📂 Category
Research & Benchmarks News Archive - Page 6 of 10
932 articles in this category • Page 6 of 10
- 501. Explainable ML Classifies Alzheimer's Early in 1,641 ADNI Subjects
- 502. MIT researchers train AI to read charts, streamlining downstream workflows
- 503. Lightweight CNN Boosts Adversarial Robustness in EEG‑Based Brain‑Computer Interfaces
- 504. Hundreds sign Leiden Declaration as AI threatens mathematicians' profession
- 505. Transformer tops Gait2Hip-60 benchmark with 0.819 R² in hip force prediction
- 506. QASM-Eval Introduces First Dataset for Training LLMs on OpenQASM-3
- 507. Parallax adds learned covariance correction to linear attention, retains softmax
- 508. Men use AI coding agents over twice as often as women; economists at 39%
- 509. Molecule-trained AI gives better chicken pairing suggestions than recipe AI
- 510. AI search agents favor confirming hits, sideline gut answers, study finds
- 511. OpenAI gives free life‑sciences AI model to aid government pandemic prep
- 512. Review paper claims code defines AI agents' reasoning and behavior
- 513. NVIDIA research moves robotics simulation to reality, revealing robot confusion
- 514. CVPR 2026 Friday Session: STARFlow‑V Video Modeling Poster #178, 4‑6 PM
- 515. USD E^3USD ‑Agent splits fast router from LLM meta‑controller for edge inference
- 516. Sakana AI's DiffusionBlocks Apply Uniform [4,4,4] Layers Across Three Blocks
- 517. AI Agent Auto-Identifies Unreadable Model Parameters from CSV Files
- 518. Learn to Build AI Projects: n8n Automation, Financial Data, Summaries, Reports
- 519. AI Agents Falter in Production as Backward Design Overburdens Model
- 520. Hugging Face releases LeRobot Humanoid: 3D‑printable legs for robot research
- 521. Synthetic 1,000‑Customer Dataset Uses Gender and Income to Test Bias
- 522. SciAtlas Introduces Large-Scale Knowledge Graph to Aid Automated Research
- 523. Google outperforms OpenAI on math benchmark, winning 9 to 1 ratio
- 524. ByteDance study: LMMs answer questions better than full-page transcription
- 525. Language Models Forecast Research Success Using 11,488 Comparative Idea Pairs
- 526. Researchers use triplet loss to train high-quality Horn logic embeddings
- 527. Positive-IC 46.4% indicates negative bias; |IC| just under 0.02 after two runs
- 528. AgentNLQ released as a general‑purpose NL2SQL agent; accuracy lags human writers
- 529. SuperAI Conference Highlights Growing AI Startup and Infrastructure Scene in Asia
- 530. CODEX Agent Adds AI‑Q Deep Research Skill from GitHub Repository
- 531. AI models learn chemistry; talent and collaborations offset location concerns
- 532. Real-Time Diffusion on Apple M3 Ultra: CoreML, Quantization, Neural Engine
- 533. Evaluating AI Agents: Does the Engine Grasp Instructions and Reason Facts?
- 534. AgentWall adds runtime safety layer for local AI agents' actions
- 535. LangSmith Engine automates agent debugging; OpenAI's Frontier offers platform
- 536. VideoWorld paper links prediction, simulation, reasoning in robotics
- 537. Channel-independent tolerate modalities but falter on within-modality gaps
- 538. Study Finds Current ToM Benchmarks Overlook First‑Person, Dynamic Interaction
- 539. Graph‑Enhanced RAG Architecture Cuts Latency in Meta‑Scale Production
- 540. OpenClaw founder runs 100 AI agents for USD 1.3M/month code, review PRs, find bugs
- 541. New benchmark shows AI video generators look realistic but lack reasoning
- 542. Researchers train AI model achieving near-full performance using 12.5% of experts
- 543. RecursiveMAS cuts multi-agent inference time 2.4×, slashes token use 75%
- 544. ArXiv to ban authors of papers with unchecked LLM‑generated content
- 545. BenchJack proposes secure-by-design AI benchmark audit with eight flaw taxonomy
- 546. 12‑Metric AI Agent Eval Harness Built in 9‑14 Days Across 100+ Deployments
- 547. Google DeepMind adds Gemini-powered cursor to Chrome for visual queries
- 548. BaLoRA adds Bayesian uncertainty to low‑rank adaptation, but lags fine‑tuning
- 549. Community review tools guide novices in AI research, study finds
