest that incorporating knowledge from multiple teachers can yield more consistent and robust improvements.

---

### T2I-BiasBench: A Multi-Metric Framework for Auditing Demographic and Cultural Bias in Text-to-Image Models

Authors: Nihal Jaiswal, Siddhartha Arjaria, Gyanendra Chaubey, Ankush Kumar, Aditya Singh, Anchal Chaurasiya
Published: 2026-04-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2604.12481v1

Text-to-image (T2I) generative models achieve impressive visual fidelity but inherit and amplify demographic imbalances and cultural biases embedded in training data. We introduce T2I-BiasBench, a unified evaluation framework of thirteen complementary metrics that jointly captures demographic bias, element omission, and cultural collapse in diffusion models - the first framework to address all three dimensions simultaneously. We evaluate three open-source models - Stable Diffusion v1.5, BK-SDM Base, and Koala Lightning - against Gemini 2.5 Flash (RLHF-aligned) as a reference baseline. The benchmark comprises 1,574 generated images across five structured prompt categories. T2I-BiasBench integrates six established metrics with seven additional measures: four newly proposed (Composite Bias Score, Grounded Missing Rate, Implicit Element Missing Rate, Cultural Accuracy Ratio) and three adapted (Hallucination Score, Vendi Score, CLIP Proxy Score). Three key findings emerge: (1) Stable Diffusion v1.5 and BK-SDM exhibit bias amplification (>1.0) in beauty-related prompts; (2) contextual constraints such as surgical PPE substantially attenuate professional-role gender bias (Doctor CBS = 0.06 for SD v1.5); and (3) all models, including RLHF-aligned Gemini, collapse to a narrow set of cultural representations (CAS: 0.54-1.00), confirming that alignment techniques do not resolve cultural coverage gaps. T2I-BiasBench is publicly released to support standardized, fine-grained bias evaluation of generative models. The project page is available at: https://gyanendrachaubey.github.io/T2I-BiasBench/

---

### Audio Source Separation in Reverberant Environments using $β$-divergence based Nonnegative Factorization

Authors: Mahmoud Fakhry, Piergiorgio Svaizer, Maurizio Omologo
Published: 2026-04-14
Categories: cs.SD, cs.AI
arXiv: https://arxiv.org/abs/2604.12480v1

In Gaussian model-based multichannel audio source separation, the likelihood of observed mixtures of source signals is parametrized by source spectral variances and by associated spatial covariance matrices. These parameters are estimated by maximizing the likelihood through an Expectation-Maximization algorithm and used to separate the signals by means of multichannel Wiener filtering. We propose to estimate these parameters by applying nonnegative factorization based on prior information on source variances. In the nonnegative factorization, spectral basis matrices can be defined as the prior information. The matrices can be either extracted or indirectly made available through a redundant library that is trained in advance. In a separate step, applying nonnegative tensor factorization, two algorithms are proposed in order to either extract or detect the basis matrices that best represent the power spectra of the source signals in the observed mixtures. The factorization is achieved by minimizing the $β$-divergence through multiplicative update rules. The sparsity of factorization can be controlled by tuning the value of $β$. Experiments show that sparsity, rather than the value assigned to $β$ in the training, is crucial in order to increase the separation performance. The proposed method was evaluated in several mixing conditions. It provides better separation quality with respect to other comparable algorithms.

---

### Meet Dynamic Individual Preferences: Resolving Conflicting Human Value with Paired Fine-Tuning

Authors: Shanyong Wang, Shuhang Lin, Yining Zhao, Xi Zhu, Yongfeng Zhang
Published: 2026-04-14
Categories: cs.CL
arXiv: https://arxiv.org/abs/2604.12479v1

Recent advances in large language models (LLMs) have significantly improved the alignment of models with general human preferences. However, a major challenge remains in adapting LLMs to individual preferences, which are not only diverse but also dynamic. In this paper, we introduce a novel framework, Preference-Paired Fine-Tuning (PFT), designed to align models with contradictory and evolving individual preferences. We present a new dataset, Value Conflict Dilemma (VCD), which includes scenarios that involve conflicting human preferences, facilitating the evaluation of our approach. Our experiments demonstrate that PFT outperforms single-preference training methods, achieving up to 96.6% accuracy in multi-choice classification tasks and the highest open-ended generation score of 8.69. PFT also shows significant improvements over DPO, SFT and some traditional training methods, especially when handling conflicting preferences. Additionally, with limited user history data, models can inferring preference vector rapidly, achieving a 44.76% improvement in user-specific preference alignment in comparison to single-preference models.

---

### Mining Large Language Models for Low-Resource Language Data: Comparing Elicitation Strategies for Hausa and Fongbe

Authors: Mahounan Pericles Adjovi, Roald Eiselen, Prasenjit Mitra
Published: 2026-04-14
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2604.12477v1

Large language models (LLMs) are trained on data contributed by low-resource language communities, yet the linguistic knowledge encoded in these models remains accessible only through commercial APIs. This paper investigates whether strategic prompting can extract usable text data from LLMs for two West African languages: Hausa (Afroasiatic, approximately 80 million speakers) and Fongbe (Niger-Congo, approximately 2 million speakers). We systematically compare six elicitation task types across two commercial LLMs (GPT-4o Mini and Gemini 2.5 Flash). GPT-4o Mini extracts 6-41 times more usable target-language words per API call than Gemini. Optimal strategies differ by language: Hausa benefits from functional text and dialogue, while Fongbe requires constrained generation prompts. We release all generated corpora and code.

---

### From Kinematics to Dynamics: Learning to Refine Hybrid Plans for Physically Feasible Execution

Authors: Lidor Erez, Shahaf S. Shperberg, Ayal Taitler
Published: 2026-04-14
Categories: cs.RO, cs.AI
arXiv: https://arxiv.org/abs/2604.12474v3

In many robotic tasks, agents must traverse a sequence of spatial regions to complete a mission. Such problems are inherently mixed discrete-continuous: a high-level action sequence and a physically feasible continuous trajectory. The resulting trajectory and action sequence must also satisfy problem constraints such as deadlines, time windows, and velocity or acceleration limits. While hybrid temporal planners attempt to address this challenge, they typically model motion using linear (first-order) dynamics, which cannot guarantee that the resulting plan respects the robot's true physical constraints. Consequently, even when the high-level action sequence is fixed, producing a dynamically feasible trajectory becomes a bi-level optimization problem. We address this problem via reinforcement learning in continuous space. We define a Markov Decision Process that explicitly incorporates analytical second-order constraints and use it to refine first-order plans generated by a hybrid planner. Our results show that this approach can reliably recover physical feasibility and effectively bridge the gap between a planner's initial first-order trajectory and the dynamics required for real execution.

---

### Beyond Single-Dimension Novelty: How Combinations of Theory, Method, and Results-based Novelty Shape Scientific Impact

Authors: Yi Zhao, Yang Chenggang, Yuzhuo Wang, Tong Bao, Zhang Heng, Chengzhi Zhang
Published: 2026-04-14
Categories: cs.DL, cs.CL, cs.IR
arXiv: https://arxiv.org/abs/2604.12471v1

Scientific novelty drives advances at the research frontier, yet it is also associated with heightened uncertainty and potential resistance from incumbent paradigms, leading to complex patterns of scientific impact. Prior studies have primarily ex-amined the relationship between a single dimension of novelty -- such as theoreti-cal, methodological, or results-based novelty -- and scientific impact. However, because scientific novelty is inherently multidimensional, focusing on isolated dimensions may obscure how different types of novelty jointly shape impact. Consequently, we know little about how combinations of novelty types influence scientific impact. To this end, we draw on a dataset of 15,322 articles published in Nature Communications. Using the DeepSeek-V3 model, we classify articles into three novelty dimensions based on the content of their Introduction sections: theoretical novelty, methodological novelty, and results-based novelty. These dimensions may coexist within the same article, forming distinct novelty configura-tions. Scientific impact is measured using five-year citation counts and indicators of whether an article belongs to the top 1% or top 10% highly cited papers. Descriptive results indicate that results-based novelty alone and the simultaneous presence of all three novelty types are the dominant configurations in the sample. Regression results further show that articles with results-based novelty only re-ceive significantly more citations and are more likely to rank among the top 1% and top 10% highly cited papers than articles exhibiting all three novelty types. These findings advance our understanding of how multidimensional novelty configurations shape knowledge diffusion.

