【AI最新動向 2026年3月12日】論文5件・GitHub5件

2026年3月12日2026年3月27日

📝 この記事のポイント

🚀 AI技術の最新動向 – 2026年3月12日世界中から収集したAI・機械学習の最新情報をお届けします 📑 目次 💻 注目のGitHubプロジェクト Kaleighc793/freqtrade-bot wanshuiyin/Auto-claude-code-research-in-sleep Genius-apple/RusTorch Aura-AI-Trading-Agent/Aura-AI-Trading-Agent caramaschiHG/awesome-ai-agents-2026 📌 関連記事もチェック 📚 最新研究論文 1. COMIC: Agentic Sketch Comedy Generation 著者: Susung Hong, Brian Curless, Ira Kemelmacher-Shlizerman We propose a fully automated AI system that produces short comedic videos similar to sketch shows such as Saturday Night Live. Starting with character references, the system employs a population of agents loosely based on real production studio roles, structured to optimize the quality and diversity… 論文を読む → 2. LiTo: Surface Light Field Tokenization 著者: Jen-Hao Rick Chang, Xiaoming Zhao, Dorian Chan We propose a 3D latent representation that jointly models object geometry and view-dependent appearance. Most prior works focus on either reconstructing 3D geometry or predicting view-independent diffuse appearance, and thus struggle to capture realistic view-dependent effects. Our approach leverage… 論文を読む → 3. Neural Field Thermal Tomography: A Differentiable Physics Framework for Non-Destructive Evaluation 著者: Tao Zhong, Yixun Hu, Dongzhe Zheng We propose Neural Field Thermal Tomography (NeFTY), a differentiable physics framework for the quantitative 3D reconstruction of material properties from transient surface temperature measurements. While traditional thermography relies on pixel-wise 1D approximations that neglect lateral diffusion, … 論文を読む → 💻 注目のGitHubプロジェクト 1. Kaleighc793/freqtrade-bot Freqtrade — Configuration and management tool for the Freqtrade open-source crypto trading bot with strategy backtesting, live/dry-run modes, 10+ exchange support (Binance, Kraken, OKX), hyperparameter optimization, FreqUI dashboard, and Rich terminal menu interface ⭐ 549 stars | 🔀 349 forks リポジトリを見る → 2. wanshuiyin/Auto-claude-code-research-in-sleep Claude Code skills for autonomous ML research: cross-model review loops via Codex MCP ⭐ 132 stars | 🔀 13 forks リポジトリを見る → 3. Genius-apple/RusTorch RusTorch is a production-grade deep learning framework re-imagined in Rust. It combines the usability you love from PyTorch with the performance, safety, and concurrency guarantees of Rust. Say goodbye to GIL locks, GC pauses, and runtime errors. Say hello to RusTorch. ⭐ 84 stars | 🔀 1 forks リポジトリを見る → 4. Aura-AI-Trading-Agent/Aura-AI-Trading-Agent AI trading agent for stocks & crypto ⭐ 53 stars | 🔀 51 forks リポジトリを見る → 5. caramaschiHG/awesome-ai-agents-2026 🤖 The most comprehensive list of AI agents, frameworks & tools in 2026. 300+ resources · 20+ categories · Updated monthly. ⭐ 49 stars | 🔀 12 forks リポジトリを見る → 📚 あわせて読みたい「プロンプトは戦略だ！」と学んで劇的効率UP！私のAI活用術私が実感！AIで物流の悩み解消、時間もコストも浮いた話 AI三つ巴！Geminiと私が出会って、ストーリー作りは変わった？。

🚀 AI技術の最新動向 – 2026年3月12日

世界中から収集したAI・機械学習の最新情報をお届けします

📑 目次

💻 注目のGitHubプロジェクト
1. Kaleighc793/freqtrade-bot
2. wanshuiyin/Auto-claude-code-research-in-sleep
3. Genius-apple/RusTorch
4. Aura-AI-Trading-Agent/Aura-AI-Trading-Agent
5. caramaschiHG/awesome-ai-agents-2026
📌 関連記事もチェック

📚 最新研究論文

1. COMIC: Agentic Sketch Comedy Generation

著者: Susung Hong, Brian Curless, Ira Kemelmacher-Shlizerman

We propose a fully automated AI system that produces short comedic videos similar to sketch shows such as Saturday Night Live. Starting with character references, the system employs a population of agents loosely based on real production studio roles, structured to optimize the quality and diversity…

論文を読む →

2. LiTo: Surface Light Field Tokenization

著者: Jen-Hao Rick Chang, Xiaoming Zhao, Dorian Chan

We propose a 3D latent representation that jointly models object geometry and view-dependent appearance. Most prior works focus on either reconstructing 3D geometry or predicting view-independent diffuse appearance, and thus struggle to capture realistic view-dependent effects. Our approach leverage…

論文を読む →

3. Neural Field Thermal Tomography: A Differentiable Physics Framework for Non-Destructive Evaluation

著者: Tao Zhong, Yixun Hu, Dongzhe Zheng

We propose Neural Field Thermal Tomography (NeFTY), a differentiable physics framework for the quantitative 3D reconstruction of material properties from transient surface temperature measurements. While traditional thermography relies on pixel-wise 1D approximations that neglect lateral diffusion, …

論文を読む →

💻 注目のGitHubプロジェクト

1. Kaleighc793/freqtrade-bot

Freqtrade — Configuration and management tool for the Freqtrade open-source crypto trading bot with strategy backtesting, live/dry-run modes, 10+ exchange support (Binance, Kraken, OKX), hyperparameter optimization, FreqUI dashboard, and Rich terminal menu interface

⭐ 549 stars | 🔀 349 forks

リポジトリを見る →

2. wanshuiyin/Auto-claude-code-research-in-sleep

Claude Code skills for autonomous ML research: cross-model review loops via Codex MCP

⭐ 132 stars | 🔀 13 forks

リポジトリを見る →

3. Genius-apple/RusTorch

RusTorch is a production-grade deep learning framework re-imagined in Rust. It combines the usability you love from PyTorch with the performance, safety, and concurrency guarantees of Rust. Say goodbye to GIL locks, GC pauses, and runtime errors. Say hello to RusTorch.

⭐ 84 stars | 🔀 1 forks

リポジトリを見る →