- 550. Tilde Research's Aurora optimizer beats Muon and NorMuon at 340M scale
- 551. OpenAI unveils Daybreak to secure Codex, with industry and government rollout
- 552. New embeddings prioritize preferential similarity over semantics for clustering
- 553. Baidu's Ernie 5.1 Cuts 94% Pre‑Training Costs Using Once‑For‑All Framework
- 554. Hermes Agent tops use as Nous Research’s self‑improving model leads OpenRouter
- 555. Palisade Research: Open‑weight AI like Qwen boost autonomous hacking
- 556. Study proposes method to curb AI reward hacking in safety tests
- 557. Build Python Vector Search with Cosine Similarity for Scale‑Invariant Matching
- 558. AI success shifts from 95% accuracy to latency, cost, and reliability
- 559. Apple Workshop Shows ML with Homomorphic Encryption, Georgia Institute, CISPA
- 560. OpenAI opens GPT-5.5-Cyber to vetted security researchers, adds three tiers
- 561. LightSeek launches TokenSpeed, cutting LLM latency by half vs TensorRT-LLM
- 562. CLIP-FP8 Model Matches CLIP-FP16 Quality; Patch Embedding Quantizers Matter
- 563. Automation updates AI context morning with active threads, key dates, note
- 564. Google DeepMind buys minority stake in EVE Online studio for AI testing
- 565. Meta AI releases NeuralBench, benchmark for 36 EEG tasks, 94 datasets
- 566. iTARFlow Shows Competitive Performance on ImageNet 64‑256px Resolutions
- 567. CreativityBench benchmark introduces 4K‑entity affordance KB to test LLM creativity
- 568. Self-Attentive Meta-Optimizer Adds Gradient Alignment and Group-Adaptive Rates
- 569. Local edits in LLM-driven NAS can trigger broader performance shifts
- 570. Groq‑Powered Agentic Assistant Uses Sub‑Agent to Catalog 2024‑25 SLMs
- 571. AI autoencoders and joint communications‑sensing rank among 6G enablers
- 572. Anthropic adds 'dreaming' feature to Claude Managed Agents for memory recall
- 573. Anthropic's USD 200 B, five‑year Google Cloud deal makes up >40% of backlog
- 574. MRC retires paths, then probes to confirm failures and recovery
- 575. eOptShrinkQ enables near‑lossless KV cache compression with spectral denoising
- 576. Harvard study finds OpenAI's o1 and 4o outdiagnose ER doctors in 76‑patient test
- 577. 2021 EDEN-unbiased quantizer beats 2026 successor in average accuracy
- 578. US benchmark shows China lagging; Deepseek model underperforms private tests
- 579. Google DeepMind AI co‑clinician beats GPT‑5.4 in blind tests, lags docs
- 580. Anthropic benchmark says Claude matches experts, 23 tasks remain ambiguous
- 581. Grok Voice Think Fast 1.0 lets non‑programmers design agents via console.x.ai
- 582. WPI professor Gerych offers solution to AI vision ‘Whac‑a‑mole’ bias dilemma
- 583. New method advances privacy‑preserving AI training on consumer devices
- 584. Musk says he was duped, warns AI could kill us, xAI to IPO via SpaceX in June
- 585. NVIDIA BioNeMo wraps CPU layer with DistributedTriangleMultiplication
- 586. DeepSeek unveils new AI breakthrough as nation tightens grip on departing firms
- 587. Poolside AI launches Laguna XS.2 and M.1, hitting 72.5% on SWE-bench Verified
- 588. Oracle abandons its legacy, pivots to AI in an unconventional approach
- 589. New Architecture Separates Execution and Review Agents for Tool-Calling
- 590. MIT study links language model scaling success to superposition of concepts
- 591. Google staff urge Sundar Pichai to reject classified military AI projects
- 592. AI framework autonomously optimizes data, models, algorithms, outperforms humans
- 593. MolClaw Introduces Autonomous Agent for Hierarchical Drug Screening
- 594. Lakehouse concept drives AI data access for thousands of enterprise users
- 595. Fine-tuning RAG embeddings may drop retrieval accuracy 40%, study finds
- 596. AI pipelines show silent failures from orchestration drift, detected weeks later
- 597. OSWorld Benchmark Evaluates LLMs on Real Computer Use, Unlike Text‑Only Tests
- 598. PageIndex Retrieves via Reasoning Using OpenAI gpt-5.4 Model
- 599. Discord Users Access Anthropic's Mythos AI Tool Without Authorization
- 600. Google DeepMind's Vision Banana Outperforms SAM 3 and Depth Anything V3