---

### Intelligent ROI-Based Vehicle Counting Framework for Automated Traffic Monitoring

Authors: Mohamed A. Abdelwahab, Zaynab Al-Ariny, Mahmoud Fakhry, El-Sayed Hasaneen
Published: 2026-04-14
Categories: cs.AI
arXiv: https://arxiv.org/abs/2604.12470v1

Accurate vehicle counting through video surveillance is crucial for efficient traffic management. However, achieving high counting accuracy while ensuring computational efficiency remains a challenge. To address this, we propose a fully automated, video-based vehicle counting framework designed to optimize both computational efficiency and counting accuracy. Our framework operates in two distinct phases: \textit{estimation} and \textit{prediction}. In the estimation phase, the optimal region of interest (ROI) is automatically determined using a novel combination of three models based on detection scores, tracking scores, and vehicle density. This adaptive approach ensures compatibility with any detection and tracking method, enhancing the framework's versatility. In the prediction phase, vehicle counting is efficiently performed within the estimated ROI. We evaluated our framework on benchmark datasets like UA-DETRAC, GRAM, CDnet 2014, and ATON. Results demonstrate exceptional accuracy, with most videos achieving 100\% accuracy, while also enhancing computational efficiency, making processing up to four times faster than full-frame processing. The framework outperforms existing techniques, especially in complex multi-road scenarios, demonstrating robustness and superior accuracy. These advancements make it a promising solution for real-time traffic monitoring.

---

### Analyzing the Effect of Noise in LLM Fine-tuning

Authors: Lingfang Li, Procheta Sen
Published: 2026-04-14
Categories: cs.LG
arXiv: https://arxiv.org/abs/2604.12469v1

Fine-tuning is the dominant paradigm for adapting pretrained large language models (LLMs) to downstream NLP tasks. In practice, fine-tuning datasets may contain various forms of noise arising from annotation errors, preprocessing artifacts, or automated data collection. While prior work has focused on designing robust learning algorithms to mitigate performance degradation under noisy conditions, comparatively little is known about how different types of noise affect the internal learning dynamics of LLMs during fine-tuning. In this work, we systematically study the impact of noise on model behavior across three pretrained model families (GPT-2, Qwen2 and Llama-2) and three diverse NLP tasks. We introduce controlled perturbations corresponding to three common real-world noise types: label noise, grammatical noise, and typographical noise. Beyond task-level performance, we analyze layer-wise representation changes and attention patterns to understand how noise propagates through the network. Our results show that corrupting labels (i.e. label noise) consistently causes the largest performance degradation, whereas grammatical noise and typographical noise can occasionally yield mild regularization benefits. We further find that noise effects are localized primarily to task-specific layers, while attention structures remain comparatively stable.

---

### Euler-inspired Decoupling Neural Operator for Efficient Pansharpening

Authors: Anqi Zhu, Mengting Ma, Yizhen Jiang, Xiangdong Li, Kai Zheng, Jiaxin Li, Wei Zhang
Published: 2026-04-14
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2604.12463v1

Pansharpening aims to synthesize high-resolution multispectral (HR-MS) images by fusing the spatial textures of panchromatic (PAN) images with the spectral information of low-resolution multispectral (LR-MS) images. While recent deep learning paradigms, especially diffusion-based operators, have pushed the performance boundaries, they often encounter spectral-spatial blurring and prohibitive computational costs due to their stochastic nature and iterative sampling. In this paper, we propose the Euler-inspired Decoupling Neural Operator (EDNO), a physics-inspired framework that redefines pansharpening as a continuous functional mapping in the frequency domain. Departing from conventional Cartesian feature processing, our EDNO leverages Euler's formula to transform features into a polar coordinate system, enabling a novel explicit-implicit interaction mechanism. Specifically, we develop the Euler Feature Interaction Layer (EFIL), which decouples the fusion task into two specialized modules: 1) Explicit Feature Interaction Module, utilizing a linear weighting scheme to simulate phase rotation for adaptive geometric alignment; and 2) Implicit Feature Interaction Module, employing a feed-forward network to model spectral distributions for superior color consistency. By operating in the frequency domain, EDNO inherently captures global receptive fields while maintaining discretization-invariance. Experimental results on the three datasets demonstrate that EDNO offers a superior efficiency-performance balance compared to heavyweight architectures.

---

### CIA: Inferring the Communication Topology from LLM-based Multi-Agent Systems

Authors: Yongxuan Wu, Xixun Lin, He Zhang, Nan Sun, Kun Wang, Chuan Zhou, Shirui Pan, Yanan Cao
Published: 2026-04-14
Categories: cs.AI
arXiv: https://arxiv.org/abs/2604.12461v1

LLM-based Multi-Agent Systems (MAS) have demonstrated remarkable capabilities in solving complex tasks. Central to MAS is the communication topology which governs how agents exchange information internally. Consequently, the security of communication topologies has attracted increasing attention. In this paper, we investigate a critical privacy risk: MAS communication topologies can be inferred under a restrictive black-box setting, exposing system vulnerabilities and posing significant intellectual property threats. To explore this risk, we propose Communication Inference Attack (CIA), a novel attack that constructs new adversarial queries to induce intermediate agents' reasoning outputs and models their semantic correlations through the proposed global bias disentanglement and LLM-guided weak supervision. Extensive experiments on MAS with optimized communication topologies demonstrate the effectiveness of CIA, achieving an average AUC of 0.87 and a peak AUC of up to 0.99, thereby revealing the substantial privacy risk in MAS.

---

### Enhancing Clustering: An Explainable Approach via Filtered Patterns

Authors: Motaz Ben Hassine, Saïd Jabbour
Published: 2026-04-14
Categories: cs.AI
arXiv: https://arxiv.org/abs/2604.12460v1

Machine learning has become a central research area, with increasing attention devoted to explainable clustering, also known as conceptual clustering, which is a knowledge-driven unsupervised learning paradigm that partitions data into $θ$ disjoint clusters, where each cluster is described by an explicit symbolic representation, typically expressed as a closed pattern or itemset. By providing human-interpretable cluster descriptions, explainable clustering plays an important role in explainable artificial intelligence and knowledge discovery. Recent work improved clustering quality by introducing k-relaxed frequent patterns (k-RFPs), a pattern model that relaxes strict coverage constraints through a generalized kcover definition. This framework integrates constraint-based reasoning, using SAT solvers for pattern generation, with combinatorial optimization, using Integer Linear Programming (ILP) for cluster selection. Despite its effectiveness, this approach suffers from a critical limitation: multiple distinct k-RFPs may induce identical k-covers, leading to redundant symbolic representations that unnecessarily enlarge the search space and increase computational complexity during cluster construction. In this paper, we address this redundancy through a pattern reduction framework. Our contributions are threefold. First, we formally characterize the conditions under which distinct k-RFPs induce identical kcovers, providing theoretical foundations for redundancy detection. Second, we propose an optimization strategy that removes redundant patterns by retaining a single representative pattern for each distinct k-cover. Third, we investigate the interpretability and representativeness of the patterns selected by the ILP model by analyzing their robustness with respect to their induced clusters. Extensive experiments conducted on several real-world datasets demonstrate that the proposed approach significantly reduces the pattern search space, improves computational effi…

---

### Operationalising the Right to be Forgotten in LLMs: A Lightweight Sequential Unlearning Framework for Privacy-Aligned Deployment in Politically Sensitive Environments

Authors: Esen Kurt, Haithem Afli
Published: 2026-04-14
Categories: cs.AI
arXiv: https://arxiv.org/abs/2604.12459v1

Large Language Models (LLMs) are increasingly deployed in politically sensitive environments, where memorisation of personal data or confidential content raises regulatory concerns under frameworks such as the GDPR and its Right to be Forgotten. Translating such legal principles into large-scale generative systems presents significant technical challenges. We introduce a lightweight sequential unlearning framework that explicitly separates retention and suppression objectives. The method first stabilises benign capabilities through positive fine-tuning, then applies layer-restricted negative fine-tuning to suppress designated sensitive patterns while preserving general language competence. Experiments on the SemEval-2025 LLM Unlearning benchmark demonstrate effective behavioural suppression with minimal impact on factual accuracy and fluency. GPT-2 exhibits greater robustness than DistilGPT-2, highlighting the role of model capacity in privacy-aligned adaptation. We position sequential unlearning as a practical and reproducible mechanism for operationalising data erasure requirements in politically deployed LLMs.

