Papers

Modern neural network training increasingly uses matrix-aware optimizers, yet their conditioned matrix step is typically added directly to the weight, jointly changing its norm and direction. This interaction matters because the current norm determines angular motion, while directional learning can drive norm growth and thereby alter later steps. We introduce RODE, which gives the radial and directional components separate update rules and step sizes. RODE explicitly updates the matrix Frobenius norm through a scalar radial rule, while its directional channel performs Newton–Schulz-conditioned updates in the tangent space. Controlled GPT-2 interventions show gains from both direct norm control and RODE's directional update. Across two language-modeling and two image-classification tasks, RODE outperforms both Muon variants in every direct comparison and ends with lower full-model norms. At 1.5B scale, using the learning rate transferred directly from the Qwen2-style LM sweep, RODE lowers loss from 4.145 to 3.346 and final global norm from 11964 to 2183 relative to Muon RMS, with fixed-radius RODE improving further. For Qwen3.5-9B full-parameter fine-tuning, all six optimizers use the same tuning budget and the same formal-training and evaluation settings; RODE outperforms both Muon variants on all four evaluation tasks and attains the highest mean on GSM8K and MATH-500. Thus, decoupling radial and directional dynamics offers a more effective and controllable approach to matrix optimization.

Word alignment has traditionally been studied between sentences, but many cross-lingual tasks increasingly require correspondences across full documents. While recent multilingual embedding models can encode long inputs, we show that applying algorithms designed for sentences directly to documents leads to performance degradation. To address this, we introduce CTFAlign, a lightweight, training-free approach for document-level word alignment. CTFAlign applies a coarse-to-fine refinement strategy that restricts the alignment search space to semantically similar regions. Additionally, we introduce MDPAlign, a simpler alternative that constrains alignments by position with a main diagonal prior. Both approaches operate directly on full documents without relying on sentence segmentation or sentence alignment. We evaluate these methods across six language pairs varying in typological distance, resourcedness, and document length. Averaged over three models, CTFAlign reduces word alignment error rate from 0.412 to 0.326. These gains transfer downstream, leading to improvements in document-level translation coverage evaluation and recognition of semantic differences. We release CTFAlign as a Python package and make the code and data to reproduce our experiments publicly available.

Micro-actions are subtle, short, low-amplitude body movements, such as a fidgeting hand or a slight head tilt, that humans perform with little conscious intent yet that reliably leak emotional and psychological state. Understanding them goes beyond assigning a label: a model must also describe which body parts move and reason, faithfully, about why a clip warrants a particular fine-grained category. We present the training-free, prompt-only system that won first place in the fine-grained understanding track (MA-Bench) of the MAC~2026 Micro-Action Challenge, where both fine-tuning and ground-truth supervision are disallowed. Built entirely upon frozen multimodal large language models (MLLMs), the system dynamically routes each of the eight sub-tasks to the MLLM empirically best suited for that task: a discriminative MLLM for closed-ended recognition tasks and a generative MLLM for open-ended description and reasoning tasks. This architecture achieves a statistically significant performance advantage on open-ended tasks, attaining an average score of 2.68 (on a five-point scale) compared to 1.44 for the second-best approach.

Real-world fault analysis in 5G and emerging 6G networks demands domain expertise to analyze free-text diagnostics, including root-cause explanations and recommended actions. LLMs have emerged as a promising approach to automating this, yet whether lightweight, edge-deployable models are capable of performing in-depth free-text diagnostics remains an open question. While existing benchmarks rely on restrictive MCQs with fixed answer keys, this paper evaluates 5G domain understanding and fault analysis in a free-text generation format. Transitioning to this paradigm requires evaluating lightweight, edge-deployable AI models on open-ended diagnostic reasoning, alongside a dependable framework to validate these text outputs at scale. To address this we evaluate three lightweight LLMs, Claude-Haiku-4.5, GPT-5.4-Mini, and Gemini-3.1-Flash-Lite, on free-text 5G domain knowledge and fault-analysis tasks across three benchmarks, TeleQNA ORAN FT, 5G-Faults FT, and TeleInter FT. Three independent frontier judges score outputs, and pairwise inter-judge agreement is measured as an empirical test of the LLM-as-Judge methodology. All three models reach at least 90% accuracy on fault diagnosis, while zero-shot recall of 3GPP and O-RAN specifications remains the critical gap, with all models scoring below 60%. Mean inter-judge agreement is at least 0.90 across all runs, indicating that multi-judge LLM scoring produces consistent, reproducible grades for open-ended telecom responses. Operationally, Gemini-3.1-Flash-Lite offers the best efficiency trade-off, combining competitive accuracy with the lowest inference cost and latency, making it the most suitable candidate for production telecom deployments.

