New SPBM method for constrained optimization in deep learning. Combines penalty methods, barrier methods, and exponential dual averaging to handle non-convexity and non-smoothness. Demonstrates effectiveness on fairness, physics-informed networks, and symbolic knowledge integration with linear overhead up to 10k constraints.
Robotic clay sculpting planning method using visually-aligned representation. System models deformable material dynamics capturing textures and lighting, enabling long-horizon planning (>100 actions) without per-goal retraining. Tested on three materials with various end-effectors.
Study on 'alignment drift': gradual process where LLM outputs become less constrained by user's current message and more shaped by interaction history, while remaining helpful. Mechanism-oriented framework distinguishes signal A/B, feedback loops, and interactive regimes to control this cumulative drift.
CATA introduces a continual machine unlearning method for vision-language models (VLMs). It represents each unlearning request as a task vector and aggregates historical vectors by suppressing conflicting components, ensuring forgetting effectiveness, model fidelity, and persistence against knowledge re-emergence.
UVTran, a transformer-based framework, solves N-sided hole filling in CAD by predicting an auxiliary projection surface via cross-attention biased toward nearby control points, voxelizing coordinates, and progressive-resolution training. On benchmark, it improves tolerance-satisfaction rate by 12% over industrial and academic baselines while producing fairer trimmed surfaces.
FedNL reformulates federated learning as a three-level nested optimization system. Embeds Titans-based linear attention for zero-shot test-time adaptation without additional training. Tested on Non-IID MMLU and long-context benchmarks with constant inference memory.
An arXiv study compares Matryoshka Representation Learning (MRL) with simple embedding truncation. Results show non-MRL embeddings remain robust up to 80% dimensionality reduction. MRL provides advantage only for heavy truncation (>80%), questioning its systematic training cost.
Paper introduces Contract-Bounded Evidence Activation (CBEA) with Lexicographic Commitment Validation (LCV) for personalized language systems. CBEA+LCV achieves zero failures at 0.49-0.60 availability versus 0.003-0.092 for baselines, with 74-75% median input payload reduction.
Paper introducing trace-based evaluation to detect when agents hit business KPIs while violating behavioral constraints. In hotel pricing with hidden competitor state, authors show PPO variants fail trace alignment while behavior cloning and Trace-Prior RL better preserve price/bid distributions and rate discipline.
StrLoRA introduces a streaming continual visual instruction tuning framework for MLLMs. Unlike existing methods restricted to predefined tasks, StrCVIT handles data streams with dynamic, interleaved tasks. StrLoRA employs two-stage expert routing with task-aware selection and token-wise weighting, stabilized via routing-stability regularization.
FIM-LoRA optimizes rank allocation in LoRA by using 8 calibration passes to estimate gradient variance per layer. This parameter-free approach matches standard LoRA performance (88.6 vs 88.7 on GLUE with DeBERTa-v3-base) while reducing memory costs by 256x compared to full Fisher estimation.
Study on automated classification of invalid bug reports and no-code fix generation. Researchers propose a standardized taxonomy and benchmark, testing vanilla LLMs, RAG, and agentic web search. RAG achieves 0.66 weighted F1 for subclassification; agentic web search reaches 68.9% Judge LLM success rate for fix generation.
UniAlign is a model-agnostic framework improving robustness of network traffic classification under distribution shifts. It combines domain alignment fine-tuning and stable model ensembling, achieving 2.51% accuracy and 2.71% F1 improvements on three public datasets, requiring only 12.4–53.9% of baseline training time.
Study evaluating three off-the-shelf LLMs to estimate difficulty of educational items without response data. Across 6 primary-school mathematics domains, Spearman correlations show moderate-to-strong alignment with empirical difficulties. Pairwise comparisons outperform absolute judgements; adding token probabilities and few-shot examples improves results.
DriveSafe is a framework for risk assessment in autonomous driving scenarios. It generates spatially grounded captions enriched with motion and depth cues, then fine-tunes a lightweight adapter to identify hazardous objects and suggest safety actions. Achieves SOTA on DRAMA benchmark.
AMR-SD introduces asymmetric meta-reflective self-distillation to improve token-level credit assignment in LLM reinforcement learning. The method compresses diagnostic signals into self-generated Socratic hints and uses Causal Information Gain with asymmetric ReLU-gated threshold for sparse token-level advantage modulation, preventing late-stage training collapse.
GA-S2S combines T5-small with a Relational Graph Attention Network (RGAT) for knowledge graph link prediction. The model jointly encodes textual features and full k-hop subgraph topology around the query entity. On CoDEx, GA-S2S outperforms Seq2Seq baselines with 19% relative accuracy gain.
Bayesian-Monte Carlo probabilistic framework for dynamic construction project schedule updating. Models activity durations with lognormal distributions, updates them via Bayesian inference, and propagates uncertainty through Monte Carlo simulation. Demonstrates improved forecasting accuracy over deterministic CPM methods on PSPLIB benchmarks.
Brain tumor segmentation method using multimodal MRI with virtual nodes and dynamic graph neural networks. One-stage framework handling missing modalities through adaptive adjacency matrices and heterogeneous weight matrices. SOTA results on BRATS-2018/2020 with incomplete modalities.
Fre-Res introduces adaptive video-token compression for video MLLMs. The framework separates spatial details (high-fidelity anchors) from temporal evolution (residual-frequency tokens via 1D-DCT). A Spatial-Guided Absorber aligns frequency dynamics with visual embeddings. Results: near full-token performance with substantial reduction in token length across short and long-video benchmarks.
A supervised method called Manifold Probe discovers representation manifolds in superposition within neural networks. Tested on Llama 2-7b, it identifies linear manifolds for time and space, and demonstrates causal control by steering model completions about release years of movies and songs.