---

### X-VC: Zero-shot Streaming Voice Conversion in Codec Space

Authors: Qixi Zheng, Yuxiang Zhao, Tianrui Wang, Wenxi Chen, Kele Xu, Yikang Li, Qinyuan Cheng, Xipeng Qiu, Kai Yu, Xie Chen
Published: 2026-04-14
Categories: eess.AS, cs.AI
arXiv: https://arxiv.org/abs/2604.12456v3

Zero-shot voice conversion (VC) aims to convert a source utterance into the voice of an unseen target speaker while preserving its linguistic content. Although recent systems have improved conversion quality, building zero-shot VC systems for interactive scenarios remains challenging because high-fidelity speaker transfer and low-latency streaming inference are difficult to achieve simultaneously. In this work, we present X-VC, a zero-shot streaming VC system that performs one-step conversion in the latent space of a pretrained neural codec. X-VC uses a dual-conditioning acoustic converter that jointly models source codec latents and frame-level acoustic conditions derived from target reference speech, while injecting utterance-level target speaker information through adaptive normalization. To reduce the mismatch between training and inference, we train the model with generated paired data and a role-assignment strategy that combines standard, reconstruction, and reversed modes. For streaming inference, we further adopt a chunkwise inference scheme with overlap smoothing that is aligned with the segment-based training paradigm of the codec. Experiments on Seed-TTS-Eval show that X-VC achieves the best streaming WER in both English and Chinese, strong speaker similarity in same-language and cross-lingual settings, and substantially lower offline real-time factor than the compared baselines. These results suggest that codec-space one-step conversion is a practical approach for building high-quality low-latency zero-shot VC systems. Our audio samples, code and checkpoints are released at https://github.com/Jerrister/X-VC.

---

### X-VC: Zero-shot Streaming Voice Conversion in Codec Space

Authors: Qixi Zheng, Yuxiang Zhao, Tianrui Wang, Wenxi Chen, Kele Xu, Yikang Li, Qinyuan Chen, Xipeng Qiu, Kai Yu, Xie Chen
Published: 2026-04-14
Categories: eess.AS, cs.AI
arXiv: https://arxiv.org/abs/2604.12456v2

Zero-shot voice conversion (VC) aims to convert a source utterance into the voice of an unseen target speaker while preserving its linguistic content. Although recent systems have improved conversion quality, building zero-shot VC systems for interactive scenarios remains challenging because high-fidelity speaker transfer and low-latency streaming inference are difficult to achieve simultaneously. In this work, we present X-VC, a zero-shot streaming VC system that performs one-step conversion in the latent space of a pretrained neural codec. X-VC uses a dual-conditioning acoustic converter that jointly models source codec latents and frame-level acoustic conditions derived from target reference speech, while injecting utterance-level target speaker information through adaptive normalization. To reduce the mismatch between training and inference, we train the model with generated paired data and a role-assignment strategy that combines standard, reconstruction, and reversed modes. For streaming inference, we further adopt a chunkwise inference scheme with overlap smoothing that is aligned with the segment-based training paradigm of the codec. Experiments on Seed-TTS-Eval show that X-VC achieves the best streaming WER in both English and Chinese, strong speaker similarity in same-language and cross-lingual settings, and substantially lower offline real-time factor than the compared baselines. These results suggest that codec-space one-step conversion is a practical approach for building high-quality low-latency zero-shot VC systems. Our audio samples, code and checkpoints are released at https://github.com/Jerrister/X-VC.

---

### Latent-Condensed Transformer for Efficient Long Context Modeling

Authors: Zeng You, Yaofo Chen, Qiuwu Chen, Ying Sun, Shuhai Zhang, Yingjian Li, Yaowei Wang, Mingkui Tan
Published: 2026-04-14
Categories: cs.CL
arXiv: https://arxiv.org/abs/2604.12452v2

Large language models (LLMs) face significant challenges in processing long contexts due to the linear growth of the key-value (KV) cache and quadratic complexity of self-attention. Existing approaches address these bottlenecks separately: Multi-head Latent Attention (MLA) reduces the KV cache by projecting tokens into a low-dimensional latent space, while sparse attention reduces computation. However, sparse methods cannot operate natively on MLA's compressed latent structure, missing opportunities for joint optimization. In this paper, we propose Latent-Condensed Attention (LCA), which directly condenses context within MLA's latent space, where the representation is disentangled into semantic latent vectors and positional keys. LCA separately aggregates semantic vectors via query-aware pooling and preserves positional keys via anchor selection. This approach jointly reduces both computational cost and KV cache without adding parameters. Beyond MLA, LCA's design is architecture-agnostic and readily extends to other attention mechanisms such as GQA. Theoretically, we prove a length-independent error bound. Experiments show LCA achieves up to 2.5$\times$ prefilling speedup and 90% KV cache reduction at 128K context while maintaining competitive performance.

---

### Scaling Exposes the Trigger: Input-Level Backdoor Detection in Text-to-Image Diffusion Models via Cross-Attention Scaling

Authors: Zida Li, Jun Li, Yuzhe Sha, Ziqiang Li, Lizhi Xiong, Zhangjie Fu
Published: 2026-04-14
Categories: cs.CR, cs.CV
arXiv: https://arxiv.org/abs/2604.12446v1

Text-to-image (T2I) diffusion models have achieved remarkable success in image synthesis, but their reliance on large-scale data and open ecosystems introduces serious backdoor security risks. Existing defenses, particularly input-level methods, are more practical for deployment but often rely on observable anomalies that become unreliable under stealthy, semantics-preserving trigger designs. As modern backdoor attacks increasingly embed triggers into natural inputs, these methods degrade substantially, raising a critical question: can more stable, implicit, and trigger-agnostic differences between benign and backdoor inputs be exploited for detection? In this work, we address this challenge from an active probing perspective. We introduce controlled scaling perturbations on cross-attention and uncover a novel phenomenon termed Cross-Attention Scaling Response Divergence (CSRD), where benign and backdoor inputs exhibit systematically different response evolution patterns across denoising steps. Building on this insight, we propose SET, an input-level backdoor detection framework that constructs response-offset features under multi-scale perturbations and learns a compact benign response space from a small set of clean samples. Detection is then performed by measuring deviations from this learned space, without requiring prior knowledge of the attack or access to model training. Extensive experiments demonstrate that SET consistently outperforms existing baselines across diverse attack methods, trigger types, and model settings, with particularly strong gains under stealthy implicit-trigger scenarios. Overall, SET improves AUROC by 9.1% and ACC by 6.5% over the best baseline, highlighting its effectiveness and robustness for practical deployment.