Quantization is widely used to deploy large language models, but its effect on uncertainty behavior, such as confidence, margins, and abstention, is rarely treated as a primary objective. We frame calibration-data selection for quantization as a target-dependent uncertainty-preservation problem. Different deployments emphasize different regions of the input distribution, yet prior work mainly optimizes accuracy-oriented compression metrics or adjusts scores after quantization. We formalize this goal with distributional and boundary preservation risks, and provide a simple mixture-mismatch argument explaining why no single calibration recipe should be expected to fit all targets. We introduce Doubt-Preserving Quantization (DPQ), a lightweight pre-quantization recipe family that uses full-precision predictions to construct target-aligned calibration mixtures of high-doubt examples and generic anchors. Across 8 language models, 9 NLP benchmarks, and 22 comparison methods, the leading fixed recipe changes with the preservation target: DPQ-r75 leads on SQuAD2 answerability-boundary preservation, while milder or single-signal variants, including DPQ-r50, confidence-only, and entropy-only, better preserve broad multiple-choice QA behavior. These results show that calibration data should be selected for the specific full-precision score behavior a deployment needs to preserve, rather than treated as a fixed quantization detail.

Generative recommendation represents each item with a sequence of discrete Semantic IDs (SIDs) and predicts the sequence to retrieve the next item. Typical systems use multi-level residual quantization, which increases autoregressive decoding cost and creates a large hierarchical space that may be sparsely occupied. Static codebooks also become misaligned with current traffic as new items arrive and exposure distributions change. We propose a single-level large semantic codebook that replaces multiple residual semantic codes with one semantic token while retaining a separate collaborative disambiguation token to reduce item collisions. We further introduce an exposure-aware dynamic update mechanism based on temporal weight decay, exponential moving-average center updates, and an exposure-weighted penalty on SID changes. We also develop an offline evaluation framework covering representation quality, code utilization, cluster load, full-SID collision, and temporal stability. On two public datasets, the two-level SID improves mean Recall@10 by 5.0%-8.8% and mean NDCG@10 by 4.1%-5.1% for OneRec-V1, and by 7.1%-8.7% and 3.8%-8.5%, respectively, for OneRec-V2. Dynamic updating provides further gains on KuaiRec. Across three serving architectures, the shorter SID reduces estimated autoregressive-decoding FLOPs by 47.93%-48.70% and increases single-card QPS by 28.57%-47.0%. A five-day online A/B test serving 2.5% of production traffic improves the primary consumption metric by 0.792%.

Ensuring that Extended Reality (XR) environments are age-appropriate is an important regulatory and safety challenge. However, current age assurance operates only at registration and cannot verify the age of the active user during a session. Face-based approaches, the dominant solution in social media and adult platforms, are impractical in XR, because they require removing the headset and taking a self-captured image, often on a mobile app. This both breaks immersion and introduces the privacy risk of sharing face pictures with third parties, which leaves XR platforms without a viable path to continuous, in-session and privacy-preserving age assurance. We propose the dorsal part of the hand as an alternative to the face, by exploiting the egocentric cameras that XR headsets inherently and naturally use to capture gesture interactions. To evaluate this, we collect an age- and sex-stratified, ethnodiverse dataset of 436 participants spanning the minor–adult boundary, captured under unconstrained lighting and orientation conditions. To characterise what is achievable with off-the-shelf methods at the minor–adult boundary, we evaluate standard neural network architectures for age assurance at the legally critical 18-year threshold. Analysis confirms performance is robust to skin-tone variation. On this dataset, the challenge-31 operating point achieves zero minor admission, making the system a viable first-stage filter for age assurance. These findings position dorsal hand morphometrics as an effective and more privacy-preserving biometric modality for in-session age assurance in XR.

