Machine learning news — deep learning, reinforcement learning, neural architecture search, diffusion models, and new ML frameworks and libraries.
Second-brain-mcp is a self-maintaining personal knowledge database that uses MCP, DuckDB, and biological memory models to automatically link, compress, and index saved papers, notes, and figures. The tool fetches full-te…
Researchers introduced Diffinity, a training-free guidance method that enables continuous diffusion language models to satisfy formal syntactic constraints defined by regular expressions. The method constructs an analyti…
London-based AI lab Inherent has raised $50 million in a seed round led by Index Ventures with participation from Radical Ventures. The company was founded by former DeepMind researchers Tantum Collins, Edward Hughes, an…
A recent study identified specific behavioral signals from sandbox executions that improve Trojan malware detection, focusing on feature selection rather than the deep learning model itself. Malware analysts typically si…
A developer's comparison of the RTX 5090 and RTX 4090 for AI workloads found that the 5090's jump to 32 GB of VRAM and 1,792 GB/s memory bandwidth is the most significant difference, not raw benchmark percentages. The an…
Researchers have developed RORA-VLM, a robust retrieval-augmented framework for vision-language models that enables accurate question-answering using external knowledge even when retrieved information contains noise. The…
Researchers have developed Differentiable Belief-based Opponent Shaping (D-BOS), a first-order method for multi-agent reinforcement learning that treats an observer's belief as the shaped opponent state and differentiate…
Researchers introduced TRACE, a training-free framework that stabilizes reconstruction trajectories in imaging inverse problems by coupling adjacent intermediate states. The method, which approximates proximal updates wi…
Researchers have introduced Embodied3DBench, a benchmark designed to evaluate low-level spatial intelligence in Vision Language Models (VLMs) within embodied 3D environments. The benchmark includes over 21,000 question-a…
Researchers have developed SalsaAgent, a multimodal embodied language model that generates expressive, full-body salsa dance motions in response to a human leader and synchronized to music. The system extends a large lan…
A team of researchers has developed a training-free multi-signal segmentation pipeline for the CVPR 2026 UG2+ Challenge's Dynamic Object Segmentation in Turbulence (DOST) track. The system combines RAFT motion estimation…
Researchers have developed TRACE, a trajectory-aware AI agent that uses large language model reasoning to optimize drug lead compounds by treating tool selection as a sequential decision-making process. The system outper…
Researchers have introduced COM, a strategy that enforces geometric constraints on time series token embeddings to preserve their inherent continuity and ordinality in large language models. The approach, detailed in a n…
A new 306M-parameter language model architecture, the Cognitive Categorical Transformer (CCT), achieved 21.27 validation perplexity on WikiText-103, a 12% relative improvement over a fine-tuned GPT-2 Small baseline. The …
Researchers have developed a lightweight multimodal large language model framework for defect grading of power transmission equipment, achieving state-of-the-art performance while reducing manual annotation costs. The ap…
Researchers have developed MechELK, a three-stage framework that uses mechanistic interpretability to extract hidden factual and reasoning knowledge from large language models. The framework, which combines sparse autoen…
A new study analyzing 17 language models from 410 million to over 100 billion parameters found that instruction-tuned systems systematically collapse linguistic entropy along discourse and structural dimensions, with mea…
Researchers have introduced Micro-Macro Retrieval (M2R), a new framework designed to reduce hallucination in large language models during long-form text generation. The system addresses the problem of factual errors by e…
A comprehensive evaluation of 14 open-source safety guard models on a benchmark of 79,331 samples found that Qwen Guard, a 4-billion-parameter model, achieved the highest recall at 83.97%, while larger models like Llama …
Researchers have introduced Thoughts-as-Planning, a framework that formalizes reasoning chain optimization in large language models as a sequential decision-making process over a latent semantic space. The method models …