---

### DiffusionPrint: Learning Generative Fingerprints for Diffusion-Based Inpainting Localization

Authors: Paschalis Giakoumoglou, Symeon Papadopoulos
Published: 2026-04-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2604.12443v1

Modern diffusion-based inpainting models pose significant challenges for image forgery localization (IFL), as their full regeneration pipelines reconstruct the entire image via a latent decoder, disrupting the camera-level noise patterns that existing forensic methods rely on. We propose DiffusionPrint, a patch-level contrastive learning framework that learns a forensic signal robust to the spectral distortions introduced by latent decoding. It exploits the fact that inpainted regions generated by the same model share a consistent generative fingerprint, using this as a self-supervisory signal. DiffusionPrint trains a convolutional backbone via a MoCo-style objective with cross-category hard negative mining and a generator-aware classification head, producing a forensic feature map that serves as a highly discriminative secondary modality in fusion-based IFL frameworks. Integrated into TruFor, MMFusion, and a lightweight fusion baseline, DiffusionPrint consistently improves localization across multiple generative models, with gains of up to +28% on mask types unseen during fine-tuning and confirmed generalization to unseen generative architectures. Code is available at https://github.com/mever-team/diffusionprint

---

### GLeMM: A large-scale multilingual dataset for morphological research

Authors: Hathout Nabil, Basilio Calderone, Fiammetta Namer, Franck Sajous
Published: 2026-04-14
Categories: cs.CL
arXiv: https://arxiv.org/abs/2604.12442v1

In derivational morphology, what mechanisms govern the variation in form-meaning relations between words? The answers to this type of questions are typically based on intuition and on observations drawn from limited data, even when a wide range of languages is considered. Many of these studies are difficult to replicate and generalize. To address this issue, we present GLeMM, a new derivational resource designed for experimentation and data-driven description in morphology. GLeMM is characterized by (i) its large size, (ii) its extensive coverage (currently amounting to seven European languages, i.e., German, English, Spanish, French, Italian, Polish, Russian, (iii) its fully automated design, identical across all languages, (iv) the automatic annotation of morphological features on each entry, as well as (v) the encoding of semantic descriptions for a significant subset of these entries. It enables researchers to address difficult questions, such as the role of form and meaning in word-formation, and to develop and experimentally test computational methods that identify the structures of derivational morphology. The article describes how GLeMM is created using Wiktionary articles and presents various case studies illustrating possible applications of the resource.

---

### IAD-Unify: A Region-Grounded Unified Model for Industrial Anomaly Segmentation, Understanding, and Generation

Authors: Haoyu Zheng, Tianwei Lin, Wei Wang, Zhuonan Wang, Wenqiao Zhang, Jiaqi Zhu, Feifei Shao
Published: 2026-04-14
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2604.12440v1

Real-world industrial inspection requires not only localizing defects, but also explaining them in natural language and generating controlled defect edits. However, existing approaches fail to jointly support all three capabilities within a unified framework and evaluation protocol. We propose IAD-Unify, a dual-encoder unified framework in which a frozen DINOv2-based region expert supplies precise anomaly evidence to a shared Qwen3.5-4B vision-language backbone via lightweight token injection, jointly enabling anomaly segmentation, region-grounded understanding, and mask-guided generation. To enable unified evaluation, we further construct Anomaly-56K, a comprehensive unified multi-task IAD evaluation platform, spanning 59,916 images across 24 categories and 104 defect variants. Controlled ablations yield four findings: (i) region grounding is the decisive mechanism for understanding, removing it degrades location accuracy by >76 pp; (ii) predicted-region performance closely matches oracle, confirming deployment viability; (iii) region-grounded generation achieves the best full-image fidelity and masked-region perceptual quality; and (iv) pre-initialized joint training improves understanding at negligible generation cost (-0.16 dB). IAD-Unify further achieves strong performance on the MMAD benchmark, including categories unseen during training, demonstrating robust cross-category generalization.

---

### A Hybrid Architecture for Benign-Malignant Classification of Mammography ROIs

Authors: Mohammed Asad, Mohit Bajpai, Sudhir Singh, Rahul Katarya
Published: 2026-04-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2604.12437v1

Accurate characterization of suspicious breast lesions in mammography is important for early diagnosis and treatment planning. While Convolutional Neural Networks (CNNs) are effective at extracting local visual patterns, they are less suited to modeling long-range dependencies. Vision Transformers (ViTs) address this limitation through self-attention, but their quadratic computational cost can be prohibitive. This paper presents a hybrid architecture that combines EfficientNetV2-M for local feature extraction with Vision Mamba, a State Space Model (SSM), for efficient global context modeling. The proposed model performs binary classification of abnormality-centered mammography regions of interest (ROIs) from the CBIS-DDSM dataset into benign and malignant classes. By combining a strong CNN backbone with a linear-complexity sequence model, the approach achieves strong lesion-level classification performance in an ROI-based setting.

---

### A Bayesian Perspective on the Role of Epistemic Uncertainty for Delayed Generalization in In-Context Learning

Authors: Abdessamed Qchohi, Simone Rossi
Published: 2026-04-14
Categories: stat.ML, cs.LG
arXiv: https://arxiv.org/abs/2604.12434v1

In-context learning enables transformers to adapt to new tasks from a few examples at inference time, while grokking highlights that this generalization can emerge abruptly only after prolonged training. We study task generalization and grokking in in-context learning using a Bayesian perspective, asking what enables the delayed transition from memorization to generalization. Concretely, we consider modular arithmetic tasks in which a transformer must infer a latent linear function solely from in-context examples and analyze how predictive uncertainty evolves during training. We combine approximate Bayesian techniques to estimate the posterior distribution and we study how uncertainty behaves across training and under changes in task diversity, context length, and context noise. We find that epistemic uncertainty collapses sharply when the model groks, making uncertainty a practical label-free diagnostic of generalization in transformers. Additionally, we provide theoretical support with a simplified Bayesian linear model, showing that asymptotically both delayed generalization and uncertainty peaks arise from the same underlying spectral mechanism, which links grokking time to uncertainty dynamics.

---

### VeriX-Anon: A Multi-Layered Framework for Mathematically Verifiable Outsourced Target-Driven Data Anonymization

Authors: Miit Daga, Swarna Priya Ramu
Published: 2026-04-14
Categories: cs.CR, cs.DB, cs.LG
arXiv: https://arxiv.org/abs/2604.12431v3

Organisations increasingly outsource privacy-sensitive data transformations to cloud providers, yet no practical mechanism lets the data owner verify that the contracted algorithm was faithfully executed. VeriX-Anon is a multi-layered verification framework for outsourced Target-Driven k-anonymization combining three orthogonal mechanisms: deterministic verification via Merkle-style hashing of an Authenticated Decision Tree, probabilistic verification via Boundary Sentinels and exact-duplicate Twins with cryptographic identifiers, and utility-based verification via Explainable AI fingerprinting that compares SHAP value distributions before and after anonymization using the Wasserstein distance. Across seven cross-domain datasets and four cloud profiles (28 scenarios), against Lazy (drops records), Dumb (fake hash), and Approximate (valid hash) adversaries, VeriX-Anon detects 25 of 28 deviations under a fixed threshold and 27 of 28 once the threshold is calibrated per dataset, with no false alarms. No single layer achieved this alone. The XAI layer was the only mechanism that caught the Approximate adversary, succeeding on six of seven datasets and missing only a high-dimensional case where honest generalization shifts SHAP as much as the attack. Target-Driven anonymization preserved significantly more utility than blind splitting, with mean F1 gaps of +0.058 to +0.362 and Wilcoxon p <= 0.001 on six of seven datasets. Client-side verification completes under one second at one million rows. The threat model covers three empirically evaluated profiles and one theoretical Informed Attacker unable to defeat the cryptographic salt. Sentinel evasion probability ranges from near-zero to 0.82 for the most imbalanced data, which the twin layer offsets in every scenario.