Mobile AR frameworks attach a metric pose prior to every casual phone capture, and turning it into reconstruction-grade poses cheaply on CPU is the step before novel-view synthesis. The least a refiner owes an accurate prior is not to make it worse. The workhorse refiner does. On 15 ScanNet++ iPhone room captures, COLMAP triangulation plus prior-seeded bundle adjustment degrades an accurate ARKit prior in all 15, 0.55 degrees to 0.74 degrees by scene-mean. The cause is the seeding. Structure is triangulated from the prior before anything is optimized, so the prior's error is baked into the structure the optimizer trusts. We remove the triangulation. Every keypoint owns a scalar depth along its own back-projected ray and each match contributes two symmetric cross-projection residuals, so structure is re-expressed at every iterate. The same solve holds the room prior at 0.57 degrees and never fails in 330 perturbed room runs, and at object scale reaches 0.265 degrees/1.80 mm from a prior at 0.456 degrees in a median of 10 s per scene on one CPU, against 2.5 GPU-hours for a learned refiner. Because no structure is committed, the objective also admits graduated non-convexity, which measures how deep the defect goes. Classical refinement collapses past 1-2 degrees of prior error, barely beyond a real ARKit prior, and no classical refinement arm survives 32 degrees. Ours recovers 425 of 425 runs through 16 degrees/80 mm and 85% at 32 degrees/160 mm, and perturbed rooms through 32 degrees. Nominal object-scale accuracy is on par rather than better, on a benchmark at its own noise floor, where classical bundle adjustment is a strong baseline absent from the literature. One scene fails for every solver already at zero perturbation. Re-mapping from position priors matches us in the prior's frame but discards it, so it cannot exploit a prior worth keeping or be warm-started.

Multimodal LLMs apply the language model interface to visual inputs, where ordinal regression tasks such as age estimation, image quality assessment, and disease grading require autoregressive decisions over ordered class labels. We ask whether MLLMs reliably convert internal ordinal evidence into ordered digit-token outputs. Across four ordinal benchmarks and four MLLM backbones, ordinal labels are linearly recoverable from hidden states with Spearman correlation up to 0.938, and a task-designed prompt further sharpens this structure. Yet native digit-token outputs weakly expose it: the unembedding matrix filters the ordinal direction, and the digit-token row space retains below 1.15% across all 16 model-dataset combinations, with a 16 to 77 absolute-point accuracy gap between linear-probe and native outputs. We introduce Ordinal Lens Alignment (OLA), a frozen-backbone inference-time method that trains lightweight W_S-anchored lenses on mid-to-deep decoder layers, fuses them into an ordinal distribution, and corrects only digit-token logits at generation. OLA outperforms the SOTA LoRA-tuned OrderChain baseline in most settings while keeping the MLLM frozen, surpasses discriminative ordinal baselines in most cells, and improves over an offline lens in every setting.

Reservoir computing (RC) couples a fixed recurrent dynamical system with a trained lightweight readout, but this efficiency is partly lost during hyperparameter selection: the recurrent gain, input scale, and leakage rate determine the reservoir's stability and temporal processing regime and are usually tuned through many rollouts. We introduce a deterministic, pilot-informed selector for leaky linear reservoirs followed by coordinate-wise nonlinear features. Free probability yields cross-lag propagation coefficients that summarize how the reservoir mixes past inputs. In the large-width limit, these coefficients define a deterministic temporal kernel that approximates the finite-reservoir feature geometry. Kernel ridge regression on a short labelled pilot sequence therefore ranks candidate operating regimes without instantiating or rolling out a reservoir, and the selected configuration transfers across widths. Across ten synthetic temporal benchmarks, zero-rollout selection obtains a mean deployment score of $0.772$, compared with $0.774$ for exhaustive simulation-based search, while avoiding $156\,600$ selection rollouts. With a small rollout budget, the proposed ranking provides the strongest mean performance at every tested budget and reaches the exhaustive reference using $4.8\%$ of its rollout cost. On four public electricity-transformer-temperature (ETT) forecasting datasets, five retained candidates recover the exhaustive operating point on three datasets. On multivariate cellular-traffic forecasting, 15 rollouts per cell reach the 462-rollout exhaustive reference and outperform random search and Bayesian optimization at low budgets. These results position free-probability kernels as deterministic surrogates for selecting reservoir operating regimes when validation rollouts are scarce.

