Search the full wire by company, model, lab, or keyword. Every story we have ever aggregated.
Robinhood’s Abhishek Fatehpuria on winning the modern financial consumer at TechCrunch Disrupt 2026. Register now to save up to $200 before September 25 at 11:59 p.m. PT.
Neural Cellular Automata transfer learning mechanism injecting pretrained hidden states to improve few-shot generalization and parameter efficiency.
AutoRecLab automates recommender-system experiments from natural-language prompts using RAG and code generation.
TrialAtlas uses memory-augmented multi-agent systems to optimize clinical trial design and reduce drug development failure rates.
Novel Poisson-process speculative sampling algorithm improves trade-off between LLM inference efficiency and watermarking for output provenance.
Fine-tuning Qwen2.5-7B and Ministral-8B on personality-labeled corpora improves consistency vs. instruction prompting for social agents.
Lightweight embedding translator and optimization sweep reduce diffusion model footprint for on-device inference on consumer GPUs.
Reusing ASR encoder states as sentence embeddings improves query-conditioned topic localization in transcripts without separate audio encoder.
EnterpriseVal benchmark measures GenAI reliability, safety, and ROI on enterprise workflows to predict deployment success vs. cancellation.
Simon Willison discusses animation techniques from 1988 film Who Framed Roger Rabbit, focusing on a pelican-bicycle scene.
Federated deep clustering networks extend unsupervised learning to distributed private data settings with improved reliability.
Touvigation combines vision-language models with spatial mapping for low-latency, embodied object acquisition guidance for blind/low-vision users.
RheoSampling fixes dynamic-tree speculative decoding collapse in stochastic sampling (T>0) to preserve context-aware topology and acceptance rates.
Bayesian VAE-SSM for personalized fluid resuscitation via uncertainty-aware hemodynamic control.
Matrix AdaGrad: structured adaptive optimization framework for matrix-valued neural network parameters.
MIST multimodal model fuses genomic and histology data via attention for cancer survival prediction.
Agentic sleep JITAI system adapts behavioral interventions via LLM review of personal activity data.
Statistical feature pool with LLM-generated signals exceeds neural baselines on TSB-AD-U time-series anomaly detection.
CASCADE: systematic study of multi-stage defense combinations against LLM jailbreaks under unified threat model.
RegKT: IRT-regularized deep knowledge tracing improves interpretability and robustness for educational applications.
Aetiology-specific dysarthria severity models using contrastive embeddings and phonological pseudo-labelling.
PARTS: subtask RL framework for real-world robot manipulation fine-tuning on long-horizon task bottlenecks.
Google expands AI & Economy team with academic advisors and fellows; no technical details on research focus or deliverables provided.
Audit of medical vision-language models (BioMedCLIP, CheXficient, MedSigLIP) on chest X-ray TB screening reveals ranking instability across datasets.
ML method for accelerating plane-wave DFT in materials science by completing neural electronic initialization under PAW formalism.
Researchers used Claude to reach an OpenAI employee account and sensitive GitHub data.
Bilevel optimization framework automating hyperparameter tuning for topology optimization design problems.
Study reveals vision-language models fail to meaningfully use patient ECGs in clinical prediction, proposing detection and mitigation strategies.
Physics-informed digital twins for robotic manipulation learn object properties from instrumented human interaction data.
Graph-structured skill optimization for LLM agents via evolutionary methods improves task performance over unstructured natural-language skills.