2026-07-16 科技日报

扫描 591 篇候选内容 · 覆盖 650 个信息源 · 纳入 579 篇

📌 今日要点

• TSCA-Net: Temporal-Spatial Clique Attention for Interpretable Multimodal Pedestrian Trajectory Prediction
• GenDiff: A Dose and Anatomy Aware Diffusion Model with Structural Prior Refinement for Low-Dose CT Reconstruction and Generalization
• Anomalous Frame Detection Using VLM-Based Description Comparison for Extracting Expert-Specific Actions and Contextual Decision-Making Scenes with Intra-Video Self-Similarity
• Contrastive Joint-Embedding Prediction for Representation Learning in Structural MRI
• SpikeDS: Dual Sparsity Spikformer for Perineural Invasion Prediction in 3D MRI
• LLM CoTs remain monitorable when being unfaithful requires computation

📝 科技简讯

• TSCA-Net: Temporal-Spatial Clique Attention for Interpretable Multimodal Pedestrian Trajectory Prediction — ArXiv CV (cs.CV)
• GenDiff: A Dose and Anatomy Aware Diffusion Model with Structural Prior Refinement for Low-Dose CT Reconstruction and Generalization — ArXiv CV (cs.CV)
• Anomalous Frame Detection Using VLM-Based Description Comparison for Extracting Expert-Specific Actions and Contextual Decision-Making Scenes with Intra-Video Self-Similarity — ArXiv CV (cs.CV)
• Contrastive Joint-Embedding Prediction for Representation Learning in Structural MRI — ArXiv CV (cs.CV)
• SpikeDS: Dual Sparsity Spikformer for Perineural Invasion Prediction in 3D MRI — ArXiv CV (cs.CV)
• LLM CoTs remain monitorable when being unfaithful requires computation — LessWrong
• Can we rely on law? — LessWrong
• Recap of bike trip/street interviews across America — LessWrong
• Extreme Power Concentration: A Map and Research Directions — LessWrong
• A Structural Similarity Between Two Open Corrigibility Questions — LessWrong
• OpenAI is now using AI to attack its own AI, and it’s working better than humans ever did — The Decoder
• GPT-5.6 Sol reportedly disproves a 30-year-old statistics conjecture in 90 minutes after humans couldn’t crack it — The Decoder
• Bonsai 27B is a full open reasoning model that fits on an iPhone — The Decoder
• Spotify bets Premium subscribers want to chat with their music player — The Decoder
• OpenAI’s Codex now encrypts instructions between AI agents, leaving developers blind to internal delegation — The Decoder
• Meta employees sue over layoffs they say were driven by discriminatory AI selection systems — The Decoder
• OpenAI’s first hardware product is a screenless AI speaker designed to feel alive — The Decoder
• Agentic orchestration: Enterprise AI organizations have a deployment problem, not a platform problem — and most are calling chatbots agents — VentureBeat AI
• Trust Is the Product: How Salesforce Thinks About Data in the AI Era — Salesforce AI Blog
• Breaking Down the Stack: Browser-Bound vs. Headless Architecture — Salesforce AI Blog
• A Practitioner’s Guide to AI-Assisted Org Discovery: Map the Org Before You Build — Salesforce AI Blog
• Storytelling in the AI Era: Bring Authenticity Back to Your Marketing — Salesforce AI Blog
• Simplify Security Visibility with Security Center Essentials — Salesforce AI Blog
• Optimal Adaptive Market Making: A Theoretical Framework for High-Yield Liquidity Provision in Perpetual Futures Markets — arXiv AI (cs.AI)
• In-Context Reinforcement Learning under Non-Stationarity: A Survey — arXiv AI (cs.AI)
• What building Shippy taught us about building agents — Hugging Face Blog
• Model Routing Is Simple. Until It Isn’t. — Hugging Face Blog
• Welcome Inkling by Thinking Machines — Hugging Face Blog
• Introducing Real World VoiceEQ: Measuring the human quality of voice AI — Hugging Face Blog
• Nokia’s AI-RAN platform: a radio comeback that runs on NVIDIA — AI News (Jack Clark)
• Why I Left Google DeepMind — AI Alignment Forum
• Built Technologies builds an AI-powered document intelligence solution on AWS to power agents across real estate finance — AWS Machine Learning
• Agentic vision: Building visual intelligence with Amazon Bedrock and MCP servers — AWS Machine Learning
• Monitor Amazon SageMaker Pipelines cross-account with custom Amazon CloudWatch dashboards — AWS Machine Learning
• EFLUX: Elastic Multi-Robot Formation Navigation and Adaptation with Agentic LLMs — ArXiv RO (cs.RO)
• Enabling 24-hour Agricultural Robotics: Unsupervised Day-to-Night Cross-Modal Image Translation for Nighttime Visual Navigation — ArXiv RO (cs.RO)
• Robust In-Hand Manipulation via Priors in Reinforcement Learning and Mechanical Design — ArXiv RO (cs.RO)
• GaitSpan: Growing Humanoid Locomotion from Walking to Running — ArXiv RO (cs.RO)
• A Behavioral State Vocabulary in Sony ERS-111 R-CODE — ArXiv RO (cs.RO)
• OmniPMNet: Bridging discrete and gridded PM10 forecasts via omni-query neural processes — ArXiv ML (cs.LG)

📊 数据概览