We propose a stabilized first-kind, Calderón preconditioned, vector-potential-only integral equation, for scattering at perfect electrical conductors. The method yields accurate low-frequency results for both the vector and the scalar potential, where the scalar potential is computed using the Lorenz gauge in the post-processing. Both the near and far fields can be accurately computed at arbitrarily low frequencies. This is achieved in a post-processing step and is independent of machine and quadrature precision. First, the low-frequency scaling of the degrees of freedom, and the near and far fields, are studied. Next, stabilized methods are derived based on the low-frequency scaling. The tangential trace of the curl of the vector-potential is stabilized leveraging quasi-Helmholtz projectors, while the normal trace of the vector-potential is stabilized with a projector based on the static scalar potential. Numerical experiments are presented showing that the stabilized method shows accurate results even for very low frequencies.

Graph Foundation Models (GFMs) on text-attributed graphs (TAGs) align graph representations with language semantics to support transferable graph learning. Despite these advantages, the backdoor vulnerability of GFMs on TAGs remains insufficiently understood, especially under graph-language alignment, where graph and text representations are trained to constrain each other in a shared semantic space. Existing backdoor attacks mainly target either the graph side or the text side, treating the two modalities independently. This makes direct adaptation ineffective: graph-only triggers can be constrained by clean text semantics, while text-only triggers alter the language view but do not directly shift the graph representation being aligned and scored. TAGs also impose a stealth challenge because triggers are exposed as both node text and local graph structure, making incoherent trigger attributes or anomalous subgraphs easy to inspect or filter. In this paper, we propose STAG, a stealthy trojan attack framework designed for the graph-language alignment interface of GFMs on TAGs. STAG coordinates a graph-trigger generator with a text-side soft prompt so that trigger-attached graph representations and triggered text representations move toward the same target-class text region. To address TAG-specific stealthiness, STAG realizes trigger nodes as readable text through candidate retrieval and regularizes the trigger-attached subgraph so that its local structure remains close to the original subgraph. Extensive experiments on multiple TAG datasets and representative GFMs demonstrate the effectiveness and stealthiness of STAG. Our code is available at https://github.com/ventr1c/STAG.

Quantization of Large Language Models (LLMs) is often hindered by the sensitivity of the self-attention mechanism to discretization errors. We identify the softmax operator as a bottleneck for quantization stability due to its sensitivity to outliers and state-dependent Jacobian. We theoretically establish that suppressing the norm of this Jacobian helps in bounding quantization-induced performance degradation. Based on this, we propose Jacobian-Guided Noise Injection, a training strategy that injects zero-mean Gaussian noise into pre-attention logits, with variance derived directly from the Jacobian Frobenius norm. Unlike prior approaches that rely on heuristic or penalise jacobian directly, our method provides a way to identify the optimal noise variance based on the local attention sensitivity. We evaluate the method on SOTA LLM architectures, where it demonstrates improved robustness over popular PTQ methods. Empirical analysis reveals that the proposed method gives up to +37% relative gains on Top-1 accuracy on ImageNet-1K for SigLIP and improves relative perplexity by upto 40% on WikiText for language models in low bit quantisation settings, proving the efficacy of the approach.

Narratives are central to how social communication is framed, making their detection critical for understanding and analysing public discourse. Prior work has explored narrative detection and extraction across diverse domains; however, migration narratives remain significantly understudied, primarily due to the absence of dedicated annotated datasets. Furthermore, public communication has recently shifted towards video-centric platforms, where narratives are conveyed through multimodal signals and consumed at scale. Despite this shift, narratives in videos remain largely unexplored. To bridge these gaps, we introduce MigrationNarrate, the first multimodal dataset for detection of migration narratives in the UK, consisting of 1,115 YouTube video transcripts annotated using a two-level taxonomy of 12 migration super-narratives and 53 narrative labels. This paper details the dataset design, collection, and annotations; together with benchmark results using a combination of pre-trained encoder models and both open- and closed-source Large Language Models. Finally, a thorough error analysis offers insights for future work.