An arXiv paper argues that responsible agentic AI requires explicit, traceable provenance across the full lifecycle. Authors formalize this through a causal attribution function and responsibility tensor, demonstrating provenance is estimable and interventionable online before irreversible harm accumulates.
arXiv paper on spatial limitations of MLLMs in multi-agent environments. Models suffer from a "Cartesian Illusion": lack grounded 3D topological understanding. Authors propose an Epistemic Sensory Bottleneck module with Anchor-Based Embodied Spatial Decomposition CoT to improve second-order spatial inference (Theory of Mind). Zero-shot baseline: 42% accuracy.
ReTAMamba is a Mamba-based model for predicting irregular clinical time series. It estimates observation reliability from missingness and elapsed time, integrates multi-resolution information via Chronological Weaving, and uses a budgeted token router. On MIMIC-IV, eICU, and PhysioNet 2012, it improves AUPRC by 7.51%, 7.80%, and 10.15% respectively.
PPR-GDE, an RL method for open-ended generation, uses pairwise preference rewards and group-based diversity to prevent diversity collapse. Without scalar rewards, it preserves subjective evaluations and encourages semantic dispersion within response groups.
Investigation of invariant and equivariant semi-supervised learning (FixMatch, Dense FixMatch) for multi-task training on partially labeled datasets. Evaluation on Cityscapes and BDD100K for object detection and semantic segmentation. Equivariant approaches outperform supervised baselines, especially with limited labeled samples per task.
arXiv study demonstrates that color features alone (RGB/HSV histograms, statistical moments) achieve 89% accuracy in binary cancer/benign classification in histopathology, excluding morphological information. Authors propose these simple features as lightweight pre-screening tool before complex deep learning models.
POST introduces an adversarial learning framework for multivariate time series anomaly detection. The model combines graph neural networks with minimax optimization over adjacency matrices to address spatial over-generalization. Evaluation on public and synthetic benchmarks with channel-wise anomaly localization.
ConsumerSimBench, a benchmark built from 1,553 Chinese social-media topics and 23,122 reaction criteria, evaluates whether LLMs can reconstruct real consumer reaction patterns. Gemini-3.1-Pro covers only 47.8% of criteria, revealing a major gap between technical performance and consumer intuition. A multi-agent pipeline improves MiMo-V2.5-Pro from 32.9% to 37.6%.
A label-wise attention model trained on combined ICD-9 and ICD-10 data improves rare medical code prediction by 27% micro F1 (18K rare codes) and macro metrics on frequent codes, despite version mismatch. Version-independent approach to automate clinical coding.
QQJ is an evaluation framework for generative AI that combines human judgment and LLMs. It uses expert-designed multi-dimensional rubrics and calibrates LLM evaluators on a small high-quality annotation set. Experiments on text and image generation show stronger alignment with human judgment than traditional automatic metrics and unconstrained LLM evaluators.
SafeLens introduces a two-tier video moderation architecture (fast-and-slow) to reduce inference costs. The framework filters SafeWatch dataset to 2.4% via influence-guided filtering and augments with Chain-of-Thought traces. It outperforms SafeWatch-8B, OmniGuard-7B, GPT-5.4, and Gemini-3.1-pro on real and AI-generated video benchmarks.
Comparative study of 29 data augmentation techniques for 2D echocardiography segmentation on U-Net. Anatomically plausible geometric transformations (affine, shift-scale-rotate, perspective, horizontal flip) improve cross-dataset generalization, while aggressive intensity augmentations degrade it. Pairwise combinations outperform individual augmentations.
arXiv study identifies three critical failure modes of vision-language models (VLMs) in clinical settings: over-reliance on text vs images, dependence on irrelevant clinical history, prompt sensitivity across semantically equivalent inputs. Testing on MIMIC-CXR shows VLM decisions dominated by text modality even when visual evidence is available.
Study of Compositional Geometry Routing Problem (CGRP), a generalization of routing problems covering points, lines, areas, and hybrid geometries. Proposes DiCon, a solver with differential attention and contrastive learning to handle asymmetry and enlarged action spaces. Results show strong performance, versatility, and superior generalization across diverse instances.
PEIRA is a non-contrastive self-supervised learning method analyzing JEPA dynamics through a regularized linear regressor. It minimizes an explicit objective based on the trace of the optimal regressor, ensuring stable non-collapsed equilibria aligned with canonical correlation subspaces. Competitive results on ImageNet-1K and CIFAR-10.
Prompt2Fingerprint introduces a framework for LLM fingerprinting via parameter generation. Instead of fine-tuning each model separately, a specialized generator maps textual descriptions to low-rank parameter increments in a single forward pass, eliminating the computational overhead of existing methods.
DocOS is a benchmark evaluating GUI agents capable of proactively searching online documentation to solve long-tailed tasks. Experiments reveal two bottlenecks: difficulty reliably locating relevant information and faithfully grounding retrieved instructions into precise GUI actions.
New zeroth-order hard-thresholding algorithm with variance reduction for ℓ0-constrained optimization. Addresses SZOHT's limitation on random directions by mitigating conflict between ZO gradient deviation and hard-thresholding expansivity. Improved convergence rates validated on ridge regression and black-box adversarial attacks.
MR-SLAM is a mixed reality system using Meta Quest 3 to teleoperate three TurtleBot3 robots in collaborative SLAM. The operator views the real world in passthrough with spatially anchored information panels. Three SLAM Toolbox instances merge occupancy grids in real time via ROS 2, achieving 94.7% cross-instance consistency and 8.83 Hz scan rate.