Machine learning news — deep learning, reinforcement learning, neural architecture search, diffusion models, and new ML frameworks and libraries.
A new arXiv preprint (2608.13889v1) presents a consensus-gated multi-agent neural architecture search (NAS) system that uses three large language models (Claude, GPT-5.1, and Gemini 2.5 Pro) to debate and author candidat…
Researchers introduced FS-TPS and LaS-TPS, two stochastic control policies for transition path sampling that improve robustness and success in generating rare molecular transition trajectories. In multi-seed experiments …
Researchers introduced CAST (Closed-form Analytic Semantic Transfer), a training-free, image-free framework that extends pre-trained classifiers to unseen classes via weight injection, achieving performance comparable to…
A new arXiv paper (2608.13608v1) proposes a label-free framework for evaluating agentic continual learning harnesses in cybersecurity, using teacher-relative convergence as a proxy for true uplift. The authors show that …
Researchers introduced GALA, a two-stage generation-aware cross-modal alignment method for text-to-time-series synthesis, which sets a new state of the art on the TSFragment-600K benchmark, ranking first in 30 of 36 metr…
Researchers released CytoBERT, an open-source foundation model for single-cell cytometry data, pretrained on 15 human datasets with over 50 million cells to learn transferable inter-marker relationships. Fine-tuning for …
A study evaluating Kolmogorov-Arnold networks for land classification from Landsat 8 imagery found that the KAN matched random forest accuracy and outperformed multilayer perceptrons on an independent Calgary, Alberta da…
Researchers propose a fast implicit neural light field representation that decomposes a 4D light field into horizontal disparity, spatial texture, and vertical disparity planes, each represented by multi-resolution low-r…
A new arXiv study (2608.14114v1) finds that an AlphaZero-inspired reinforcement learning approach using Monte Carlo Tree Search achieves 98.43% peak survivability in power grid topological control, significantly outperfo…
Researchers introduced XIGL, an architecture-agnostic human-in-the-loop strategy that uses active explanation guidance to remove shortcut learning from Graph Neural Networks (GNNs), improving their reliability in out-of-…
Researchers introduced Attention-Aware Transform Coding (AATC), a KV cache compression method that reduces memory use by approximately 5.8x while maintaining near-lossless accuracy on models like Llama-3.1-8B-Instruct an…
A new expert-guided neuro-symbolic pipeline that combines a large language model with a Sugeno fuzzy inference system to assess compliance with sepsis treatment protocols found that antibiotic timing is the most critical…
A large-scale empirical study from arXiv (2608.13698v1) finds that training language models to reason in their native language via Group Relative Policy Optimization (GRPO) leaves only a small gap to English reasoning, w…
A study from arXiv (2608.14186) finds that a two-hidden-layer Deep Boltzmann Machine (DBM) matches the Autoencoder baseline on UCI Bank Marketing and statistically beats it on NSL-KDD, while outperforming seven other bas…
A new arXiv study (2608.14067v1) finds that in diffusion-based time series forecasting, continued low-noise refinement can introduce statistical drift and degrade forecasts, whereas general temporal structure is recovere…
Researchers introduced a reinforcement learning pipeline that uses execution-based supervision to improve multilingual code translation across 600 directed language pairs, achieving a 13% average improvement on the new H…
Researchers propose Batch-wise Adaptive Pruning, a training-free method for batched inference in Large Reasoning Models (LRMs), achieving a 39.7 percentage point average accuracy improvement over prior adaptive pruning a…
A new geometric filtering framework for selecting LLM-generated training samples improves few-shot text classification by +2.61 percentage points over SMOTE across 13 datasets, 5 classifiers, 10 augmentation methods, and…
A new arXiv paper (2608.13621v1) reveals that the Predictive Information Bottleneck VJEPA (PIB-VJEPA) model is mathematically equivalent to a hidden Markov model (HMM), with its stochastic context encoder acting as an am…
Researchers introduced a novel automata-based approach for reinforcement learning from Signal Temporal Logic (STL) specifications, constructing a timed alternating automaton to provide efficient memory and Markovian rewa…