False information spreads rapidly on social media during disasters and can undermine emergency response efforts, public trust, and crisis communication. Existing research primarily focuses on determining whether social media posts contain false information, but provides limited insight into the specific false claims embedded within posts and the severity of individual false claims. To address the limitations, we propose a two-stage framework to assess the severity of false claims during disasters. In the first stage, we develop a false claim extraction agent that identifies false claims from multimodal social media posts containing text, images, videos, and links. A subsequent verification step validates extracted claims with supporting evidence. In the second stage, we define false claim severity as the combination of two complementary dimensions: believability, which determines the likelihood that a claim will be believed, and harmfulness, which captures the potential consequences if it is believed. Human annotators assess both dimensions to construct a claim-level severity benchmark using false claims extracted from Reddit posts related to hurricanes and wildfires. Building upon this benchmark, we investigate false claim severity assessment as a human-AI alignment problem, evaluating whether models can reproduce human judgments under a shared evaluation rubric rather than merely predicting severity labels. Experiments on the benchmark show that traditional supervised models exhibit limited alignment with human judgments, whereas Large Language Models (LLMs) achieve substantially stronger performance. Among the evaluated strategies, in-context learning consistently achieves the strongest alignment with human judgments, highlighting the importance of human examples and shared decision criteria for severity assessment.

Graph neural networks (GNNs) are routinely employed for spatiotemporal forecasting, yet their performance across widely used benchmark datasets is inconsistent. Here, we perform an audit of dataset properties and baseline models to assess the quality of the benchmarks, and the robustness of the conclusions drawn from them. Using classical statistical tools, we characterise spatiotemporal lagged dependencies in benchmarks, and examine how temporal differencing changes these relationships and affects model rankings. Motivated by this, we re-evaluate temporal linear baselines, significantly reducing the apparent gains from GNNs on several benchmarks, and surpassing GNNs on others. Suspecting that GNNs struggle to extract linear, node-wise signals, we find that supplying them with autoregressive residuals improves their performance particularly on non-traffic benchmarks. Finally, controlled synthetic experiments reveal that GNNs are sensitive to heterogeneity in temporal dynamics and spatial graph interactions. Together, our findings demonstrate that baseline specification, data pre-processing and system heterogeneity shape the interpretations drawn from benchmark rankings, informing the design and robust evaluation of GNNs.

Effective social interaction requires agents to translate mental state inferences into coordinated behavioral signals across verbal and nonverbal channels simultaneously. Yet existing benchmarks evaluate theory of mind (ToM) reasoning and embodied behavior in isolation, leaving unmeasured the gap between social inference and social action. We introduce MOSAIC (Multimodal Orchestration of Social Action, Inference, and Communication), a controlled benchmark in which two embodied agents interact across cooperative and competitive scenarios requiring integration of verbal statements, spatial trajectories, gaze direction, and facial expression under systematically varied ToM constraints. Evaluating 13 models, including 11 VLMs, across 200 trials per model, we find that VLMs fail to produce behaviors consistent with the expected outcomes under ToM-order constraints, and that imposing explicit ToM-order constraints produces no reliable behavioral change aligned with the specified reasoning level. Signal-level analysis reveals two sequential bottlenecks: most models cannot produce directionally coherent nonverbal signals, and even when signals are present, VLM agents fail to interpret others behaviors and react to them. PCM-LLM, included as a structured architectural reference point with an explicit ToM module, succeeds across all conditions, suggesting that explicit belief-action coupling is a sufficient ingredient for this class of tasks.

Video Joint Embedding Predictive Architecture (V-JEPA) learns powerful spatiotemporal representations from video through self-supervised latent feature prediction. However, V-JEPA is built around random-mask completion and deterministic regression, making it fundamentally ill-suited for autonomous driving planning that demands future-directed prediction tightly coupled with action. To address this, we rethink the V-JEPA paradigm and present WA-JEPA, a V-JEPA-native world-action model designed for autonomous driving planning. Instead of random spatiotemporal masking, WA-JEPA employs hybrid future-masked pre-training, where the model infers future latents from observed context. Departing from deterministic regression, we recast future prediction as conditional flow matching over latent futures, which substantially improves the model's ability to generate plausible future latents for downstream planning. Finally, a joint future-action predictor is proposed to denoise future scene tokens and ego trajectories together in a unified spatiotemporal latent space, allowing action supervision to directly shape planning-relevant world representations. Pre-trained on nuPlan videos and fine-tuned on NAVSIM, WA-JEPA reaches 91.7 EPDMS on NAVSIM-v2, surpassing the strongest end-to-end and world-action baselines by 1.6 and 1.3 EPDMS, and, without HUGSIM-specific fine-tuning, attains the best HD-Score of 0.4462 on the closed-loop HUGSIM benchmark under the same evaluation protocol. These results validate V-JEPA-native world-action modeling as a powerful and scalable paradigm for autonomous driving planning. Code is available at https://github.com/AFARI-Research/WA-JEPA.

