Machine learning news — deep learning, reinforcement learning, neural architecture search, diffusion models, and new ML frameworks and libraries.
A new arXiv paper (2608.18177v1) proposes reversible forgetting, a framework for managing obsolete knowledge in enterprise AI agents, introducing three memory states—active, dormant, and retired—and a Hysteretic Reversib…
Researchers introduced HB-SJD, a batched Speculative Jacobi Decoding rollout backend for visual on-policy distillation, which accelerates training of compact visual autoregressive models by processing multiple tokens in …
Researchers propose DyG$^2$T, a dynamics modeling framework that uses a 3D Gaussian Temporal-Spatial Particle Graph Transformer to predict object motion trajectories from limited visual observations, achieving accurate d…
Researchers introduced Entropy-Constrained Adaptive Stochastic Quantization (ECASQ), a method that jointly selects adaptive quantization values to minimize Mean Squared Error under an entropy budget and unbiasedness cons…
Researchers propose TTSD-FAR, a test-time self-distillation framework with Fisher-Anchored Restoration, to handle missing-modality emotion recognition in large video-language models (LVLMs). The method uses a frozen teac…
A new configurable privacy-preserving MRI processing workflow extends deep learning-based brain extraction with adaptive anatomical preservation, interactive preservation selection, and integrated quality control, accord…
Researchers at the Center for Computational Neuroscience, Flatiron Institute, and New York University developed a computational framework integrating spectral regularization with bidirectional predictivity analyses, demo…
Researchers introduced Top-K prompting and the C3LM (Chemistry Constraint-Consistent Language Model), trained on the CREED-CCV-2+USPTO-XL dataset of ~45.6 million verified reactions, achieving state-of-the-art performanc…
A new arXiv paper (2608.18339v1) proposes a test-time adaptation method for vision-language models (VLMs) that unifies inference and adaptation objectives via optimal transport, improving accuracy by up to 7% over state-…
A new open-source benchmark from researchers at arXiv introduces 954 olympiad geometry problems, including a 297-problem hard subset, each paired with a human-authored Asymptote diagram, to measure diagrammatic reasoning…
Researchers introduced SparsePR, a training-free sparse attention method that combines Response-Coupled Partitioning with Probe-Fitted Residual Reconstruction to accelerate video transformers. Across four video generatio…
Researchers introduced BudgetDoc, the first multimodal benchmark for model-budget-performance trade-offs in document tasks, and trained DRB (Document-Reasoning Balancer), a 1B-parameter pre-flight estimator using SigLIP-…
Google's Gemini 2.5 Flash and text-embedding-004 models, combined with ChromaDB and PyMuPDF, form a RAG pipeline that processes approximately 2GB of data (500,000 text chunks) for technical documentation retrieval. The p…
Researchers introduced Abra, a controlled family of flow-matching transformer models, in a paper submitted to arXiv on August 18, 2026, reporting that text-to-image diffusion models scale predictably across three orders …
A developer has published a guide explaining how to deploy a Retrieval-Augmented Generation (RAG) API on AWS App Runner, using Node.js 22, DynamoDB vector search, and Claude for answer generation. The post details the us…
A six-author arXiv paper submitted on August 18, 2026, proposes Agentic ESOpt, a full-parameter fine-tuning framework for long-horizon language-model agents that uses evolution strategies instead of backpropagation-based…
A developer shares detailed notes on Scott Young's book 'Ultralearning', outlining strategies for self-directed learning. The notes cover defining clear goals, breaking down subjects into concepts, facts, and procedures,…
Researchers Zhikai Ding and Ziyi Ye propose Transfer-aware Dynamic Curriculum Sampling (TDCS), a method that adjusts the mix of training examples by difficulty level based on estimated transfer between levels, for large …
An arXiv preprint submitted August 18, 2026, and revised August 19 introduces Co-RL, a cooperative multi-agent reinforcement-learning framework that uses peer-generated rewards instead of ground-truth labels, reporting a…
An eight-author research team proposes Task Specialization Fine-Tuning (TSFT), a method that uses integer linear programming to allocate a limited fine-tuning budget across related task regions, starting from a single pr…