Text Models Spot Evasion Better Than They Read Vocal Confidence
Models performed strongly on written signs of evasion but struggled more with vocal unconfidence, and speaker-level calibration helped only modestly.
Scientific publication4 min read
Subject
Artificial Intelligence
Machine learning, neural networks, language models, and computer vision.
548 published articles6 specialties
Latest in AI
Brain MRI Review Finds Safety Checks Rarely Test Small Lesions
A systematic review finds that most deep-learning brain MRI reconstruction studies did not compare image-fidelity scores with radiologist assessments on the same data, leaving key safety questions unanswered.
Scientific publication4 min read
Medical AI method links uncertainty to features in each prediction
A preprint introduces egRUE, a method that combines uncertainty scoring with feature-level explanations; its reported gains came from benchmark tests and a small BloodMNIST expert study.
Scientific publication5 min read
AI Models Struggle to Rebuild How Chinese Internet Words Are Made
A new benchmark finds that language models often handle the meaning of Chinese internet neologisms better than the sounds, characters and source forms used to create them.
Scientific publication5 min read
AI model scores higher at finding camouflaged copper-tube defects
CF-YOLO, a YOLOv11-based detector built to refine context and fine detail, scored higher than YOLOv11n on CTDD but delivered mixed results on external NEU-DET data.
Scientific publication4 min read
Simple color forecast beat trained models on unseen copper
A tiny two-recording test found that a parameter-free color extrapolator beat copying the last frame and every trained comparator on held-out copper in both directions. The air-to-chamber advantage was individually separated from zero, while the chamber-to-air interval included zero.
Scientific publication5 min read
Preprint reports higher LLM scores with NVFP4 scale refinement
H-Scale uses calibration activations to choose hardware-valid NVFP4 scales, with higher average scores reported across Qwen3 and LLaMA tests but a narrow evaluation scope.
Scientific publication4 min read
Adaptive forecast weighting leads three datasets, but not all
Gibbs-family methods led the aggregate scores in Traffic Hourly, Electricity Hourly and Solar Weekly, but classical baselines remained on top in some M4 frequency and disagreement groups.
Scientific publication4 min read
AI agents score higher with verifier-guided training, preprint reports
VICT uses a task's terminal verifier to trace credit through long action sequences instead of using it only as a final pass-or-fail signal. In ALFWorld and WebShop tests, it beat GRPO and achieved higher validation AUC over the same 300 updates, while the authors caution that verifier-defined links do not establish causal necessity.
Scientific publication4 min read
Video AI Retains Near-Full Scores With Just 10% of Visual Tokens
A video AI method called Token-Budget Distillation retained strong benchmark scores after aggressive visual-token compression, although its training still depended on the uncompressed teacher model.
Scientific publication5 min read
Chest X-ray model reports stronger case retrieval for less common findings
An anatomy-aware system called CheXtriev reported stronger case-retrieval scores than global and local comparison methods on selected chest radiographs. The gains were especially notable for several lower-prevalence findings.
Scientific publication4 min read
AI models miss subtle power cues in French and Egyptian Arabic dialogue
A preprint examining French and Egyptian Arabic movie dialogue finds that six AI systems align more closely with humans on visible social cues than on subtle power relationships, while multimodal results are limited by incomplete coverage.
Scientific publication4 min read
Geometry changes how much softmax attention can be compressed
A mathematical study links the worst-case rank needed to approximate normalized attention to support geometry, while a fixed BERT-base calibration reports lower effective dimensions in some tested cells.
Scientific publication5 min read
Overnight Bed Sensors Offer Modest Clues to Next-Day Agitation
The model with the highest pooled scores retained minute-level overnight patterns, but the small hospital sample and limited calibration do not support individual care use.
Scientific publication4 min read
AI Models Score High on Easy Finance Questions, Then Stumble on Hard Ones
A new preprint benchmarks language models on questions aligned with CFA Levels I to III and FRM Parts I to II. Leading systems exceeded 97% on Easy items but fell sharply on Hard cases, while the gated result was 0.39 percentage points higher on held-out questions.
Scientific publication5 min read
AI model reaches 97% accuracy identifying Bangladeshi mango varieties
EfficientNetB0 correctly classified 97.36% of 303 held-out mango images, with eight errors, in a study that also deployed the model through a public web app. The result is an initial within-dataset benchmark, not evidence that the tool will deliver the same performance across regions or improve agricultural decisions.
Journal of Bangladesh Academy of Sciences, vol. 50, Supplement 1, p. 114, 20265 min read
Researchers Propose an Operating System for Shared Human-AI Work
The paper presents CrabOS as an operating-system approach to human-AI handoffs, using shared text objects and common capability controls, but reports architecture and case studies rather than measured evidence of better results.
Scientific publication4 min read
Scientific Text Reconstructs Much of Reactome's Global Shape
A text-only reconstruction of Reactome preserved its large-scale network pattern, but the study measured graph similarity rather than biological correctness.
Scientific publication3 min read
Video AI Cuts Video-Localization Time 79-Fold in Benchmark Tests
The fixed-round PTD model reported sharp speed gains and higher scores across several video benchmarks, while leaving disjoint events and multiple matching targets largely unexplored.