---

### VeriX-Anon: A Multi-Layered Framework for Mathematically Verifiable Outsourced Target-Driven Data Anonymization

Authors: Miit Daga, Swarna Priya Ramu
Published: 2026-04-14
Categories: cs.CR, cs.DB, cs.LG
arXiv: https://arxiv.org/abs/2604.12431v2

Organisations increasingly outsource privacy-sensitive data transformations to cloud providers, yet no practical mechanism lets the data owner verify that the contracted algorithm was faithfully executed. VeriX-Anon is a multi-layered verification framework for outsourced Target-Driven k-anonymization combining three orthogonal mechanisms: deterministic verification via Merkle-style hashing of an Authenticated Decision Tree, probabilistic verification via Boundary Sentinels and exact-duplicate Twins with cryptographic identifiers, and utility-based verification via Explainable AI fingerprinting that compares SHAP value distributions before and after anonymization using the Wasserstein distance. Across seven cross-domain datasets and four cloud profiles (28 scenarios), against Lazy (drops records), Dumb (fake hash), and Approximate (valid hash) adversaries, VeriX-Anon detects 25 of 28 deviations under a fixed threshold and 27 of 28 once the threshold is calibrated per dataset, with no false alarms. No single layer achieved this alone. The XAI layer was the only mechanism that caught the Approximate adversary, succeeding on six of seven datasets and missing only a high-dimensional case where honest generalization shifts SHAP as much as the attack. Target-Driven anonymization preserved significantly more utility than blind splitting, with mean F1 gaps of +0.058 to +0.362 and Wilcoxon p <= 0.001 on six of seven datasets. Client-side verification completes under one second at one million rows. The threat model covers three empirically evaluated profiles and one theoretical Informed Attacker unable to defeat the cryptographic salt. Sentinel evasion probability ranges from near-zero to 0.82 for the most imbalanced data, which the twin layer offsets in every scenario.

---

### Do Transformers Use their Depth Adaptively? Evidence from a Relational Reasoning Task

Authors: Alicia Curth, Rachel Lawrence, Sushrut Karmalkar, Niranjani Prasad
Published: 2026-04-14
Categories: cs.LG, cs.CL
arXiv: https://arxiv.org/abs/2604.12426v1

We investigate whether transformers use their depth adaptively across tasks of increasing difficulty. Using a controlled multi-hop relational reasoning task based on family stories, where difficulty is determined by the number of relationship hops that must be composed, we monitor (i) how predictions evolve across layers via early readouts (the logit lens) and (ii) how task-relevant information is integrated across tokens via causal patching. For pretrained models, we find some limited evidence for adaptive depth use: some larger models need fewer layers to arrive at plausible answers for easier tasks, and models generally use more layers to integrate information across tokens as chain length increases. For models finetuned on the task, we find clearer and more consistent evidence of adaptive depth use, with the effect being stronger for less constrained finetuning regimes that do not preserve general language modeling abilities.

---

### Forecasting the Past: Gradient-Based Distribution Shift Detection in Trajectory Prediction

Authors: Michele De Vita, Julian Wiederer, Vasileios Belagiannis
Published: 2026-04-14
Categories: cs.LG
arXiv: https://arxiv.org/abs/2604.12425v1

Trajectory prediction models often fail in real-world automated driving due to distributional shifts between training and test conditions. Such distributional shifts, whether behavioural or environmental, pose a critical risk by causing the model to make incorrect forecasts in unfamiliar situations. We propose a self-supervised method that trains a decoder in a post-hoc fashion on the self-supervised task of forecasting the second half of observed trajectories from the first half. The L2 norm of the gradient of this forecasting loss with respect to the decoder's final layer defines a score to identify distribution shifts. Our approach, first, does not affect the trajectory prediction model, ensuring no interference with original prediction performance and second, demonstrates substantial improvements on distribution shift detection for trajectory prediction on the Shifts and Argoverse datasets. Moreover, we show that this method can also be used to early detect collisions of a deep Q-Network motion planner in the Highway simulator. Source code is available at https://github.com/Michedev/forecasting-the-past.

---

### Decoding by Perturbation: Mitigating MLLM Hallucinations via Dynamic Textual Perturbation

Authors: Sihang Jia, Shuliang Liu, Songbo Yang, Yibo Yan, Xin Zou, Xuming Hu
Published: 2026-04-14
Categories: cs.CL, cs.AI, cs.CV
arXiv: https://arxiv.org/abs/2604.12424v1

Multimodal Large Language Models frequently suffer from inference hallucinations, partially stemming from language priors dominating visual evidence. Existing training-free mitigation methods either perturb the visual representation and deviate from the natural image distribution, or enforce intrusive manipulations that compromise the model's inherent generative fluency. We introduce a novel perspective that multimodal hallucination manifests as the hypersensitivity of visual grounding to textual phrasing during the decoding phase. Building on this insight, we propose Decoding by Perturbation (DeP), a training-free framework mitigating prior-induced hallucinations via controlled textual interventions. DeP employs a dynamic probe applying multi-level textual perturbations to elicit latent language priors. Leveraging attention variance, it enhances stable evidence regions while suppressing suspicious noise in the feature space. Furthermore, it constructs an interpretable prior drift direction using logits statistics to counteract probability biases from textual co-occurrences. Extensive experiments confirm DeP effectively reduces hallucinations and achieves superior performance across multiple benchmarks.

---

### Agentic Insight Generation in VSM Simulations

Authors: Micha Selak, Dirk Krechel, Adrian Ulges, Sven Spieckermann, Niklas Stoehr, Andreas Loehr
Published: 2026-04-14
Categories: cs.CL
arXiv: https://arxiv.org/abs/2604.12421v1

Extracting actionable insights from complex value stream map simulations can be challenging, time-consuming, and error-prone. Recent advances in large language models offer new avenues to support users with this task. While existing approaches excel at processing raw data to gain information, they are structurally unfit to pick up on subtle situational differences needed to distinguish similar data sources in this domain. To address this issue, we propose a decoupled, two-step agentic architecture. By separating orchestration from data analysis, the system leverages progressive data discovery infused with domain expert knowledge. This architecture allows the orchestration to intelligently select data sources and perform multi-hop reasoning across data structures while maintaining a slim internal context. Results from multiple state-of-the-art large language models demonstrate the framework's viability: with top-tier models achieving accuracies of up to 86% and demonstrating high robustness across evaluation runs.

---

### RACF: A Resilient Autonomous Car Framework with Object Distance Correction

Authors: Chieh Tsai, Hossein Rastgoftar, Salim Hariri
Published: 2026-04-14
Categories: cs.RO, cs.AI
arXiv: https://arxiv.org/abs/2604.12418v1

Autonomous vehicles are increasingly deployed in safety-critical applications, where sensing failures or cyberphysical attacks can lead to unsafe operations resulting in human loss and/or severe physical damages. Reliable real-time perception is therefore critically important for their safe operations and acceptability. For example, vision-based distance estimation is vulnerable to environmental degradation and adversarial perturbations, and existing defenses are often reactive and too slow to promptly mitigate their impacts on safe operations. We present a Resilient Autonomous Car Framework (RACF) that incorporates an Object Distance Correction Algorithm (ODCA) to improve perception-layer robustness through redundancy and diversity across a depth camera, LiDAR, and physics-based kinematics. Within this framework, when obstacle distance estimation produced by depth camera is inconsistent, a cross-sensor gate activates the correction algorithm to fix the detected inconsistency. We have experiment with the proposed resilient car framework and evaluate its performance on a testbed implemented using the Quanser QCar 2 platform. The presented framework achieved up to 35% RMSE reduction under strong corruption and improves stop compliance and braking latency, while operating in real time. These results demonstrate a practical and lightweight approach to resilient perception for safety-critical autonomous driving

---

### Machine learning for four-dimensional SU(3) lattice gauge theories

Authors: Urs Wenger
Published: 2026-04-14
Categories: hep-lat, cs.LG
arXiv: https://arxiv.org/abs/2604.12416v1

In this review I summarize how machine learning can be used in lattice gauge theory simulations and what ap\-proaches are currently available to improve the sampling of gauge field configurations, with a focus on applications in four-dimensional SU(3) gauge theories. These include approaches based on generative machine-learning models such as (stochastic) normalizing flows and diffusion processes, and an approach based on renormalization group (RG) transformations, more specifically the machine learning of RG-improved gauge actions using gauge-equivariant convolutional neural networks. In particular, I present scaling results for a machine-learned fixed-point action in four-dimensional SU(3) gauge theory towards the continuum limit. The results include observables based on the classically perfect gradient-flow scales, which are free of tree-level lattice artefacts to all orders, and quantities related to the static potential and the deconfinement transition.

