2026-10-02 科技日报

扫描 444 篇候选内容 · 覆盖 650 个信息源 · 纳入 427 篇

📌 今日要点

• SynIL: Leveraging Synergy for Offline Imitation Learning from Imperfect Demonstration Datasets
• A Two-Echelon Covering Tour Vehicle Routing Problem with Drones for Post-Disaster Relief
• TALK-Dem: Benchmarking Embodied Task Planning under Dementia-Associated Communication Patterns
• The eternal complement
• How Albertsons Companies is reimagining retail from the inside out
• 何恺明团队新作：看猫片就能学会 ARC 挑战

📝 科技简讯

• SynIL: Leveraging Synergy for Offline Imitation Learning from Imperfect Demonstration Datasets — ArXiv RO (cs.RO)
• A Two-Echelon Covering Tour Vehicle Routing Problem with Drones for Post-Disaster Relief — ArXiv RO (cs.RO)
• TALK-Dem: Benchmarking Embodied Task Planning under Dementia-Associated Communication Patterns — ArXiv RO (cs.RO)
• The eternal complement — OpenAI Blog
• How Albertsons Companies is reimagining retail from the inside out — OpenAI Blog
• 何恺明团队新作：看猫片就能学会 ARC 挑战 — 量子位
• 谷歌 Gemini 4 突然发布！RSI 加持，GPT 和 Opus 都让让 — 量子位
• Evaluating Multi-Task Morphological Concept Learning for Pulmonary Nodule Malignancy Assessment in 3D CT — ArXiv CV (cs.CV)
• Masked Swingers: Harnessing Data Augmentation to Advance Autoencoders for Self-Supervised Learning — ArXiv CV (cs.CV)
• GaugeVLM: Structuring Spatial Supervision with Measured Geometric Interventions — ArXiv CV (cs.CV)
• It Takes Little to Rewrite Perception: Targeted Semantic Substitution in Vision-Language Models at $\epsilon \leq 4/255$ — ArXiv CV (cs.CV)
• Strike a Chord! Modal Kinetic Typography — ArXiv CV (cs.CV)
• 刚刚，GPT-6 Astra 破解拿破仑百年悬案！217 年密信曝光，信息量太大了 — 新智元
• 谷歌 DeepMind 研究员宣称 Gemini 4 已实现 RSI！ — 新智元
• 突发！特朗普签行政令：美国政府全面封杀「AI」 — 新智元
• Astra 成功通关「我不是机器人」！人类最后一道防线，崩塌了？ — 新智元
• NeurIPS’26 | 港科大：从 CPU 到 GPU，AI 能否真正加速全流程？ — 新智元
• Agents Can Mimic Good Design. Ours Know Why It Works — Salesforce AI Blog
• What’s New in B2C Commerce: The September 2026 Release — Salesforce AI Blog
• Founder-led Sales Strategies: The Startups and Small Business Edition — Salesforce AI Blog
• Improving OCR Faithfulness via Gated and Attenuated On-Policy Distillation — arXiv AI (cs.AI)
• AREX-2: Advancing Self-Improving Agents through Long-Horizon Reflective Tasks — arXiv AI (cs.AI)
• MoFlow: Multi-Objective Agentic Workflow Generation — arXiv AI (cs.AI)
• AI Agents are Vulnerable to Radicalization — arXiv AI (cs.AI)
• CARAT: Do Materials LLMs Reason or Recite? — arXiv AI (cs.AI)
• ChatGPT can now virtually try on clothes for you — TechCrunch AI
• Google thinks SpaceX’s Starship has to launch 1,800 times before space data centers get off the ground — TechCrunch AI
• OpenAI cuts ties with 3 safety researchers, WSJ reports — TechCrunch AI
• Opus 5.5 loves to tell you ‘this matters’ (and other AI writing tells) — TechCrunch AI
• Amazon releases its own Jev clone as decision models flood the web — TechCrunch AI
• Shopify debuts Canvas, a way to build online stores by chatting with AI — TechCrunch AI
• Autoencoders vs. PCA: I Rigged the Test and PCA Still Won — Towards Data Science
• Can an Apartment Search Agent Call the Model Fewer Times and Still Find Good Matches? — Towards Data Science
• What the ReLU Revolution Revealed About Biological Plausibility — Towards Data Science
• Your AI Bill Is a Toll Booth. Stop Paying Twice. — Towards Data Science
• Nearly half of test subjects mistook Tavus’ AI video avatar for a real person on a one-minute call — The Decoder
• Ideogram says its new model can edit part of an image without messing up the rest — The Decoder
• Anthropic brings Claude to civilian agencies as its fight with the Pentagon drags on — The Decoder
• Security startup finds more than 13,000 internal company screenshots that AI agents uploaded publicly — The Decoder
• AI beats Stratego’s greatest player, ending one of the last human strongholds in board games — The Decoder

📊 数据概览