Machine learning news — deep learning, reinforcement learning, neural architecture search, diffusion models, and new ML frameworks and libraries.
FalkorDB's GraphRAG SDK is positioned as a key tool for network analysis applications in 2026, with ten production use cases including knowledge graph construction for RAG, fraud ring detection, and supply chain optimiza…
JetBrains released PyCharm 2026.2.1, adding AI agent support for Jupyter notebooks with live kernel execution and an environment coordinator that directs agents to the project's configured Python interpreter and tool (uv…
JetBrains introduced a new Jupyter skill in PyCharm that lets AI agents operate a live Jupyter kernel, preserving state across cells and avoiding notebook corruption. In tests with Claude Opus 5 on twelve MLGym tasks, th…
A study published in Water Resources Research by Yang et al. (2026) applies generative diffusion models—the technology behind AI photo erasers—to streamflow prediction, achieving improved performance over baseline and ex…
A new tutorial by Emeka Okafor demonstrates building a local Python pipeline that uses three open-source Hugging Face classifiers to detect AI-generated images, audio, and text, emitting a JSON verdict and exit code to g…
A developer's blog post series on building production RAG systems continues with a focus on the retrieval layer, arguing that vector search alone is insufficient and that hybrid search with reranking is necessary. The po…
DeepSeek, a Chinese AI lab, has developed a novel training technique that reduces the cost of training large language models by up to 90%, according to a technical report published on the company's website. The method, w…
Thinking Machines' 150M-parameter BDH-CQ reasoning model, which combines in-context learning with recurrent latent reasoning, achieves 29.5% pass@2 on the public ARC-AGI-1 evaluation set at a computed inference cost of $…
Tencent HY LLM Frontier researchers and collaborators demonstrated that AI agent training tasks can be generated synthetically for $0.05 each, producing 1,000 accepted tasks for $50. Fine-tuning open-weights models on th…
Reinforcement learning researcher Hyunsoo Chung argues that, like RL agents, people should stop imitating others and instead learn from their own experiences to surpass their teachers. Chung, who works on deep research a…
Project 0, in partnership with Gamma, launched two AI-powered vaults on Solana that use machine learning to automate yield strategies, targeting SOL/LST and stablecoin yield. The vaults leverage Project 0's multi-venue p…
Data scientist Lauren Leek's essay 'Temperature Zero for Culture: Why Everything Is Starting to Look the Same' argues that in a data-driven society, predictions become self-fulfilling prophecies, leading to cultural homo…
A developer's blog post explains how CatBoost's ordered target statistics prevent target leakage in categorical feature encoding, a common issue in tabular machine learning. The post illustrates the problem with an analo…
Rockwell Automation announced on August 11 an API-enabled integration between its Plex Quality Management System (QMS) and FactoryTalk Analytics VisionAI, bringing AI-based visual inspection results into quality-manageme…
A new thesis introduces scalable tools for nested optimization in deep learning, addressing bilevel problems such as hyperparameter optimization and generative adversarial networks, where classical methods fail at scale.
Researchers propose JacNet, a neural network architecture that directly learns the Jacobian of an input-output function, enabling easy enforcement of structural priors such as invertibility and k-Lipschitz continuity. Th…
Temple Compute released Pantheon, a collection of production-ready scientific workflows for its Horus Runtime, covering molecular dynamics, virtual screening, docking, and LLM training. The workflows are multi-stage pipe…
A new study from researchers demonstrates that large language models (LLMs) can match or outperform traditional hyperparameter optimization (HPO) methods such as Bayesian optimization within constrained search budgets. B…
A team at Cua has developed a process-scoped Metal capability shim that unlocks 11-16x faster LLM inference in macOS virtual machines on Apple Silicon, bringing VM performance to 94-99% of bare-metal speeds. The bottlene…
A 32-attempt Claude Code evaluation on August 11, 2026, found that Langfuse was selected in every tracing and production-monitoring task but in none of the prompt comparison and release-gate tasks. However, the analysis …