Machine learning news — deep learning, reinforcement learning, neural architecture search, diffusion models, and new ML frameworks and libraries.
Researchers propose DocOCR-Eval, an annotation-free framework for automatically evaluating and selecting OCR tools without ground-truth labels. The framework uses a three-stage correction and ranking strategy that aggreg…
Researchers introduce Alignment-Aware Bridge Matching (A²BM), a bridge matching method that leverages image pair alignment during training to improve image-to-image translation fidelity with weakly aligned data. A²BM con…
A new survey on arXiv (2607.16210v1) provides a unifying perspective on verification methods for reinforcement learning policies, introducing a taxonomy along three axes: verification paradigm, temporal scope, and guaran…
A new survey from arXiv provides a comprehensive review of Graph Neural Network (GNN)-based link prediction, proposing a taxonomy that categorizes advances by technique (GCN, GAE, GAT, GFormer) and application (knowledge…
Researchers introduce masked diffusion language models (MDLMs) as steerable text-based world models for reinforcement learning, achieving up to 47% absolute gains over baselines in zero-shot transfer across three out-of-…
Researchers propose trace-based on-policy distillation (TOPD), a teacher-supervised framework that transfers reasoning ability to a diffusion large language model (dLLM) without reward estimation. TOPD enables SDAR-4B-Ch…
A new benchmark, CIB-Med-1, reveals that diffusion-based medical image editors can appear successful by altering non-target findings rather than isolating intended pathology, a failure mode termed reward-hacking. Researc…
Researchers introduced GenSyn10, a CIFAR-10-aligned synthetic image dataset of 60,000 images generated using three architecturally diverse models (FLUX.2-dev, HunyuanImage-3.0, and Qwen-Image-2512) to benchmark AI-genera…
Researchers propose FedDP-PALD, a privacy-preserving federated latent diffusion framework for multimodal medical data synthesis that uses differentially private prototype aggregation to defend against membership inferenc…
Researchers propose W2SPO, an off-policy reinforcement learning method that uses short auxiliary segments from a weaker model to improve reasoning in large language models, achieving a Pass@1 improvement from 62.3% to 64…
A new study from arXiv evaluates nine open-weight small language models between 135 million and 3 billion parameters on a 1,085-example, 16-topic benchmark, finding that Qwen Coder 3B achieves the highest base accuracy a…
Researchers propose Group-Contrastive Forward-Forward (GCFF), a biologically plausible learning algorithm that yields monosemantic neurons organized into hierarchies of increasing abstraction without sparsity constraints…
Researchers have developed a Generalist Controller, a learning-based controller that uses attention mechanisms and a mixture-of-experts architecture to control systems of varying orders and dynamics without system-specif…
Researchers have released MFGNet-Gear, a synthetic 3D dataset of 24,000 paired polygon meshes and point clouds across 12 gear designs and 4 quality classes, to support deep learning for manufacturing quality inspection. …
Researchers at arXiv propose Contextual Information-Gain Policy Optimization (CIGPO) to fix a zero-advantage lock-in failure in outcome-only reinforcement learning for multi-turn evidence-reading agents. In HotpotQA expe…
Researchers propose Multi-level Context Fusion MOE (MCF-MOE), a framework that improves expert selection consistency in Mixture-of-Experts Transformer models by integrating cross-layer semantic aggregation and local toke…
Researchers released OpenLanguageModel (OLM), an open-source PyTorch library for building and pretraining small language models with readable, composable code. The library includes 27 presets across nine model families, …
Researchers from arXiv identify a structured confound in Reinforcement Learning from Human Feedback (RLHF) where pairwise preference labels may reflect the rater's state during annotation rather than just output quality,…
Researchers propose SelKV, a training-free framework for compressing the key-value (KV) cache in large language models (LLMs) that uses a soft cosine gate to selectively merge or discard tokens and an attention-ratio com…
Researchers have developed a token-level cross-modal transformer with contrastive multi-task learning that jointly performs breast cancer subtype classification and survival prediction by enabling fine-grained interactio…