We investigate how the realism of synthetic room impulse response (RIR) datasets affects the training of DeepFilterNet3 for single-channel speech enhancement. We compare a DNS4 image-source-method (ISM) RIR dataset with a higher-acoustic-fidelity dataset generated using hybrid wave-based and geometrical acoustics simulation. Rather than isolating individual simulation factors, we compare complete RIR generation pipelines while keeping the enhancement model unchanged. Models are evaluated on unseen measured RIRs using objective speech enhancement metrics and downstream automatic speech recognition (ASR). Training with the higher-fidelity dataset consistently yields modest improvements in objective metrics and substantially lower ASR word error rates than the ISM dataset. Although the experiments do not attribute these gains to individual modelling components, they show that increasing the overall realism of synthetic acoustic training data improves the generalization of DeepFilterNet3 to unseen measured environments.

Recent work in mechanistic interpretability has studied how large language models recall facts stored in their weights. This paper argues that factual recall points to something broader: a general kind of operation in deep learning models, which I call feature recall. The core observation is that a linear projection can be read as retrieving stored information scaled by input activations. I define feature recall, show it applies across architectures, and contrast it with the established paradigm of feature combination. I also consider how cases of feature recall might be mechanistically identified. The account gives philosophers a new conceptual tool for understanding deep learning, and points to empirical directions for mechanistic interpretability research.

Multimodal clinical AI is limited by weakly aligned inputs and the absence of domain-specific interpretable representations, particularly when learning from dense video stream, structured time-series, and template-based kinematic text. Here we present ScoliDetect, an explainable framework for adolescent idiopathic scoliosis screening from monocular gait video, built around a kinematic knowledge map (KKM) and complementary template-based kinematic text derived from per-sequence pose statics. KKM is a fixed-index structured representation that encodes gait features across absolute motion, self-skeleton configuration and joint-joint signal correlation, providing anchor-referenced multimodal fusion and factor-level interpretation. We integrate video, KKM, and template-based kinematic text through bidirectional cross-attention with latent-bottleneck aggregation. In a multicenter cohort (n = 1,858 after exclusions), prespecified supervised ablations on an external screening cohort show that KKM-mediated multimodal fusion outperforms unimodal models and late concatenation. Under a staged training protocol, trimodal contrastive pretraining is applied after architecture selection as representation initialization, improving external ROC-AUC from 0.961 to 0.972. Furthermore, the structured nature of the KKM provides inherent, factor-level attributions mapped directly to specific kinematic phases and skeletal indices, offering verifiable interpretability. The results demonstrate that embedding explicit structural topologies into latent spaces significantly enhances both the generalization and explainability of multimodal pattern analysis systems.

Accurate soft tissue simulation is essential for surgical training, pre-operative planning, and haptic feedback systems. While learning-based surrogate models trained on data using the finite element method (FEM) offer a promising path to real-time inference, their reliability depends on well-calibrated constitutive models. Existing approaches neither provide systematic guidance on model selection across stiffness levels, nor generalize across different tissue stiffnesses or geometries. We perform a comprehensive calibration of hyperelastic constitutive models in the SOFA Framework using gravity-loaded silicone beams with different stiffnesses. Using calibrated simulations as training data, we use a softness conditioned equivariant graph neural network, enabling deformation and force prediction across multiple tissue types and unseen geometries. Our model achieves sub-millimeter mean deformation accuracy at 0.010s inference time, while showing that force prediction quality is directly tied to upstream calibration consistency.