---

### DeferredSeg:A Multi-Expert Deferral Framework for Medical Image Segmentation

Authors: Qiuyu Tian, Haoliang Sun, Yunshan Wang, Yinghuan Shi, Yilong Yin
Published: 2026-04-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2604.12411v2

Segmentation models based on deep neural networks demonstrate strong generalization for medical image segmentation. However, they often exhibit overconfidence or underconfidence, leading to unreliable confidence scores for segmentation masks, especially in ambiguous regions. This undermines the trustworthiness required for clinical deployment. Motivated by the learning-to-defer (L2D) paradigm, we introduce DeferredSeg, a deferral-aware segmentation framework, i.e., a Human--AI collaboration system that determines whether to defer predictions to human experts in specific regions. DeferredSeg extends the base segmentor with an aggregated deferral predictor and additional routing channels that dynamically route each pixel to either the base segmentor or a human expert. To train this routing efficiently, we introduce a pixel-wise surrogate collaboration loss that supervises deferral decisions. In addition, to preserve spatial coherence within deferral regions, we propose a spatial-coherence loss that enforces smooth deferral masks, thereby enhancing reliability. Beyond single-expert deferral, we further extend the framework to a multi-expert setting by introducing multiple discrepancy experts for collaborative decision-making. To prevent overloading or underutilizing individual experts, we further design a load-balancing penalty that evenly distributes workload across expert branches. We evaluate DeferredSeg on three challenging medical datasets using MedSAM and CENet as the base segmentor for fair comparison. Experimental results show that DeferredSeg consistently outperforms the baseline, demonstrating its effectiveness for trustworthy dense medical segmentation. Moreover, the proposed framework is model-agnostic and can be readily applied to other segmentation architectures.

---

### DeferredSeg: A Multi-Expert Deferral Framework for Trustworthy Medical Image Segmentation

Authors: Qiuyu Tian, Haoliang Sun, Yunshan Wang, Yinghuan Shi, Yilong Yin
Published: 2026-04-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2604.12411v1

Segmentation models based on deep neural networks demonstrate strong generalization for medical image segmentation. However, they often exhibit overconfidence or underconfidence, leading to unreliable confidence scores for segmentation masks, especially in ambiguous regions. This undermines the trustworthiness required for clinical deployment. Motivated by the learning-to-defer (L2D) paradigm, we introduce DeferredSeg, a deferral-aware segmentation framework, i.e., a Human--AI collaboration system that determines whether to defer predictions to human experts in specific regions. DeferredSeg extends the base segmentor with an aggregated deferral predictor and additional routing channels that dynamically route each pixel to either the base segmentor or a human expert. To train this routing efficiently, we introduce a pixel-wise surrogate collaboration loss that supervises deferral decisions. In addition, to preserve spatial coherence within deferral regions, we propose a spatial-coherence loss that enforces smooth deferral masks, thereby enhancing reliability. Beyond single-expert deferral, we further extend the framework to a multi-expert setting by introducing multiple discrepancy experts for collaborative decision-making. To prevent overloading or underutilizing individual experts, we further design a load-balancing penalty that evenly distributes workload across expert branches. We evaluate DeferredSeg on three challenging medical datasets using MedSAM and CENet as the base segmentor for fair comparison. Experimental results show that DeferredSeg consistently outperforms the baseline, demonstrating its effectiveness for trustworthy dense medical segmentation. Moreover, the proposed framework is model-agnostic and can be readily applied to other segmentation architectures.

---

### Security and Resilience in Autonomous Vehicles: A Proactive Design Approach

Authors: Chieh Tsai, Murad Mehrab Abrar, Salim Hariri
Published: 2026-04-14
Categories: cs.CR, cs.AI
arXiv: https://arxiv.org/abs/2604.12408v1

Autonomous vehicles (AVs) promise efficient, clean and cost-effective transportation systems, but their reliance on sensors, wireless communications, and decision-making systems makes them vulnerable to cyberattacks and physical threats. This chapter presents novel design techniques to strengthen the security and resilience of AVs. We first provide a taxonomy of potential attacks across different architectural layers, from perception and control manipulation to Vehicle-to-Any (V2X) communication exploits and software supply chain compromises. Building on this analysis, we present an AV Resilient architecture that integrates redundancy, diversity, and adaptive reconfiguration strategies, supported by anomaly- and hash-based intrusion detection techniques. Experimental validation on the Quanser QCar platform demonstrates the effectiveness of these methods in detecting depth camera blinding attacks and software tampering of perception modules. The results highlight how fast anomaly detection combined with fallback and backup mechanisms ensures operational continuity, even under adversarial conditions. By linking layered threat modeling with practical defense implementations, this work advances AV resilience strategies for safer and more trustworthy autonomous vehicles.

---

### Dual-Modality Anchor-Guided Filtering for Test-time Prompt Tuning

Authors: Jungwon Choi, Eunwoo Kim
Published: 2026-04-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2604.12403v2

Test-Time Prompt Tuning (TPT) adapts vision-language models using augmented views, but its effectiveness is hindered by the challenge of determining which views are beneficial. Standard entropy-based filtering relies on the internal confidence scores of the model, which are often miscalibrated under distribution shift, assigning high confidence to irrelevant crops or background regions while ignoring semantic content. To address this, we propose a dual-modality anchor-guided framework that grounds view selection in semantic evidence. We introduce a text anchor from attribute-rich descriptions, to provide fine-grained class semantics, and an adaptive image anchor that captures evolving test-time statistics. Using these anchors, we filter views based on alignment and confidence, ensuring that only informative views guide adaptation. Moreover, we treat the anchors as auxiliary predictive heads and combine their predictions with the original output in a confidence-weighted ensemble, yielding a stable supervision signal for prompt updates. Extensive experiments on 15 benchmark datasets demonstrate new state-of-the-art performance, highlighting the contribution of anchor-guided supervision as a foundation for robust prompt updates.

---

### Dual-Modality Anchor-Guided Filtering for Test-time Prompt Tuning

Authors: Jungwon Choi, Eunwoo Kim
Published: 2026-04-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2604.12403v1

Test-Time Prompt Tuning (TPT) adapts vision-language models using augmented views, but its effectiveness is hindered by the challenge of determining which views are beneficial. Standard entropy-based filtering relies on the internal confidence scores of the model, which are often miscalibrated under distribution shift, assigning high confidence to irrelevant crops or background regions while ignoring semantic content. To address this, we propose a dual-modality anchor-guided framework that grounds view selection in semantic evidence. We introduce a text anchor from attribute-rich descriptions, to provide fine-grained class semantics, and an adaptive image anchor that captures evolving test-time statistics. Using these anchors, we filter views based on alignment and confidence, ensuring that only informative views guide adaptation. Moreover, we treat the anchors as auxiliary predictive heads and combine their predictions with the original output in a confidence-weighted ensemble, yielding a stable supervision signal for prompt updates. Extensive experiments on 15 benchmark datasets demonstrate new state-of-the-art performance, highlighting the contribution of anchor-guided supervision as a foundation for robust prompt updates.

---

### KoCo: Conditioning Language Model Pre-training on Knowledge Coordinates

Authors: Yudong Li, Jiawei Cai, Linlin Shen
Published: 2026-04-14
Categories: cs.CL
arXiv: https://arxiv.org/abs/2604.12397v1

Standard Large Language Model (LLM) pre-training typically treats corpora as flattened token sequences, often overlooking the real-world context that humans naturally rely on to contextualize information. To bridge this gap, we introduce Knowledge Coordinate Conditioning (KoCo), a simple method that maps every document into a three-dimensional semantic coordinate. By prepending these coordinates as textual prefixes for pre-training, we aim to equip the model with explicit contextual awareness to learn the documents within the real-world knowledge structure. Experiment results demonstrate that KoCo significantly enhances performance across 10 downstream tasks and accelerates pre-training convergence by approximately 30\%. Furthermore, our analysis indicates that explicitly modeling knowledge coordinates helps the model distinguish stable facts from noise, effectively mitigating hallucination in generated outputs.