Scientific publication4 min read
One OCR model reports fewer errors across 13 Indic scripts
A preprint describes one OCR system trained across 13 Indic scripts; its reported overall character error rate was 6.9%, compared with 8.6% for monolingual models.
Scientific publication4 min read
More privacy may improve long-term data use, model suggests
A theoretical and simulation study suggests that private mean estimation can outperform non-private estimation over time when leakage-participation feedback is strong enough.
Scientific publication5 min read
Image generators lose count as crowds grow, preprint finds
Tests across nine systems found a sharp drop in numeric precision as prompts asked for more objects. Layout, composition and appearance also mattered, but high-count results were harder to validate.
Scientific publication5 min read
New AI decoder guarantees valid syntax, but task scores still vary
The preprint reports perfect syntax across the tested settings, while task-accuracy scores for SQL and linear temporal logic remained uneven.
Scientific publication4 min read
Language model gains on early forecasts of process steps and time
D-TAIA combined parameter-efficient language-model adaptation, domain-aware pre-training and retrieval, then matched or improved on LLM and RNN baselines across four event logs.
Scientific publication5 min read
AI routing was more accurate when user profiles mattered
A benchmark study found the largest routing gap when user-profile details changed which skill best matched a task, while profile-insensitive cases stayed similar.
Scientific publication3 min read
Spectral Signals Lead BCG Detection of Sleep Apnea Events
Spectral and frequency-based features gave the clearest signal for detecting respiratory events in ballistocardiography recordings, with nonlinear classifiers performing strongly when each patient was held out for testing. The results point toward compact systems, but remain limited to one hospital-based sensor setup and cohort.
Scientific publication5 min read
Flow Matching outperforms diffusion in sparse-view CT test
A preprint reports stronger image-quality results from Flow Matching than diffusion on sparse-view chest CT, with FlowDPS leading the reconstruction comparison.
Scientific publication4 min read
AI System Reports Higher Scores With First-Error Tracking
A diagnose-then-correct system called DoCtOR reported higher success rates on HotPotQA, ChartQAPro and Mind2Web. Its attribution model was less accurate on held-out failures, and open-ended, multilingual and larger-team settings were not tested.
Scientific publication4 min read
Residual-guided neural networks show higher accuracy on 71 benchmarks
A preprint reports higher average accuracy, lower variability and better average ranks for a residual-guided neural-network procedure tested against RVFL, ELM and BLS baselines on 71 UCI datasets.
Scientific publication4 min read
Study reports lower blood-pressure errors using buffered sample selection
A buffered sample-selection method called RECAST reported lower blood-pressure estimation error than TTC and no adaptation on PulseDB and MC-MED. The results came from simulated streaming benchmarks, with extra but sub-second latency and no prospective clinical evaluation reported.
Scientific publication5 min read
Urban Forecasting Model Reports Lower Errors With Sparse Ride Data
A computational preprint reports that TransMod had the lowest listed forecasting errors in New York City and Chicago and remained competitive when ride-hailing data were limited.
Scientific publication4 min read
Preprint reports proof of determinant conjecture for real-node matrices
A mathematical preprint reports a proof of Colombo's determinant conjecture for even-dimensional difference-power matrices with pairwise-distinct real nodes. It also gives an exact rank formula and determinant sign rule, while the formal verification described covers the odd-exponent derivation.
Scientific publication5 min read
AI Argument Search Showed Better Stance Matching, With a Topic Trade-Off
A preprint reports better support-versus-attack retrieval in tests of three embedding models, alongside evidence of reduced sensitivity to the claim topic in some fine-tuned comparisons.
Scientific publication5 min read
UAV mapping model reports higher 3D reconstruction scores
GeoFF3D, a feed-forward system for large UAV image collections, reported higher reconstruction scores than Pi3X + SLRF across nine aerial mapping blocks.
Scientific publication4 min read
Vision-Language Encoders Shift Which Clues Matter During Search
A controlled test suggests that visual evidence does not have one fixed ranking: the next useful clue can depend on what has already been acquired and on the target being sought.
Scientific publication4 min read
Bangla medical system scores 0.95 on a crafted test set
A Bangla question-answering system reported strong results on a manually built test set, while a small student review found high satisfaction. The study did not test patient outcomes, clinical safety, or performance on an external benchmark.
Scientific publication4 min read
A new sequence encoding keeps time shifts exact in one test
A methods preprint compares three real-valued position encodings, finding similar classification averages and an exact, faster algebraic shift for its Sinusoid variant.
Scientific publication4 min read
Training Helps AI Match Medical Images Across Wavelengths
A computer-vision study reports that a sensor-agnostic training approach improved dense image matching under spectral mismatch, but its real medical demonstrations were qualitative.
Scientific publication4 min read
Preprint links denoising to lower 3D crop reconstruction error
A two-stage system that denoises partial plant scans before completing missing structure showed lower real-data reconstruction error, while synthetic rankings varied by metric and dataset.
Scientific publication5 min read
Study reports a more targeted way to select what language models forget
GRACE, a gradient-guided method for selecting compact forget and retain sets, reported higher forget-sample retrieval accuracy than RASLIK and the highest model utility in six of eight algorithm-model combinations, while forget-quality scores varied little across selectors.
Scientific publication4 min read