Large language models (LLMs) are increasingly used in cybersecurity workflows, yet it remains unclear whether they can perform structured security reasoning or merely rely on superficial cues and prior knowledge. We study this question in the context of defence selection over attack graphs derived from real-world threat scenarios, including ransomware, supply-chain compromise, cloud abuse, Kubernetes attacks, POS malware, and ICS/OT intrusion. Given a budget constraint, LLMs must select security controls to minimise attacker success. We compare their strategies against each other and against a game-theoretic optimization baseline used as a normative reference for structured reasoning. Our results show that LLMs exhibit conditional competence. When explicit attack-graph structure is provided, they often produce coherent strategies close to the optimization baseline. However, their capabilities are fragile. LLM behaviour becomes increasingly fragile with graph complexity and is highly sensitive to framing. Small prompt changes can substantially alter rankings, and merely relabeling a poor strategy as “optimal” dramatically improves its evaluation. We further observe a non-monotonic relationship between formal risk and LLM judgement: strategies closest to the optimum are not necessarily ranked highest by LLM evaluators. To further probe reasoning ability, we ask LLMs to generate solvers for the same optimization problem. While the generated implementations recover the correct high-level formulation, they scale poorly compared to a purpose-built solver. Overall, our findings show that LLMs can approximate structured cybersecurity reasoning under controlled representations, but do not apply it robustly. This has important implications for the design and evaluation of AI-assisted security decision-support systems.

We define an atomic generation fact f=(u,tau,omega,z;rho), recording the origin, realized transformation, concrete occurrence, generated result and relation role. Compiled into a Generation-Fact Graph (GFG), these facts provide an AI-native, compilable scientific fact substrate preserving generation histories. We establish a GFG-based recursive scientific process in which analysis, intervention, replay and validation form facts for later cycles. Using nanoGPT, we establish unified training-learning dynamics. Training is the evolution of a parameter-optimizer system with state and memory: each actual training action enters the receiving state and produces a finite-amplitude nonlinear functional response conditioned by that state and target-specific update geometry. Learning is the persistent reorganization of distributed functional support by these responses; capability formation, maintenance, decline or recovery becomes observable when target-specific states are evaluated against their readout boundaries. Three primary coordinates - target-boundary state, target-specific update geometry and parameter-Adam receiving state - yield a second-order predictor operating before post-update outputs are read. On held-out runs, it achieved 91.43% accuracy and 91.49% macro-averaged recall across four transitions. We further establish inference as a frozen projection of training-learning dynamics. Component gating and rollback show causal recruitment and non-additive combination of query-conditioned support formed during training, deriving organizational conditions realized by Attention. Controlled feedback indicates possible double-edged reinforcement effects. ResNet/CIFAR-100 and diffusion/CIFAR-10 experiments confirm receiving-state-conditioned responses, persistent support reorganization and frozen inference projection beyond nanoGPT.

In this research, we introduce SAraBERT, an enhanced version of AraBERT which proposes inter-sentence transformer layers for extractive summarization tasks. To ensure that the summaries generated by SAraBERT achieve a high coverage of the document's main ideas, we propose Semantic Siamese Similarity, a novel evaluation metric that measures the level of similarity between two text inputs. We validated using BLEU, ROUGE, and Semantic Siamese similarity on Sarabert and published related models. Simulation results showed the effectiveness of our proposed model and motivate follow on research.

Large language models increasingly generate complete web applications from natural-language prompts, raising the question of whether explicitly requesting security best practice improves the result. We study six functionally distinct web applications, each generated in two prompt variants that are identical except for an appended security-requirements section: a baseline (A) and a security-aware (B) variant. All twelve programs were produced by the same agentic coding assistant and the same model version in a single, non-iterative generation round, and were then analyzed with static, dependency, dynamic and manual techniques, yielding 75 confirmed findings out of 85 candidates. The security-aware variant produced fewer confirmed findings in every application (24 versus 51) and contained no Critical or High issues; the most severe finding was detected only by manual testing. Because the corpus is small and each variant was generated once, we report descriptive observations rather than statistically established effects, and position the work as a preliminary study whose pipeline is being scaled to multiple models and repeated runs.