---

### Chain-of-Models Pre-Training: Rethinking Training Acceleration of Vision Foundation Models

Authors: Jiawei Fan, Shigeng Wang, Chao Li, Xiaolong Liu, Anbang Yao
Published: 2026-04-14
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2604.12391v1

In this paper, we present Chain-of-Models Pre-Training (CoM-PT), a novel performance-lossless training acceleration method for vision foundation models (VFMs). This approach fundamentally differs from existing acceleration methods in its core motivation: rather than optimizing each model individually, CoM-PT is designed to accelerate the training pipeline at the model family level, scaling efficiently as the model family expands. Specifically, CoM-PT establishes a pre-training sequence for the model family, arranged in ascending order of model size, called model chain. In this chain, only the smallest model undergoes standard individual pre-training, while the other models are efficiently trained through sequential inverse knowledge transfer from their smaller predecessors by jointly reusing the knowledge in the parameter space and the feature space. As a result, CoM-PT enables all models to achieve performance that is mostly superior to standard individual training while significantly reducing training cost, and this is extensively validated across 45 datasets spanning zero-shot and fine-tuning tasks. Notably, its efficient scaling property yields a remarkable phenomenon: training more models even results in higher efficiency. For instance, when pre-training on CC3M: i) given ViT-L as the largest model, progressively prepending smaller models to the model chain reduces computational complexity by up to 72%; ii) within a fixed model size range, as the VFM family scales across 3, 4, and 7 models, the acceleration ratio of CoM-PT exhibits a striking leap: from 4.13X to 5.68X and 7.09X. Since CoM-PT is naturally agnostic to specific pre-training paradigms, we open-source the code to spur further extensions in more computationally intensive scenarios, such as large language model pre-training.

---

### Heuristic Classification of Thoughts Prompting (HCoT): Integrating Expert System Heuristics for Structured Reasoning into Large Language Models

Authors: Lei Lin, Jizhao Zhu, Yong Liu, Donghong Sun, Hongbo He, Yihua Du
Published: 2026-04-14
Categories: cs.AI
arXiv: https://arxiv.org/abs/2604.12390v3

This paper addresses two limitations of large language models (LLMs) in solving complex problems: (1) their reasoning processes exhibit Bayesian-like stochastic generation, where each token is sampled from a context-dependent probability distribution, leading to inherently random decision trajectories rather than deterministic planning; (2) the reasoning and decision-making mechanisms are statically decoupled, meaning dynamically retrieved domain knowledge fails to dynamically adjust the underlying reasoning strategy. These dual deficiencies result in initial decisions lacking strategic anchoring and reasoning chains often failing to converge on correct solutions, as stochastic generation lacks mechanisms for trajectory correction or knowledge-guided optimization during sequential reasoning. To resolve these issues, we propose a problem-solving method integrated into the LLM's generation process to guide reasoning. This method, compatible with numerous LLMs and featuring reusable solutions, is grounded in a novel Heuristic-Classification-of-Thoughts prompting schema (HCoT). HCoT synergizes the LLM's reasoning ability with a structured problem space via a heuristic classification model that controls the reasoning process and provides reusable abstract solutions. Evaluated on two complex inductive reasoning tasks with ill-defined search spaces, HCoT outperforms existing approaches (e.g., Tree-of-Thoughts and Chain-of-Thoughts prompting) in performance. On the well-structured 24 Game task, HCoT demonstrates significantly higher token efficiency compared to the state-of-the-art Tree-of-Thoughts-Breadth-First-Search. In terms of both accuracy and token usage, HCoT achieves a Pareto frontier balance, offering a strong trade-off between performance and computational cost.

---

### From Myopic Selection to Long-Horizon Awareness: Sequential LLM Routing for Multi-Turn Dialogue

Authors: Jiarui Zhang, Xiangyu Liu, Yong Hu, Chaoyue Niu, Hang Zeng, Shaojie Tang, Fan Wu, Guihai Chen
Published: 2026-04-14
Categories: cs.CL
arXiv: https://arxiv.org/abs/2604.12385v1

Multi-turn dialogue is the predominant form of interaction with large language models (LLMs). While LLM routing is effective in single-turn settings, existing methods fail to maximize cumulative performance in multi-turn dialogue due to interaction dynamics and delayed rewards. To address this challenge, we move from myopic, single-turn selection to long-horizon sequential routing for multi-turn dialogue. Accordingly, we propose DialRouter, which first performs MCTS to explore dialogue branches induced by different LLM selections and collect trajectories with high cumulative rewards. DialRouter then learns a lightweight routing policy from search-derived data, augmented with retrieval-based future state approximation, enabling multi-turn routing without online search. Experiments on both open-domain and domain-specific dialogue tasks across diverse candidate sets of both open-source and closed-source LLMs demonstrate that DialRouter significantly outperforms single LLMs and existing routing baselines in task success rate, while achieving a superior performance-cost trade-off when combined with a cost-aware reward.

---

### Preventing Safety Drift in Large Language Models via Coupled Weight and Activation Constraints

Authors: Songping Peng, Zhiheng Zhang, Daojian Zeng, Lincheng Jiang, Xieping Gao
Published: 2026-04-14
Categories: cs.AI
arXiv: https://arxiv.org/abs/2604.12384v1

Safety alignment in Large Language Models (LLMs) remains highly fragile during fine-tuning, where even benign adaptation can degrade pre-trained refusal behaviors and enable harmful responses. Existing defenses typically constrain either weights or activations in isolation, without considering their coupled effects on safety. In this paper, we first theoretically demonstrate that constraining either weights or activations alone is insufficient for safety preservation. To robustly preserve safety alignment, we propose Coupled Weight and Activation Constraints (CWAC), a novel approach that simultaneously enforces a precomputed safety subspace on weight updates and applies targeted regularization to safety-critical features identified by sparse autoencoders. Extensive experiments across four widely used LLMs and diverse downstream tasks show that CWAC consistently achieves the lowest harmful scores with minimal impact on fine-tuning accuracy, substantially outperforming strong baselines even under high harmful data ratios.

---

### Modality-Agnostic Prompt Learning for Multi-Modal Camouflaged Object Detection

Authors: Hao Wang, Jiqing Zhang, Xin Yang, Baocai Yin, Lu Jiang, Zetian Mi, Huibing Wang
Published: 2026-04-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2604.12380v1

Camouflaged Object Detection (COD) aims to segment objects that blend seamlessly into complex backgrounds, with growing interest in exploiting additional visual modalities to enhance robustness through complementary information. However, most existing approaches generally rely on modality-specific architectures or customized fusion strategies, which limit scalability and cross-modal generalization. To address this, we propose a novel framework that generates modality-agnostic multi-modal prompts for the Segment Anything Model (SAM), enabling parameter-efficient adaptation to arbitrary auxiliary modalities and significantly improving overall performance on COD tasks. Specifically, we model multi-modal learning through interactions between a data-driven content domain and a knowledge-driven prompt domain, distilling task-relevant cues into unified prompts for SAM decoding. We further introduce a lightweight Mask Refine Module to calibrate coarse predictions by incorporating fine-grained prompt cues, leading to more accurate camouflaged object boundaries. Extensive experiments on RGB-Depth, RGB-Thermal, and RGB-Polarization benchmarks validate the effectiveness and generalization of our modality-agnostic framework.

