Machine Learning News

TAM-Chain, a multi-scale thyroid cytology classification framework combining Absorbing Markov Chain theory with Shannon Entropy-based uncertainty quantification, achieved a Macro F1 score of 0.9741 and an absolute False-…

A new benchmark called RECLAIM, built from 100 NeurIPS 2025 papers, found that the best AI agent reproduced only 41% of Run-tier papers, 27% at the Retrain tier, and 15% at the Reimplement tier, according to the arXiv pa…

Researchers posted arXiv:2609.28859v1, a paper introducing SCALE, a sequential cost-aware policy that combines selective AI scoring with adaptive human escalation to run valid hypothesis tests at minimum cost. The paper …

Forecast-Dojo, a replayable environment for benchmarking and training LLM forecasting agents, was introduced in arXiv paper 2609.28876v1, combining 1,568 Polymarket events split by time into training and evaluation perio…

Researchers introduced PFArena, a benchmark with four controlled task interfaces covering single-mutant generation and multi-mutant ranking, to compare protein language models (PLMs), large language models (LLMs), and LL…

SMILESGNN, a multimodal architecture that fuses a SMILES Transformer encoder with a GATv2 graph encoder via cross-attention, achieved AUC-ROC 0.987 and F1 0.906 on ClinTox with only 0.4M parameters, according to an arXiv…

A new arXiv paper (2609.28561v1) introduces CARE (Condition-Aware REpresentation regularization), a lightweight plug-and-play framework that dynamically modulates feature distribution based on condition similarity in dif…

A physics-informed self-supervised learning framework jointly performs wire calibration and interaction position reconstruction in Multi-Wire Parallel Plate Avalanche Counters (MWPPACs) without labelled position measurem…

UO-FIE, a parameter-efficient system built on Qwen3.5-9B with LoRA, ranked first in the fine-tuning track of the Factivity Inference Evaluation 2026 (FIE2026) with a macro utility of 0.8316, according to the paper's abst…

A new arXiv paper (2609.28625v1) maps entropy-regularized reinforcement learning with verifiable rewards (RLVR) over myopic tabular policies onto a spin-glass energy model and shows that, for a wide class of models and t…

A region-held-out evaluation of 316L stainless steel SEM micrographs found that a local binary pattern plus support vector machine (LBP+SVM) classifier achieved a balanced accuracy of 0.79, H2 recall of 0.69, and H2 prec…

A new arXiv paper (2609.28581v1) reports that rule-set overlap between independently trained reinforcement-learning policies does not imply behavioral agreement, with policies sharing symbolic rules while choosing near-c…

LabFactory, a framework presented in arXiv paper 2609.28697v1, lets an AI builder turn a scientific brief into an executable AI lab that a separate host then runs on held-out inputs and scores under the task's protocol. …

A morphogenetic graph-generation framework that grows mechanical lattices node-by-node on a 3x3x3 dot matrix containing 27 candidate nodes, coupled to a three-layer graph convolutional neural network (GCNN) surrogate, ac…

A composable observation-interface method that conditions a single pretrained 13-variable atmospheric diffusion model on sparse radiosonde, clustered aircraft, and dense irregular surface-station data reduced RMSE agains…

Researchers proposed Predictive Action Chunk Learning (PACL), a method that lets robot manipulation policies learn from naturally accumulated mixed-quality autonomous rollouts without human corrections or exploratory int…

Fine-tuned XLM-RoBERTa achieved an F1-score of 0.5959 on Bangla medical named entity recognition, setting a new state of the art above the previously reported best of 0.5848, according to an arXiv benchmark evaluating Ba…

Researchers Zhuo Zhang, Shun Zou, Canqun Yang and Xi Yang introduced the Transformer-Mamba for Flow Field (TM4FF) framework, a physics-constrained operator learning model for computational fluid dynamics, in a paper post…

Researchers introduced m-WCN, an end-to-end deep learning framework that neuralizes multi-wavelet decomposition by approximating the classical GHM multi-wavelet transform with trainable convolutional operators under orth…

Researchers introduced FB-GDM, a fully-Bayesian guided diffusion method that eliminates the per-task scalar hyperparameter calibration required by Diffusion Posterior Sampling (DPS) and Pseudoinverse-Guided Diffusion Mod…