In household manipulation, pose uncertainty often results in off-centre or partial initial contact, making reliable object acquisition difficult. Roller-based grippers can actively draw objects inward but often provide limited post-capture stability, whereas granular-jamming grippers require sufficient contact before jamming to achieve strong retention. This paper presents a hybrid roller-jamming gripper that integrates active object intake and post-capture retention within a single gripper. The proposed gripper uses inward roller rotation to increase contact and draw the object toward the gripper centre, followed by vacuum-induced granular jamming to stiffen the rollers and stabilise the grasp. The paper also presents a simplified geometric analysis of the gripper and a bench-level characterisation of the prototype's force capability. The gripper prototype was mounted on a 7-DoF robotic arm and evaluated using eight test objects. Furthermore, controlled planar position and orientation offsets were applied, with each condition repeated three times. The main evaluation comprised 840 grasp trials, including 216 planar-offset trials and 624 orientation-offset trials. Overall, the gripper succeeded in 812/840 trials: 215/216 planar-offset trials and 597/624 orientation-offset trials. The ablation evaluation comprised 162 trials on three objects. The roller-only and jamming-only conditions achieved 54/81 and 24/81 successes, respectively, showing their different contributions. These results provide initial mechanism-level evidence that hybrid roller-jamming is a promising strategy for improving acquisition and retention after imperfect first contact.

Modern open models are hybrids: most layers are linear-attention (Gated DeltaNet, GDN) layers carrying a small fixed-size recurrent state instead of a growing key-value (KV) cache. This makes ordinary decoding memory-efficient, but hurts speculative decoding. To verify a batch of draft tokens and then roll back the rejected ones, today's systems snapshot the full recurrent state at every draft position for GDN layers, and those snapshots cannot be shared across branches of a draft tree, so a wide, high-acceptance tree becomes memory-infeasible. We remove the snapshots. Using a tree-structured WY transform of the gated delta rule, we compute every draft node's output with a single triangular solve and reconstruct only the one accepted state on commit, storing a small pseudo-value matrix instead of per-node states; the derivation depends only on the gated delta rule, not on any other architectural detail. In serving benchmarks on two scales of one hybrid model family (Qwen3.5 35B and 397B) this cuts speculative recurrent-state memory and KV-cache pressure at essentially identical acceptance length, turning the freed HBM into higher throughput and much lower time-to-first-token (TTFT) wherever memory binds, and costing a few percent where it does not. For tree width the same memory buys affordability: a wider, higher-acceptance draft becomes possible, though not yet a throughput win.

Language models (LMs) are trained on static scientific corpora, whereas scientific knowledge continuously evolves through correction and revision. Scientific claims encoded within these models may later become retracted, disproven, or updated by subsequent research, creating the risk of disseminating outdated information in scientific workflows. This creates a need for LMs to forget obsolete scientific claims. Machine unlearning offers a promising solution by enabling knowledge removal while maintaining overall model utility. Existing studies primarily investigate instance-level forgetting; however, scientific claims introduce additional challenges because they are interconnected, and continually evolving. To address this gap, we introduce the task of Scientific Claim Unlearning and present a new benchmark, SciUnlearn. We show that current unlearning approaches are unable to effectively eliminate claim-level knowledge and often achieve only superficial suppression, highlighting the need for specialized methods designed for structured knowledge removal.

E-commerce live streaming requires omni-modal understanding of noisy, temporally extended streams, where product facts are distributed across speech, video frames, product images, overlaid text, and user queries. We present TLive-Omni, an omni-modal understanding model tailored to live-commerce scenarios. It maps image, video, audio, and text inputs into a unified representation space. For long-form live streaming analysis, we introduce Per-vGrid, a timestamped token organization that groups each video grid with its temporally corresponding audio within explicit boundary tokens to facilitate temporal alignment. We design a three-stage supervised training recipe that progressively develops live-commerce understanding, from omni-modal perception to instruction-following responses. We then propose Faithful-RFT, a reinforcement fine-tuning stage that further improves answer faithfulness and expression quality while meeting real-time demands, scoring final responses directly with task-verifiable feedback rather than optimizing for reasoning-style exploration during rollout. Moreover, TLive-Omni is supported by a scenario-oriented atomic capability taxonomy and a compact data production engine that converts live-commerce audio, image, and video streams into training signals for speech recognition, speaker analysis, product visual grounding, text recognition, temporal grounding, video dense caption, and omni-modal QA, etc. For scalable training, a synchronized length-grouped sampler reduces padding while preserving comparable workloads across workers, while a lightweight dynamic sampling strategy regenerates rollout groups with near-zero reward variance to maintain meaningful relative advantages for GRPO. Experiments on e-commerce live streaming benchmarks demonstrate strong performance across live-commerce domain tasks, together with excellent generalization on general benchmarks.