---

### Beyond Output Correctness: Benchmarking and Evaluating Large Language Model Reasoning in Coding Tasks

Authors: Yuangang Li, Justin Tian Jin Chen, Ethan Yu, David Hong, Iftekhar Ahmed
Published: 2026-04-14
Categories: cs.SE, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2604.12379v2

Large language models (LLMs) increasingly rely on explicit reasoning to solve coding tasks, yet evaluating the quality of this reasoning remains challenging. Existing reasoning evaluators are not designed for coding, and current benchmarks focus primarily on code generation, leaving other coding tasks largely unexplored. We introduce CodeRQ-Bench, the first benchmark for evaluating LLM reasoning quality across three coding task categories: generation, summarization, and classification. Using this benchmark, we analyze 1,069 mismatch cases from existing evaluators, identify five recurring limitations, and derive four design insights for reasoning evaluation in coding tasks. Guided by these insights, we propose VERA, a two-stage evaluator that combines evidence-grounded verification with ambiguity-aware score correction. Experiments on CodeRQ-Bench show that VERA consistently outperforms strong baselines across four datasets, improving AUCROC by up to 0.26 and AUPRC by up to 0.21. We release CodeRQ-Bench at https://github.com/MrLYG/CodeRQ-Bench, supporting future investigations.

---

### Beyond Output Correctness: Benchmarking and Evaluating Large Language Model Reasoning in Coding Tasks

Authors: Yuangang Li, Justin Tian Jin Chen, Ethan Yu, David Hong, Iftekhar Ahmed
Published: 2026-04-14
Categories: cs.SE, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2604.12379v1

Large language models (LLMs) increasingly rely on explicit reasoning to solve coding tasks, yet evaluating the quality of this reasoning remains challenging. Existing reasoning evaluators are not designed for coding, and current benchmarks focus primarily on code generation, leaving other coding tasks largely unexplored. We introduce CodeRQ-Bench, the first benchmark for evaluating LLM reasoning quality across three coding task categories: generation, summarization, and classification. Using this benchmark, we analyze 1,069 mismatch cases from existing evaluators, identify five recurring limitations, and derive four design insights for reasoning evaluation in coding tasks. Guided by these insights, we propose VERA, a two-stage evaluator that combines evidence-grounded verification with ambiguity-aware score correction. Experiments on CodeRQ-Bench show that VERA consistently outperforms strong baselines across four datasets, improving AUCROC by up to 0.26 and AUPRC by up to 0.21. We release CodeRQ-Bench at https://github.com/MrLYG/CodeRQ-Bench, supporting future investigations.

---

### ReasonXL: Shifting LLM Reasoning Language Without Sacrificing Performance

Authors: Daniil Gurgurov, Tom Röhr, Sebastian von Rohrscheidt, Josef van Genabith, Alexander Löser, Simon Ostermann
Published: 2026-04-14
Categories: cs.CL
arXiv: https://arxiv.org/abs/2604.12378v1

Despite advances in multilingual capabilities, most large language models (LLMs) remain English-centric in their training and, crucially, in their production of reasoning traces. Even when tasked with non-English problems, these models predominantly reason in English, creating a fundamental mismatch for non-English usage scenarios. We address this disparity directly with three contributions. (i) We introduce ReasonXL, the first large-scale parallel corpus of cross-domain reasoning traces spanning five European languages (English, German, French, Italian, and Spanish), with over two million aligned samples per language, each comprising prompts, reasoning traces, and final outputs, enabling direct supervision of language-specific reasoning. (ii) Using ReasonXL, we demonstrate that LLMs can be adapted to reason entirely in a desired target language, using a simple two-stage pipeline of supervised fine-tuning (SFT) followed by reinforcement learning with verifiable rewards (RLVR). The resulting models match or exceed baseline performance, with minimal loss in general knowledge and broadly preserved cross-lingual transfer. (iii) We conduct an extensive representational analysis of the adaptation and find a clear functional division across model depth: early layers contain an activation bottleneck that causally determines language identity, while upper layers concentrate the weight and activation changes driven by adaptation. We further find that RLVR achieves greater behavioral divergence from the base model with smaller parameter updates than SFT, suggesting a more efficient representational rerouting despite much smaller weight updates.

---

### SCRIPT: A Subcharacter Compositional Representation Injection Module for Korean Pre-Trained Language Models

Authors: SungHo Kim, Juhyeong Park, Eda Atalay, SangKeun Lee
Published: 2026-04-14
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2604.12377v1

Korean is a morphologically rich language with a featural writing system in which each character is systematically composed of subcharacter units known as Jamo. These subcharacters not only determine the visual structure of Korean but also encode frequent and linguistically meaningful morphophonological processes. However, most current Korean language models (LMs) are based on subword tokenization schemes, which are not explicitly designed to capture the internal compositional structure of characters. To address this limitation, we propose SCRIPT, a model-agnostic module that injects subcharacter compositional knowledge into Korean PLMs. SCRIPT allows to enhance subword embeddings with structural granularity, without requiring architectural changes or additional pre-training. As a result, SCRIPT enhances all baselines across various Korean natural language understanding (NLU) and generation (NLG) tasks. Moreover, beyond performance gains, detailed linguistic analyses show that SCRIPT reshapes the embedding space in a way that better captures grammatical regularities and semantically cohesive variations. Our code is available at https://github.com/SungHo3268/SCRIPT.

---

### Cooperative Memory Paging with Keyword Bookmarks for Long-Horizon LLM Conversations

Authors: Ziyang Liu
Published: 2026-04-14
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2604.12376v2

When LLM conversations grow beyond the context window, old content must be evicted -- but how does the model recover it when needed? We propose cooperative paging: evicted segments are replaced with minimal keyword bookmarks ((pN:keywords), ~8-24 tokens each), and the model is given a recall() tool to retrieve full content on demand. On the LoCoMo benchmark (10 real multi-session conversations, 300+ turns), cooperative paging achieves the highest answer quality among six methods -- outperforming truncation, BM25, word-overlap retrieval, a search-tool baseline, and full context -- on four models (GPT-4o-mini, DeepSeek-v3.2, Claude Haiku, GLM-5), confirmed by four independent LLM judges ($p=0.017$, paired bootstrap). We then study the paging design space with a 5x4 ablation over boundary strategies and eviction policies (3,176 synthetic probes, 1,600 LoCoMo probes). Key findings: (1) coarse fixed-size pages (fixed_20) reach 96.7% while content-aware topic_shift collapses to 56.7%; (2) eviction policy choice is data-dependent (FIFO best on synthetic, LFU on LoCoMo); (3) two bookmark generation strategies improve over the heuristic baseline (+4.4 and +8.7 E2E points); (4) the remaining bottleneck is bookmark discrimination -- the model triggers recall() 96% of the time but selects the correct page only 57% when bookmarks are insufficiently distinctive. Keyword specificity alone accounts for a 25 percentage point accuracy difference.

---

### Nemotron 3 Super: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning

Authors: NVIDIA, :, Aakshita Chandiramani, Aaron Blakeman, Abdullahi Olaoye, Abhibha Gupta, Abhilash Somasamudramath, Abhinav Khattar, Adeola Adesoba, Adi Renduchintala, Adil Asif, Aditya Agrawal et al.
Published: 2026-04-14
Categories: cs.LG, cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2604.12374v1

We describe the pre-training, post-training, and quantization of Nemotron 3 Super, a 120 billion (active 12 billion) parameter hybrid Mamba-Attention Mixture-of-Experts model. Nemotron 3 Super is the first model in the Nemotron 3 family to 1) be pre-trained in NVFP4, 2) leverage LatentMoE, a new Mixture-of-Experts architecture that optimizes for both accuracy per FLOP and accuracy per parameter, and 3) include MTP layers for inference acceleration through native speculative decoding. We pre-trained Nemotron 3 Super on 25 trillion tokens followed by post-training using supervised fine tuning (SFT) and reinforcement learning (RL). The final model supports up to 1M context length and achieves comparable accuracy on common benchmarks, while also achieving up to 2.2x and 7.5x higher inference throughput compared to GPT-OSS-120B and Qwen3.5-122B, respectively. Nemotron 3 Super datasets, along with the base, post-trained, and quantized checkpoints, are open-sourced on HuggingFace.

---